See discussions, stats, and author profiles for this publication at: [Link]

net/publication/387970653

Artificial Intelligence, Machine Learning, and Deep Learning Applications in the

Engineering Fields -A Comprehensive Review

Preprint · January 2025

DOI: 10.14744/sigma.2024.00126

CITATIONS READS

0 526

3 authors, including:

Haneen Darwish Hiba H. Darwish

5 PUBLICATIONS 4 CITATIONS 12 PUBLICATIONS 49 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Hiba H. Darwish on 13 January 2025.

The user has requested enhancement of the downloaded file.

 Sigma Journal of Engineering and Natural Sciences
Web page info: [Link]
DOI: 10.14744/sigma.2024.00126

Review Article

Artificial Intelligence, Machine Learning, and Deep Learning

Applications in the Engineering Fields – A Comprehensive Review
Haneen Darwish Walaa Darwish Hiba Darwish

Abstract

This research paper addresses the critical issue of understanding the diverse applications of Artificial Intelligence (AI),
Machine Learning (ML), and Deep Learning (DL) across 14 distinct engineering disciplines, a topic of growing
significance as these technologies increasingly drive innovation in engineering. The stud y is motivated by the need to
clarify the hierarchical relationships and specific roles of AI, ML, and DL within these fields, which remains a
challenge in the current literature. A systematic review methodology was employed, analyzing 150 peer-reviewed
studies to identify the prevalence and effectiveness of AI, ML, and DL techniques. The analysis considered key factors
such as the specific technique used (AI, ML, or DL), the tools or software employed, the methodologies adopted, and
the outcomes reported. Notably, ML emerged as the predominant technique, utilized in approximately 73% of the
studies. This preference for ML is attributed to its versatility across various domains, refined algorithms developed
over decades, and superior interpretability compared to DL. The findings suggest that ML's widespread adoption in
engineering research is due to its ability to provide actionable insights across multiple disciplines, with DL being
applied primarily in cases where high-dimensional data or complex pattern recognition is essential. The research
highlights the significant contribution of ML in enhancing predictive accuracy and optimizing engineering processes,
underscoring its dominance in contemporary studies.

This study goes beyond previous efforts by not only providing a comprehensive overview of AI, ML, and DL
applications but also elucidating the hierarchical relationships between these technologies. It clarifies that AI
encompasses the broader field, ML focuses on algorithms that enable systems to learn from data, and DL represents
a subset of ML using neural networks for complex feature extraction. The novelty of this work lies in its systematic
comparison of these techniques across a wide range of engineering disciplines, offering new insights into their
respective contributions and interactions. These findings provide a clearer framework for understanding how AI, ML,
and DL can be effectively leveraged in engineering research, guiding future studies and applications in the field.
Keywords: Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL), Applications, and engineering.

1. Introduction
Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) have become integral to
modern life, profoundly influencing our interactions with technology and information. AI refers to the
simulation of human intelligence in machines, enabling them to perform tasks that typically require human
cognitive functions. ML, a subset of AI, focuses on systems that learn from data and improve performance
over time without explicit programming. DL, a subfield of ML, involves neural networks with multiple
layers, mimicking the human brain's structure for more sophisticated processing. These technologies are
pervasive, found in applications ranging from voice assistants and recommendation systems to image

1
recognition and autonomous vehicles. They enhance efficiency, automate tasks, and provide personalized
experiences, transforming industries such as healthcare, finance, and education. In essence, AI, ML, and
DL are essential tools driving innovation and efficiency, significantly impacting various aspects of
contemporary life. However, despite their frequent interchangeable use, these terms do not convey identical
meanings. To better understand the distinctions among AI, ML, and DL, Figure 1 presents a graph
elucidating the key differences. The following subsections provide a more detailed discussion of AI, ML,
DL, and RF, along with the algorithms employed in these fields.

Artificial Intelligence

Targeted marketing
M arket forecasting
Population growth
Machine Learning
prediction Clustering
Regression

Customer
Weather
Unsupervised segmentation
forecasting Deep Learning Learning
S upervised
Learning

Diagnostics
Dimensionality

Image classification reduction

Classification
Feature Big data
Real time decision elicitation visualization
Customer Retention
Identity fraud
detection Reinforcement
Learning
Robot navigation
Learning tasks

Figure 1: AI, M L, and DL configuration

This comprehensive review delves into the transformative impact of AI, which is the umbrella of ML,
and DL across 14 diverse engineering domains. Our contributions are summarized in the following three
points.
 Each engineering type is meticulously examined, unraveling the employed methodologies,
innovative applications, and succinct summaries of outcomes;

2
  This research illuminates the pervasive influence of cutting-edge technologies in reshaping and
optimizing various facets of engineering, offering a valuable synthesis of advancements that bridge
theory and practical implementation; and
 A statistical section is also provided that clarifies the percentages of Al, ML, and DL application
in the engineering fields.
These contributions collectively advance the understanding of AI, ML, and DL's role in engineering,
providing a foundational reference for future research and development, which is the novelty of this work.

1.1 Artificial Intelligence

In simple words, AI is the process of transferring the information, data, human intelligence to
machines. The fundamental objective of artificial intelligence is the development of autonomous entities
endowed with the capacity for human-like thought processes and actions. Such entities execute tasks by
engaging in problem-solving through research and emulation of human behaviors. The majority of AI
systems emulate natural intelligence in order to address intricate challenges [1].
The word AI refers to the computer algorithms, which simulate cognitive processes or activities
observed in living organisms. Nowadays, AI is integrated into the scientific discovery for many purposes
such as collect and interpret large datasets, design and guide experiments, helping scientists to introduce
some hypotheses, accelerate research process, provide reliable and actionable models integrated with the
scientific workflows, and refine measurements. Moreover, AI can be used in decision-making, suggestion,
problem-solving, estimating, understanding, and learning [2-3]. AI along with the Internet of Things (IoT)
are mandatorily required to handle massive data effectively in less time by applying the theorem of
algorithms even though the training along with the execution of the algorithms requires substantial storage,
power, and massive data to be trained [4-5]. Nowadays several AI tool are used not only in engineering
field but in general, the following subsection discuss the most known one, which is helpful for the
researchers in general.
1.1.1 ChatGPT

Open AI released the ChatGPT (Chat Generative Pretrained Transformer) [6], which is defined as
an intelligent chatting robot that has the ability to give an immediate response based on the instruction
provided by the user [7]. For instance, Figure 2 shows a request from the user asking ChatGPT to give
knowledge about AI in only one paragraph and the response was exactly as requested. ChatGPT showed an
incredibly powerful in generating several tasks extended from trivial tasks to several language
understanding. Some of these tasks are story writing, debugging, grammatical correction, and multilingual
machine translation [7]. Moreover, natural language processing problems could be also solved by ChatGPT
such as semantic annotation and acceptance of the text going towards natural language understanding
(NLU) like word sense disambiguation (WSD), providing answers for the provided questions in the input
text box, a relatively simple binary classification of texts like spam, aggression detection, sarcasm, and
other tasks like a more complex multi-class and multi-label classification of texts such as sentiment
analysis, emotion recognition [8]. These days, ChatGPT has gained much attention due to its ability to
provide high-quality as well as fluent responses [9]. However, this tool is very helpful if and only if it is
used wisely, the educational facilities can encourage academic integrity and avoid the disruptive impacts
of this technology by introducing guidelines and clear policies [10]. GPT is being improved continuously
started from GPT 1 to GPT 4, Table 1 clarifies the differences between these versions [7].

3
 Figure 2: An example of how ChatGPT works

Table 1: Comparison between GPT-1, GPT-2, GPT-3, and GPT-4

Released year Context S ize of pre- Data source Learning

window training data target

GPT-1 2018 512 tokens Around 5 GB Book corpus Unsupervised

and Wikipedia learning

GPT-2 2019 1024 tokens 40 GB Web text M ulti-task

learning

GPT-3 2020 2048 tokens 45 TB Common crawl In-context

learning

GPT-4 2023 8195 tokens Unavailable Unavailable M ultimodal

information information learning

1.2 Machine Learning

The field of machine learning (ML) has undergone substantial advancements over the preceding
two decades, transitioning from an experimental pursuit confined to laboratories to a pragmatic technology
entrenched in pervasive commercial application. Within the domain of AI, ML has ascended as the
preeminent approach for the formulation of functional software across diverse domains such as computer
vision, speech recognition, natural language processing, robot control, and various other applications [11].
ML has emerged as a crucial method for harnessing the capabilities of data, enabling businesses to enhance
their innovation, efficiency, and sustainability [12].
ML involves the examination of data and the creation of algorithms to make predictions based on
the, previously, trained data. ML programs rely on a computer's ability to learn without explicit
programming. The ML model may either be predictive or descriptive, to gather information about input
data [13].
Constructing an effective machine learning model involves two crucial stages: training and testing.
During the training phase, a significant and research-intensive process, the system is presented with labeled
or unlabeled inputs. These inputs are stored in the feature space for future reference in making predictions.
Subsequently, in the testing phase, the system is presented with an unlabeled input, and it must predict the
correct output based on its training [14]. ML enhances task performance and productivity by incorporating
learning mechanisms and self-correction in response to new data by five main steps, which are clarified in
Figure 3 [15].

4
 Figure 3: steps of applying M L

In the first step, the required data will be collected from several sources such as World bank or
Kaggle, and in the second step some analysis for the data will be made such as the missed data, or handling
the invalid data by several analytical techniques and finally the data will be preprocessing for feature
extraction, feature analysis, and data visualization. The data then will be divided into two main parts for
two main goals; training and testing. In the third step, the testing part of the available dataset will be used
to train the model by various ML algorithms. In the fourth step, the accuracy of the model will be tested by
the second part (testing part of the data). And in the fifth step, model deployment, the ML model will be
integrated with an existing production environment that takes the input and gives the output to make
decision making or to achieve any another goal. Several technologies can be used to deploy the ML model
such as: Azure ML Service, Docker, Kubernetes, MLFlow, AWS SageMaker. Finally, the last step the
obtained model will be monitored from several aspects such as the crashes, errors and to assure that the
model still provides the desired performance [15]. In general, ML have wide advantages and benefits, which
are shown in Figure 4, and accordingly this promotes it to be applied in several applications as represented
in Figure 5 [15]. ML is divided into three main types, and each type is familiar with specific sorts of dataset.
These three types are clarified in Figure 6 [16].

Increased in
sales in the
e-commerce
market
ML benefits
in the medical
Work domain for
Automation enhancing
medical
diagnosis
Main benefits of
ML

Modeling the ML is used in

data to make robotic
useful medical
decisions M L in finance surgery
increases
productivity,
enhances
revenue and
gives secure
transactions
Figure 4: M ain benefits of M L

5
 Automatic
Faster route language Ads Recomm- Smartphone
selection in translation in Speech endation with face
Recognition
Google Map Google system recognition
Translate

Stock Driverless/
Medical Weather Fraud
market Chatbot Self-
trading Diagnosis Prediction Detection driving car

Figure 5: Real-World Applications of M L

Machine learning types

Supervised learning Unsupervised learning Reinforcement learning

Classification Regression Clustering

Figure 6: M achine learning types

 Supervised Learning
The preeminent ML type is the supervised learning model, primarily employed for addressing practical
applications in the real world [17]. This model predicts results based on specific input sets and pairs of
input/output examples. Each training dataset in supervised learning comprises input goals, an input vector,
and a desired output value known as a supervisory signal. These examples train machine learning
algorithms, leading to the derivation of a classifier function through dataset analysis. The goal of supervised
learning training algorithms is to forecast one or multiple outcomes using different input features. While
supervised learning is effective, it has the limitation of necessitating a substantial amount of labeled data

6
for the creation of a comprehensive labeled dataset [18]. Supervised learning models are widely employed
for tasks involving classification and regression as shown in Figure 6.
The supervised learning is divided into two main parts, classification and regression. In the realm of
Classification, a computer program undergoes training using a designated dataset. Through this training
process, the program organizes the data into distinct class labels. This algorithm proves valuable for
predicting discrete values, such as distinguishing between male and female, determining the truth or falsity
of a statement, or identifying whether an item is spam or not. On the other hand, the regression algorithm's
objective is to discover a mapping function that links input variables (x) to a continuous output variable (y).
These algorithms are employed to forecast continuous values, such as predicting prices, salaries, ages,
marks, and similar variables.
The primary goal of ML – supervised type is classification or prediction. These methods categorize and
predict class labels using a predefined set of examples. Classification samples either entirely belong to a
specific class or do not belong to any class. Partial membership in a class is not considered. The presence
of missing values can hinder the accuracy of classification and prediction processes. Classification comes
in two forms: binary and multiclass. Binary classification involves two sets of classes, where input data are
organized into these two categories. For instance, making a YES/NO prediction or classifying emails as
spam or non-spam falls under binary classification, represented as 0 and 1. On the other hand, multiclass
classification deals with three or more predictable classes, such as identifying different cancer stages, where
classes are denoted as 0, 1, 2, and so forth.
Figure 7 demonstrates the process of addressing a given problem through supervised learning [14], and
Figure 8 shows how the input data will be handled by this approach. This approach typically involves
specific steps. Initially, it is crucial to identify the type of training examples relevant to the problem.
Subsequently, a training set is compiled, sourced either from human experts or through measurements. The
training set should accurately represent real-world scenarios where the function will be applied.
Determining the representation of input features is the next step, ensuring it contains sufficient information
for precise output predictions. Following this, the selection of an appropriate learning algorithm is essential.
Once the design is complete, the learning algorithm is applied to the assembled training set. During this
stage, some supervised learning algorithms may necessitate user input to determine specific control
parameters, particularly for prediction-related challenges [19].

Figure 7: Followed steps in supervised learning

7
 Model

Input data Output

Figure 8: Principle of supervised learning [20]

 Unsupervised Learning
In the second type, unsupervised learning, the algorithm learns from an unlabeled dataset and endeavors
to comprehend the data by autonomously extracting features, identifying co-occurrences, and revealing
underlying patterns.
Unsupervised machine learning serves the purpose of discovering concealed patterns in data without
predefined labels. Various effective applications have employed this approach, but their evaluation can be
challenging. This difficulty arises from a deficiency in training related to unsupervised machine learning,
leading to a lack of clear measures such as error or reward indicators for assessing potential solutions. The
absence of a reward signal is a notable distinction between supervised and unsupervised machine learning.
In statistical applications, unsupervised learning finds utility in approximating data density.
Neural network (NN) models [21], self-organizing maps (SOM) [22], and adaptive resonance theory
(ART) [23] are examples of unsupervised learning techniques. Unsupervised learning involves two main
processes: data transformation and clustering. During data transformation, the dataset undergoes
modifications to represent information in a different, more comprehensible format for both humans and
machine algorithms. On the other hand, clustering algorithms categorize datasets into meaningful groups
of related objects. The K-means clustering algorithm, recognized as the simplest and most widely used
unsupervised method, identifies clusters of similar data. The algorithm involves two steps: initially
assigning each data point to the nearest cluster center and subsequently updating all cluster centers to the
mean of the data points assigned to them.
On the other hand, a fundamental challenge in unsupervised learning lies in assessing its effectiveness.
Success in unsupervised learning signifies whether the algorithm has acquired meaningful insights. Unlike
supervised learning, unsupervised learning lacks predefined labels or outputs, making it challenging to

8
ascertain the correctness of the results. Consequently, the evaluation of algorithm performance becomes
complex. As a result, unsupervised learning is predominantly employed for exploratory purposes, such as
enhancing data understanding. Another noteworthy aspect of unsupervised algorithms is their role in
preprocessing for supervised algorithms. The exploration of alternative data representations can contribute
to enhancing the accuracy of supervised algorithms.
To figure out the difference between the classification, regression, and clustering, Figure 9 shows a
representation for them and how they can be applied in the data analysis. Most of the ML models can apply
both types of data, the labelled and the unlabeled, in some certain conditions. However, labelled data are
employed in supervised learning [14].

Classification Regression Clustering

Figure 9: Representation of classification, regression, and clustering

 Reinforcement
The third type, reinforcement learning (RL), is a subset of machine learning focused on guiding
intelligent agents in their actions within an environment to maximize rewards over time. The environment
is the context in which the agent learns and makes decisions. The environment responds to the agent's
actions by providing rewards or moving to a different state. In the reinforcement learning approach, the
emphasis is not on instructing the agent on how to perform a specific task but on providing incentives,
either positive or negative, based on the agent's behavior [24]. In this type, the model acquires behavior in
an environment by taking actions and assessing the resulting reactions. In RL, the objective is to select
actions that optimize positive outcomes in specific situations. The reinforcement model determines which
actions to undertake for a given task, emphasizing the importance of learning from experience.
1.2.1 Used Algorithms
In this study, a quick comprehensive exploration of ML algorithms is undertaken, encompassing
the three primary types: supervised learning, unsupervised learning, and reinforcement learning. Supervised
learning involves algorithms such as Support Vector Machines (SVM), Decision Trees, and Neural
Networks, where models are trained on labeled datasets to predict outcomes for new, unseen data.
Unsupervised learning, on the other hand, employs algorithms like K-means clustering, Principal
Component Analysis (PCA), and Gaussian Mixture Models (GMM) to uncover hidden patterns and
structures within unlabeled data. Reinforcement learning introduces a distinct set of algorithms, including

9
Q-Learning, Deep Q Networks (DQN), and Policy Gradient Methods, which enable intelligent agents to
learn optimal actions in dynamic environments through a system of rewards and penalties. This section
aims to provide a comprehensive overview of the diverse ML algorithms utilized in the research field.
Figure 10 shows the used algorithms in each type [15] [25].

Figure 10: List of used algorithms used in M L

 Decision Tree

Decision tree algorithm serves as a visual roadmap for navigating a series of options or decisions guided
by specific conditions. As a subtype of the classification tree, it organizes instances into various categories
based on their property values. Each node within the decision tree embodies a unique characteristic of an
instance essential for classification [26]. Figure 11 shows the schematic representation of the decision tree
algorithm [27].

Condition 1

Yes No

Condition 2 Condition 3

Yes No Yes No

Decision 1 Decision 2 Decision 3 Decision 4

Figure 11: Decision tree algorithm principle

10
  Logistic Regression

Logistic regression is a statistical methodology employed for predictive modeling and classification
purposes. It offers a structured approach particularly conducive to categorical outcome prediction,
accommodating both categorical and continuous predictor variables. The primary objective of logistic
regression is to assess the probability of an event occurrence based on the provided independent
variables. As the output signifies probability, the dependent variable is constrained within the range of
0 to 1. This logistic regression model is formally expressed through Equation (1) [28] and Figure 12
clarifies the representation of this algorithm [29].
𝑝
𝐿𝑜𝑔(1−𝑝) = 𝛽𝑜 + 𝛽1 𝑋1 + 𝛽2 𝑋2 + ⋯ + 𝛽𝑛 𝑋𝑛 + 𝛽12 𝑋1 𝑋2 + ⋯ + 𝛽𝑛−1,𝑛 𝑋𝑛−1 𝑋𝑛 (1)

Where p is the probability which is associated with an event occurrence, 𝛽𝑜 is the coefficient of the
intercept term, and 𝛽1 to 𝛽𝑛 are the coefficients of the independent variables (𝑋1 𝑡𝑜 𝑋𝑛 ).

Threshold Value

Figure 12: Logistic regression algorithm representation

 Support Vector Classifier (SVC)

SVC stands as a pivotal tool in supervised machine learning, tasked with scrutinizing input data to
discern patterns, thereby facilitating classification or regression analysis. Its utility spans across diverse
domains, encompassing digit recognition, handwriting analysis, face detection, cancer classification, and
time series forecasting [30].
The SVC algorithm, a seminal creation by Vapnik and Cortes, epitomizes an efficacious machine learning
approach renowned for its adeptness in data analysis and pattern recognition. Endowed with the inherent
advantages of global optimization and robustness against overfitting, SVM emerges as the cornerstone for
both binary and multi-class classification tasks. Its applicability extends to varied domains such as text
categorization, handwriting recognition, gene-function prediction, and remote sensing classification,
underscoring its versatility. Although SVM typically operates optimally with modest datasets, it exhibits
sensitivity to kernel functions and outliers. Moreover, the efficacy of SVM hinges significantly upon
judicious selection and fine-tuning of kernel functions [31]. Figure 13 shows the representation of this
algorithm.

11
 SVM hyperplane

Support vectors

Figure 13: representation for SVC algorithm

 Random Forest Classifier

The random forest algorithm serves as a strategic countermeasure against the prevalent risk of
overfitting commonly encountered in singular decision tree implementations. It embodies an ensemble
learning paradigm, whereby multiple decision trees are trained concurrently. The culmination of this
process involves aggregating the outputs from each constituent tree to derive the algorithm's final
prediction. In essence, every decision tree autonomously generates its prediction, which is subsequently
harmonized through either averaging for regression purposes or employing majority voting for
classification endeavors [28]. Refer to Figure 14 for a graphical depiction of this algorithm.

Dataset

Feature 1 Feature 2 Feature N

Tree 1 Tree 2 Tree N

Class A Class A Class B

Majority voting

Final class
12
Figure 14: representation for random forest tree algorithm
1.3 Deep Learning
The notion of deep learning (DL) has a longstanding history, but recent years have witnessed its
transformation due to the accessibility of extensive datasets and advancements in computing capabilities.
DL represents a swiftly evolving field that has gained widespread acceptance in recent times [32]. This
approach employs algorithms to acquire knowledge from data, aiming to enhance the precision and
efficiency of predictions or decision-making processes. Neural networks are pivotal in DL, enabling the
learning process from extensive datasets [33-41]. The evolution of DL has paved the way for new avenues
in research and innovation, resulting in substantial progress across various domains such as healthcare,
finance, transportation, and beyond [32].

Though ML and DL are frequently used interchangeably, they represent discrete facets of artificial
intelligence. ML entails creating algorithms that empower machines to learn from data without explicit
programming. In contrast, DL is a subset of machine learning that employs neural networks to emulate the
structure and functioning of the human brain. These neural networks excel at learning from unstructured
data, proving highly proficient in addressing intricate challenges like image and speech recognition. [42-
48].

Based on the previous sections, and based on [49-67] a comparative Table 3 for AI, ML, and DL is
shown for better understanding.

Table 2: Advantages and disadvantages of AI, M L, and DL

Technique Advantages Disadvantages

AL AI encompasses a broad range of Disadvantages: Developing AI
techniques, including ML and systems can be resource-
DL, to simulate human-like intensive, requiring substantial
intelligence. It can handle computing power and expertise.
complex decision-making and is They may struggle with adapting
versatile across different to new scenarios without
domains. extensive reprogramming.
ML ML algorithms learn from data ML models depend heavily on
patterns to make predictions and labeled data for training, and
automate tasks. They find their complex nature.
applications in various industries
for improving efficiency and
decision-making.
DL DL excels in processing large, Training deep learning models
unstructured datasets like images requires significant
and text. It automatically extracts computational resources and can
features, achieving state-of-the- over fit with limited data. The
art performance in tasks such as lack of transparency in how
image recognition and natural decisions are made (black box
language processing. problem) poses interpretability
challenges.

13
 2. Applications of the Aforementioned Techniques in Engineering
As stated, this research concentrated on showing the applications of AI, ML, and DL in engineering
fields. Table 3 illustrates how these techniques are applied in 14 fields of engineering science.
Table 3: Comprehensive review of AI, M L, and DL applications in engineering fields

Automobile Engineering
Used
Ref technique Used Methodology Summary of the outcomes Comments Year
#
tool
(AI/ ML/
DL)

[68] ML --- In this study, the authors applied The outcomes showed that the VRP The background, problem 2023
a combination method between can be enhanced by ML models, and statement, the research
ML and analytical methods to the corresponding performance of objective, and the
improve the vehicle routing the vehicle could be improved. methodology were
problems (VRP). introduced in a very good
way assuming that the reader
has a brief knowledge of this
field.

[69] DL Matlab In this study, the authors The outcomes showed that the built The study was, totally, 2023
developed a novel model, an four models achieved a better conducted theoretically.
improved hybrid ANFIS, to accuracy of 0.5% in comparison Future research may apply
design a practical, accurate, and with the 2% stated in the previous and test the adopted
simple Lithium-Ion battery state studies. methodology practically.
of charge estimator.

[70] ML GT-Suite In this study, the authors The proposed method showed a In general, the analyzed 2023
developed a parametric ML good performance with an accuracy different cycles should
model with the aid of GT-Suit e of 89% in being able to predict the include the worst scenario.
software. The parameters state of charge of the battery, and For instance, the distances
included in the study are energy this can help automobile engineers traveled scenarios in the
consumption, distance traveled, to design scalable and more efficien t study varied between 4.1 km
vehicle speed, state of charge of electrical vehicles. and 20.1 km. it would be
battery, and motor power. better if the study
investigated longer
distances.

[71] AI and Matlab This study proposed a method to The outcomes showed that the best this study did not consider 2022
ML predict the required time for EV model was GWO-based ML as it the challenges in charging
charging based on collected data achieved better results compared that are met.
from 500 EVs. The proposed with the rest ones.
method depended on using the
ML algorithms and the AI was
also adopted to optimize the
parameters of ML.

[72] ML ---- In this study, the authors The outcomes showed that Long The proposed method has 2023
investigated the charging system Short-Term Memory (LSTM), could several limitations, which
of EV in a smart grid by using be employed to provide EV control

14
 several ML algorithms to in particular circumstances in are already included in the
achieve the goal of reducing load addition to keeping the billing costs paper.
fluctuation, power loss, and minimal.
expenses. Then, the proposed
algorithms were evaluated and
compared.
Agricultural Engineering
Used
Ref technique Used Methodology Summary of the outcomes Comments Year
#
tool
(AI/ ML/
DL)

[73] ML --- In this study, the authors The key outcomes from this study The paper suggested to apply 2020
analyzed the existing ML models showed that the farm managemen t real-time on new
and applications in agriculture should be evolved with the reality by applications as it is the gap in
engineering starting from the linking the related data of it with this research field.
process till the outcomes. each sensors to provide better suggestions
application was discussed with that can enhance the final products
its advantages and its from the agriculture
corresponding weaknesses.

[74] ML WEKA In this study, the authors The processing of data from ARFF The study linked how the 2023
designed and applied an ML format to its final representation as a animal affect the agriculture,
workbench, WEKA, that admits set of rules or a decision tree was which is a good addition.
rapid experimentation on a given facilitated by the WEKA
dataset by using several ML workbench. A variety of state-of-
schemes. the-art, robust, and efficient ML
schemes were provided for
experimentation, and a
concentration on simplifying access
to standard schemes for performance
evaluation and comparison was
maintained.

[75] AI and --- The study investigated and At the end of the study, it was stated This paper investigated the 2023
showed the applications of ML that AI, ML, and DL can be applied applications of ML and AI in
ML
and AI in agriculture engineering to enhance the production of crops several aspects besides the
along with some other sectors. and design a reliable product quality agriculture such as the health
the methodology of this study monitoring system. care, disease prediction, and
can be simply introduced by water waste controlling in a
discussing several techniques readable well-organized way
and algorithms of ML and AI.

[76] AI --- This study analyzed several The outcomes from this study A deep discussion was 2023
relevant articles on AI showed that AI technologies, which provided for each
applications in the agriculture use data like precipitation, application.
sector. For instance, some temperature, solar radiation, and
applications such as detecting wind speed in combined with some
the smallest insects in the farms ML algorithms and images taken by
by AI, how to provide proper drones or any other technologies can
guidance on water management effectively predict crop
by AI, how to Increase sustainability and, accordingly, ease
agricultural yield and the process for the farmers to

15
 productivity, and how to evaluate farms for the presence of
Increase farmer profitability diseases or pests and inadequate
were investigated in this paper. plant nutrition.

[77] DL Matlab This study employed some data The outcomes from this study The resolutions of the figures 2023
related to the humidity and showed that the yield production and tables are not well
temperature in some algorithms depends mainly on the soil nutrients. enough.
such as KNN to find a model that Moreover, it was concluded that
could be used for climat e Remote Sensing area and data
prediction based on a previous mining are used to provide some
observed data. algorithms and prognosis strategies
for soil attribute classification.
Aerospace Engineering
Used
Ref technique Used Methodology Summary of the outcomes Comments Year
#
tool
(AI/ ML/
DL)

[78] AI --- This study proposed a novel The outcomes showed that the In this field, the cost is one of 2023
artificial bee colony algorith m proposed method was effective the key points that should be
that was implemented as part of considered. However, this
a hybrid method to solve flexib le paper did not investigate
job-shop scheduling problems how the cost will be affected.
by extracting the key
manufacturing informatio n
through an aerospace workshop.

[79] ML python The authors developed a new The outcomes showed that the This paper handled the 2023
ML algorithm to predict the proposed algorithm gave good complexity of the nonlinear
fatigue damage of the material performance and accuracy. equations effectively.
by preprocessing the received
signals

[80] ML and Teachable The authors proposed a DL The outcomes showed the The cost was considered in 2023
M achine framework to identify faulty importance of the proposed method this paper as a key objective,
DL
components due to especially in managing the operation which is a good contribution.
and
malfunctioning in the aerospace cost along with reducing the
ANSYS industry. This could help the reducing rework-induced delays.
engineers in focusing on The statistical outcomes showed a
choosing the most proper 34.32% and 52.88% reduction in
material in the manufacturing total cost and time delays
stages. Moreover, the study respectively.
employed the CNN approach to
recognize the intermediat e
abnormal situations.

[81] ML Python This paper proposed a model to It was found that random forest was a large number of the 2023
predict fatal occurrences in the most accurate model as it limitations was considered
aviation events by applying achieved an accuracy of 90%. and discussed in this
some data on ML models. This is research, which is a good
very helpful for the engineers as contribution.
the safety standards are very
rigorous,

16
[82] ML --- This paper compared between The outcomes showed that to No clear objective function 2023
the ML models used in aircraft enhance the aircraft maintenance was provided in this study.
maintenance systems. system, several should be considered
during the evaluation and
comparison between the models
such as the cost as well the time.
Some model showed their
superiority from the cost aspect
while others showed their
superiority from the time aspect.

Architectural Engineering

Ref Used Used Methodology Summary of the outcomes Comments Year

# technique
tool
(AI/ ML/
DL)
[83] ML M ATLAB This study proposes a real-time The proposed method, validated in further validation on a 2023
seismic damage simulatio n two city-scale cases, offers accurate broader range of urban
method for urban building predictions of maximu m inter-story settings would strengthen its
portfolios by integrating a drift and floor acceleration with credibility. Overall, the
multiple degree-of-freedo m MDOF models, significantly method's efficiency and
(MDOF) shear model, flexural- improving computational efficiency accuracy make it a
shear model, and machine compared to time-history analysis. noteworthy contribution to
learning. The research The machine learning surrogate seismic risk assessment in
introduces formulas for model further enhances efficiency , urban environments.
predicting periods of various achieving up to 76,600 times higher
structure types and establishes efficiency than traditional methods.
surrogate models to optimize
MDOF models.
[84] AI ---
The methodology integrates Results from city-scale cases The study underscores the 2023
multiple degree-of-freedo m showcase the method's capability to efficiency and effectiveness
(MDOF) shear and flexural- achieve real-time seismic damage of the AI-driven
shear models with machine simulation with acceptable accuracy. methodology in enhancing
learning for real-time seismic The proposed approach eliminates seismic risk assessment and
damage simulation of urban the need for extensive retraining mitigation for urban building
building portfolios. Formulas when applied to other urban building portfolios.
predicting periods for different portfolios, demonstrating its
structures are introduced, and potential for widespread application.
surrogate models are established
to optimize MDOF models.

[85] AI --- The paper investigates the The paper emphasizes the strong Current limitations in AI art 2023
applicability of diffusion-based potential of AI tools in architecture, tools for architecture include
AI art platforms, specifically particularly for tasks like ideation, struggles with complex
Midjourney, DALL·E 2, and collages, building and style variants. prompts and a reliance on
Stable Diffusion, to common use While acknowledging the current specific design trends.
cases in architectural design. The challenges with more complex Improved semantic
study compares the capabilities prompts, the authors foresee understanding and
of these public AI art platforms improvements with new diffusion integration with

17
 and outlines the requirements for models trained on Buildin g architectural workflows are
AI art platforms in supporting Information Models (BIM) datasets areas for development.
tasks in civil engineering and Ongoing research in
architecture structural dynamics and
construction technologies
may pave the way for a new
era of AI-influenced
architectural styles.

[86] AI --- The research employed a IoT devices seamlessly integrated The research identified 2023
comprehensive approach to into construction equipment and several challenges, including
assess the integration of AI, IoT, structures facilitated real-time data technical issues like data
and Big Data technologies in the gathering, leading to AI-driven integration, security, and
Architecture, Engineering, and predictive maintenance and scalability, as well as ethical
Construction (AEC) industry. optimized workflows concerns related to
The study involved an in-depth workforce displacement and
analysis of various AI models biases in AI algorithms.
and techniques shaping
Construction 4.0 and 5.0, along
with the societal changes in
Society 5.0.

Biomedical Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[87] ML Matlab The proposed method used KNN It was found that KNN gave a very A comparative analysis 2023
algorithm to segment the good accuracy in images based on MSE, PSNR, and
medical images, and extract the classification, which is a very SSIM were applied for only
features from the image, and important phase in medical imaging . the proposed algorithm.
finally analyze the data by
classifying it based on NN.

[88] ML Matlab In this study heart disease was The results showed that the proposed Only three types of disease 2022
predicted by applying the ML method can be used to accurately were included along with
techniques. The proposed diagnose the stage of a disease and their symptoms, and
method was created after a its treatment using this approach. prevention.
comparison phase between many
ML algorithms by Matlab
coding, since the effective ML
strategy, in general, requires
precise identification of the
problem.

[89] DL --- This study proposed a new IoT The experiment results proved the This paper has a strong 2021
and DL enabled healthcare supremacy of the IoTDL-HDD mathematical foundation
disease diagnosis model by using model with an accuracy of 93.452%. which supported the
biomedical signals. The outcomes.

18
 proposed model used a BiLSTM
feature extraction to extract the
feature to discover CVDs
existence.

[90] ML --- In this study, the authors The analyses showed that 43% of the The paper has a well- 2023
analyzed the most recent studies studies used the IoT. Moreover, the organized structure
that investigated the applications most used DL algorithm was the
of ML techniques in medical CNN.
data processing. 27 articles were
discussed and the following
aspects were included: deployed
methods, proposed framewo rk,
the advantages and the
disadvantages.

[91] ML and --- This study presented systematic The outcomes showed that the ML This paper used some studies 2022
studies related to applications of was used more than DL in medical from 2014, which could be
DL
ML and DL in the detection diagnosis replaced with the most recent
along with classification of ones.
multiple diseases. 40 studies
were collected from journals and
conferences from 2014 to 2022.
Civil Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[92] ML --- This study employs six machine The findings confirm the high While the study successfully 2021
learning techniques to predict the accuracy of all six machine learning demonstrates the efficacy of
compressive strength of self- models in predicting the machine learning techniques
compacting concrete. These compressive strength of self- in predicting compressive
include established AI compacting concrete. ELM tuned strength in self-compacting
algorithms like artificial neural with MOFA consistently concrete, it would benefit
network (ANN), adaptive neuro- outperforms other models across from a clearer exploration of
fuzzy inference system (ANFIS), various metrics, demonstrating its potential limitations or
and extreme learning machine superiority. Visual representations challenges faced during the
(ELM), coupled with nature- validate the models' effectiveness, implementation of the
inspired optimization algorithms suggesting potential benefits for proposed models in real-
like moth flame optimizatio n quality control, cost reduction, and world construction
algorithm (MOFA) and wild enhanced safety in construction scenarios.
horse optimizer (WHO). processes.

[93] ML --- The study introduces The research showcases that the the study acknowledges the 2024
Mechanics-Informed Surrogate developed Mechanics-Informed need for dedicated hardware
Models (MISM) for structural Surrogate Models (MISM), and a slightly longer training
analysis, combining empirical incorporating Graph Neural phase, suggesting a trade-off
data and mechanics prior Networks, outperform standard between computational
knowledge, and utilizes Grap h surrogate models in predicting effort and improved
Neural Networks (GNNs) to deformations for structural systems. accuracy, which could be a
represent and embed knowledge Despite requiring dedicated consideration for practical
about truss structures. Applied to hardware and a slightly longer implementation.

19
 bi-dimensional and tri- training phase, the study highlights
dimensional truss structures, the the trade-off between computational
proposed approach demonstrates effort and enhanced predictive
superior predictive accuracy accuracy, suggesting the potential of
compared to standard surrogate GNNs in advancing structural
models, emphasizing the engineering applications and
integration of mechanics for predicting nonlinear behavior in
improved model interpretability. existing structures.

[94] ML Python This study integrates advanced The regression analysis results of The comprehensive 2023
image processing and machine this study exhibit a notable level of methodology, employing
and
learning techniques, specifically predictive accuracy and strong Python and MATLAB for
MATLAB computer vision approaches, to correlations between identified data analysis and imag e
predict structural performance crack patterns and structural damage segmentation, demonstrates
and assess damage levels in in exterior beam-column joints. The robust predictive accuracy
reinforced concrete exterio r application of machine learning across diverse experimen t al
beam-column joints. Using regression techniques proves conditions.
Python for data analysis and effective in estimating damage states
MATLAB for imag e based on surface crack patterns,
segmentation, the study showcasing consistent performance
processes 500 images from 141 across various experimen t al
shear-critical specimens, conditions and configurations. The
employing thresholding and successful linkage between surface
edge detection to extract crack observations and quantitative
patterns' textural and geometric estimation of damage levels,
attributes. considering stiffness, load, and
displacement, indicates the potential
feasibility of employing image -
based damage recognition for visual
inspections of exterior beam-colu mn
joints in structural assessments.

[95] ML MATLAB This study addresses the Five seismic ground motions are The correlation established 2023
deterioration of civil structures prepared, and the autoencoder is between reconstruction error
in Japan by proposing an trained using linear displacement magnitude and the degree of
autoencoder-based damage responses from adjusted motions. damage demonstrates the
detection approach using The study demonstrates a method's potential for
machine learning. Utilizing a relationship between the accurate and efficient
Multi-Degree-Of-Freedom reconstruction error magnitude and damage detection.
(MDOF) system model the degree of structural damage,
subjected to seismic responses, indicating the potential of the
the study explores the correlation proposed method for detecting
between the magnitude of damage in civil structures.
reconstruction error and the
degree of structural damage. The
autoencoder, a three-layer neural
network trained on normal data,
is employed for unsupervised
learning to distinguish between
normal and abnormal structural
responses.

20
 Computer Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[96] ML Python At the beginning, the authors It was found that ML can be No sufficient discussion was 2023
explained how a physical employed to minimize or even provided regarding the
concept of ML can be translated eliminate the errors on noisy applied NN.
into practical work in intermediate-scale quantum devices.
engineering. They explored the
ability of hybrid quantum-
classical NN to enhance the
accuracy of a model and reduce
computational resources.

[97] AI --- The author proposed a method to It was found that AI along with This paper analyzed the 2023
investigate and identify how the knowledge management can be used already existing data, but no
implementation of knowledge to improve software engineering prediction step was
management principles in activities such as error tracking, conducted regarding the
software engineering could version control, monitoring software future situation.
enhance the future documentation, and onboarding new
transformation of software- members.
related jobs. Some quantitative
data were analyzed to evaluate
the AI role.

[98] DL Python The authors conducted an The outcomes showed that DL tool It needs more discussion and 2023
empirical study of how the DL could achieve the target effectively. clarification regarding the
can be used for sentiment error categories of DL
detection tools for software
engineering.

[99] ML and Python This article proposed a method The outcomes showed that the The methodology and the 2023
that combines several algorithms phishing attack could be detected by result sections were
DL
and made the evaluation as well software engineering by the ML discussed in detail, which is
the validation based on several algorithms such as NB and SVM in a good contribution.
criteria such as the precision, F1 addition to DL algorithm such as
score and the recall. LSTM.
Chemical Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[100] ML --- The study constructed a database The utilization of ML techniques The study's comprehensive 2022
of Ionic Liquid encapsulated revealed essential factors database creation and
Metal-Organic Framewo rk influencing CO2 /N2 selectivity and utilization of molecular
(IL@MOF) composites. CO2 permeability, showcasing simulation and machine
Molecular simulation was substantial improvements in CO2 learning techniques offer a

21
 employed to identify the top- permeability and selectivity, robust framework for
performing composites and surpassing known performance identifying optimal
elucidate the gas separation boundaries. IL@MOF composites.
mechanism.

[101] ML --- The study employs high- The highest working capacity and This study contribute 2020
throughput computational coefficient of performance achieved significantly to the field's
screening of 6013 metal-organic under heat pump operating understanding of efficient
frameworks (MOFs) to identify conditions were 512.86 mg/g and working fluid and adsorbent
optimal methanol-MOF pairs for 1.83, respectively, and the study combinations for enhanced
adsorption-driven heat pumps provides insights into optimal ranges heat pump applications.
and chillers, establishing of MOF descriptors for enhancing
structure-performance efficiency in adsorption-driven heat
relationships based on 6 MOF pump applications using methanol as
descriptors and using four a working fluid.
machine learning algorithms to
predict performance metrics.

[102] ML --- The study computed and Univariate analysis highlighted the The study demonstrates a 2023
simulated metal–organic pivotal role of pore limiting diameter novel methodology for
framework membranes MOFs in separation, and comparisons evaluating numerous MOF
from the CoRE-MOF database between 3D MOF membranes membranes for Kr/Xe
for Kr/Xe separation under delineated optimal permeability and separation, offering a
atmospheric pressure. Univariate permselectivity ranges, offering promising direction for
analysis identified key structural critical insights for membran e future membrane design and
descriptors, and five machine design enhancement. improvement.
learning algorithms were applied
for classification and regression.

[103] ML --- The study employed machine The results showcased the successful The integration of ML 2023
learning models, particularly the application of machine learning models offers a significant
gradient boosting decision tree predictions in guiding the design and advantage by streamlining
(GBDT) algorithm, to predict the creation of polymer membranes , the design and development
influence of polymer functional ultimately leading to improved of specialized membranes
groups on the conductivities of battery performance metrics in the for battery applications,
zinc ions (Zn2+) and protons. Zn/MnO2 flow battery system. providing a targeted and
efficient approach in
material synthesis.

[104] ML --- The study leveraged high- The study identified top-performin g --- 2022
throughput computational metal–organic framewo rk
screening (HTCS) and ML to membranes (MOFMs) for H2
assess hydrogen separation separation, indicating potential
efficiency across computation- practical application. Integration of
ready metal–organic framewo rk high-throughput computational
membranes (CoRE-MOFMs) for screening (HTCS) and machin e
various gas pairs. Eight learning (ML) accelerated CoRE-
supervised ML algorithms were MOFMs evaluation, offering a rapid
employed to predict the material discovery approach for gas
separation performance. separation technologies.

22
 Electrical Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[105] ML LSTM This study aimed to show how The study showed that the existing The study involved a fuzzy 2023
the data since can by applied in traditional management and control logic control system but no
electrical engineering field. The (M&C) concepts of smart power sufficient description were
role and the application of the plants should be improved. In provided.
data science (DS) and ML on addition, this paper pointed out the
edge, fog, and cloud levels of advantages of the vertical IoT
vertical industrial IoT concepts concept and described all IoT levels
of power plants were presented (edge, fog, and cloud).
by investigating the
requirements and expectations of
IoT. Moreover, the limitations in
signal conversion were also
considered described.

[106] AI and --- This study proposed a Recurrent According to the evaluation metrics , This study did not include 2023
Neural Network (RNN) RMSE, MAE, and R2, the proposed anything about data
ML
forecasting model incorporating RNN-DFBER-based model collection and preprocessing
a Dynamic Fitness Al-Biru n i outperforms the other models phases.
Earth Radius (DFBER) considered. This suggests that the
algorithm in order to forecast and RNN model, combined with the
predict wind power data DFBER algorithm, predicts wind
patterns. Several popular models power data patterns more effectively
were used to test the than the alternative models.
performance of the model such
as Grey Wolf Optimizer
(GWO), Whale Optimizat io n
Algorithm (WOA), Particle
Swarm Optimization (PSO), Fire
Hawk Optimizer (FHO), and
Jaya Algorithm (JAYA).

[107] DL Python, In this study, a novel approach The outcomes showed that the This study combined the 2023
was proposed for online cyber introduction section with the
proposed method could identify
attack situational awareness that related works. In general, it
Linux
improves the reliability of the active is better to separate these two
hping3, power grid. The proposed sections to give a good
approach employed a knowledge for the reader
and Linux attack locations (control center,
combination of DL then introducing the
within substations, and wide area
bwm-ng convolutional network for time previous related work in the
network) with an accuracy
series classification-based following section.
percentage exceeding 96%.
anomaly detection and DL
model of Graph Convolutional
Long Short-Term Memory

23
[108] DL --- This study applied the DL in It was found that it is not easy to This study suffered from a 2023
power quality enhancement, it apply the DL in the power quality lack of novelty. Moreover, it
showed that most of the and the experts need to enrich their mentioned the research gabs
applications of DL depend on knowledge even with automatic in this field but it did not
supervised learning and feature extraction by DL handle it.
unrealistic synthetic data with no approaches. Further, the research
proper labeling. gaps in this field were introduced in
this study which are related to the
semi-supervised learning
applications, explainable DL and
hybrid approaches combining DL
with expert systems.

[109] ML and --- This study aimed to forecast It was found that the hybrid models, The study showed the gaps 2023
renewable energy (RE) which consist of traditional time - existing in the research field
DL
generation by using ML and DL series analysis along with ML and without solving them.
techniques by reviewing the DL techniques are used for RE
different methods and model that forecasting purposes effectively. On
were used for RE prediction. The the other hand, the design of a hybrid
limitations, advantages, and the model is not a trivial task.
strengths of each model were
discussed too.

[110] AI and Matlab This paper aimed to estimate the The study showed the superiority of In this study, the authors 2022
extractable power from wind in SA compared with the rest showed their contribution in
ML
several sites in Jordan by algorithms. Moreover, Normal PDF employing several
applying the AI techniques (GA , showed its good performance in techniques in wind energy
BFOA, SA, and ANFIS) based estimating the extractable energy estimating.
on Weibull and Normal PDFs. from wind.
Moreover, a Matlab tool was
used to estimate the accuracy of
the parameters of the
aforementioned PDFs.
Electronic Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[111] ML Matlab The authors demonstrated how The outcomes showed that there is No related works were 2021
effective is the application of the an optimal size for the solder joint included in this study.
ML in the solder joint geometry that will maximize the lifetime of the
of electronic components by electronic component.
showing how the lifetime of
these components will be
enhanced.

[112] ML Pymatgen The authors proposed a method the outcomes showed that the No related works were 2023
to predict the band gap of proposed method has an accuracy included in this study.
graphene in heterostructures. rate of 88.3% and the difference
They employed an ML learning between the predicted and the
model on a dataset of 2076 calculated values was ignorable.

24
 binary compounds, followed by
(Python the usage of two main ML
models to screen semiconductor
Materials
components and form
Genomics) heterostructures with graphene.

Environmental Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[113] AI and --- The study compared a method This study showed how employing The study addresses a crucial 2020
based on deep neural networks deep neural networks outperformed issue in WWTPs by focusing
DL
with statistical and traditional traditional methods, obtaining a fault on fault detection, essential
machine learning methods to detection rate exceeding 92% in for optimizing treatment
record the temporal patterns in WWTPs. This gives the ability to efficiency and energy
sensor data present in enhance monitoring efficiency , consumption.
wastewater treatment plants enabling timely responses to faults
(WWTPs) and optimizing treatment processes
while minimizing environmental
impact.

[114] AI and --- The authors used neural network The study demonstrated a successful The study showcased 2021
ML to predict the amount of waste; waste prediction and reduction considerable promise for
and machine learning algorith m models. It achieved a substantial enhancing waste
to improve waste collection. reduction of waste quantity by 90%, management efficiency and
landfill analysis by 40%, and advancing the cause of
transportation by 15%. sustainability within society.

[115] AI and --- The study employed an The model effectively captured The existing IFFNN model 2022
ML improved feedforward neural complex nonlinear connections and doesn't include an optimal
network (IFFNN) integrated extrapolated information accurately. control system to
with an optimization algorith m These findings indicated significant simultaneously reduce real-
to predict WWTP effluent enhancements in prediction time operational expenses in
quality by comparing it with the accuracy and and suggested WWTP while meet the
traditional feedforward neural potential applicability for immediat e effluent criteria.
network (FFNN). regulatory decision-making in
WWTP operations.

[116] AI and --- The study used an electrically- The integrated system, combining --- 2016
ML enhanced membrane bioreactor electrochemical treatment and
setup to treat medium strength membrane bioreactor, significantly
wastewater. An artificial neural reduced wastewater contaminant
networks (ANNs) based concentrations, achieving high
ensemble model effectively removal rates for COD (98%),
simulated experimen t al PO43−-P (99%), and NH4+-N
outcomes, showing strong (98%), highlighting its effectiveness
correlations. in treating municipal wastewater.

[117] AI --- The authors developed artificial The ANN models demonstrated The study's acknowledgment 2004
neural networks models to efficiency in predicting BOD and SS of lower accuracy in SS
forecast the efficiency of a concentrations in the plant effluent, predictions due to potential

25
 WWTP using historical data. achieving R2 values between 0.63 to data noise underscores the
They evaluated biochemical 0.81 for BOD and 0.45 to 0.65 for SS promising capability of
oxygen demand (BOD) and across multiple data groups. ANNs as a valuable tool for
suspended solids (SS) forecasting wastewater
concentrations. treatment plant performance.
This indicates the potential
for improved accuracy with
the availability of more
comprehensive and less
noisy datasets in future
research.

[118] ML and Python The study developed an Implementation of the VSL-based --- 2023
automated framework called ML models effectively improved
DL
variation sliding layer (VSL) for aeration control efficiency in
precise aeration control in WWTPs, showing notable
WWTPs by applying ML. The enhancements in predictive
VSL was integrated into classic accuracy for air demand compared
ML, deep learning, and to traditional mechanistic models.
ensemble learning models to The study showcased the potential
enhance aeration control to revolutionize AI technology
efficiency. application in WWTPs for precise
air demand prediction and control,
aiming at energy conservation and
reduction in consumption.
Industrial Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[119] ML --- The research employs a Addressing the need for zero-defect --- 2018
knowledge-based intelligent policies, the study introduces a
supervisory system for rare strategy for rare quality event
quality event detection in detection. Validated in automotive
manufacturing. Using l1- manufacturing, the approach,
regularized logistic regression as utilizing l1-regularized logistic
the learning algorithm, a hybrid regression, achieves 100% defect
feature elimination algorith m detection. By eliminating irrelevant
and optimal threshold search features and redundancy, the
support defect detection. proposed strategy enhances
classifier performance, contributing
to advancing quality standards in
manufacturing.

[120] ML --- The paper employs supervised The machine learning model --- 2020
machine learning classification demonstrates excellent accuracy
to predict production batch (0.99) and reliability (Cohen’s
compliance quality in Kappa of 0.91) in predicting unit
manufacturing. Utilizin g batch compliance quality.
classification algorithms, the Emphasizing the significance of
method incorporates feature feature construction and dataset
construction and dataset knowledge, the proposed method

26
 knowledge, validated through a proves effective in preemptively
multi-model appliance managing defective product batches,
production line case study. The leading to substantial cost savings in
model achieves high accuracy recalls and logistics.
(0.99) and Cohen’s Kappa (0.91)
for predicting compliance
quality.
Mechanical Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[121] ML Python The study compared Design of Low-cost scenarios favored random The study provides a 2023
Experiments (DOE) with Active sampling, while limited experimen t s comprehensive comparison
Learning (AL) in mechanical showed superiority for Latin between DOE and AL
engineering data collection, Hypercube Sampling (LHS) and AL methods in data collection
using three cases to evaluate methods. LHS displayed consistent for ML models, offering
sample efficiency, stability, and accuracy and stability but required practical insights into
predictive accuracy of ML prior sample count definition , sampling strategies for
models generated by AL and whereas QBC proved most efficien t engineers and data scientists.
DOE techniques. and stable when compatible with However, it's focus on
bagging techniques, especially in specific techniques like LHS
high-cost data acquisition settings. and QBC might limit its
generalizability.

[122] ML --- The study employed ML PCA efficiently reduced features The use of various ML 2020
techniques on sensor data to without significantly impactin g methods to handle large
evaluate machining process and accuracies, with classification datasets and identify failure
machine tool health, including techniques achieving nearly 100% modes is commendable,
supervised and unsupervised accuracy for most failure modes. offering a potential path for
techniques like principal Gaussian Mixture Models (GM M ) real-time detection of faults
component analysis (PCA) and showed consistency in categorizing in production environments.
continuous wavelet transform failures, and custom-made CNNs
(CWT), with custom-made and demonstrated comparable or higher
pre-trained convolutional neural accuracy than pre-trained networks,
networks (CNNs). though simulations might limit real-
world performance evaluation.

[123] ML --- The study diagnoses gear box It concludes that proximal support The comparative analysis of 2021
faults by analyzing vibration vector machines (PSVM) exhibit s classification techniques
signals through wavelet-based superior classification accuracy over offers insights into the
features, captured under various artificial neural networks (ANN) efficiency of PSVM over
gear conditions. It evaluats when classifying the extracted ANN for accurate fault
Morlet wavelet's effectiveness in features, emphasizing PSVM's identification, which is a
feature extraction. advantage in accurately identifying notable strength.
faults in the gear box.

[124] ML --- The study focuses on gear and The developed ANN-based system It effectively underscores the 2007
bearing fault recognition in demonstrated high efficiency and importance of preprocessing
gearboxes using a multi-lay er 100% accuracy in identifying gear methods like wavelet
perceptron neural network. It failures and detecting bearing transform for enhanced fault
innovates a feature vector defects, employing a small-sized detection accuracy,

27
 through wavelet packet multi-layer perceptron network contributing to the field's
coefficients' standard deviation structure. advancement.
from vibration signals,
facilitating fault identification
with high accuracy.

[125] DL --- The study proposes NSAE-LCN, The research showcases NSA E- the reliance on manually 2018
integrating normalized sparse LCN's success in generating determined parameters in
autoencoder (NSAE) within a dissimilar and meaningful learned NSAE-LCN could benefit
local connection network (LCN), features while accurately from future optimizatio n
to automate feature extractio n recognizing mechanical health using techniques for
and fault diagnosis in machines conditions. Comparative analysis improved adaptability and
using vibration signals. with conventional diagnostic performance, highlighting an
methods validates the superiority of avenue for further
NSAE-LCN, indicating its promise development.
in intelligent fault diagnosis for
machinery.
Telecommunication Engineering
Ref Used Used Methodology Summary of the outcomes Comments Year
# technique
tool
(AI/ ML/
DL)
[126] ML Python In this study, the authors The outcomes showed that the This study discussed the 2023
proposed a method for proposed method achieved methodology section
telecommunication fraud 89.98%F1- score. Moreover, it was stepwise in a perfect way.
detection by a convolutional stated that the performance of the
neural network (CNN). The data model can be improved to
used were images that were 91.09%F1-score.
applied to computer vision
techniques such as image
augmentation.

[127] ML Python In this work, the authors The proposed system showed a good In this study, the cross 2023
and C++ proposed a distributed performance and the objective was validation was set to 20,
congestion control mechanis m achieved. which gives an accurate
for generic multi-hop systems. model.
Several data types were
considered and applied by ML
algorithms, certainly, the
CatBoost algorithm.

[128] ML Python The authors provided an The outcomes showed that The gaps in the mentioned 2023
overview of 5G network slicing telecommunication engineers could related works were not
and applied some ML models, create optimized network slices to overcome. However, the
the accuracy of each model, of minimize the response time with paper explained the code of
how effective it is in detecting high throughput for applications. applying the ML in this field
network slices, was investigated. from scratch.

[129] DL Python In this study, the authors The outcomes showed that the The related work section did 2023
proposed a hybrid quantum DL proposed model achieved an not cover wide range of
model that compared between accuracy of 97.16% recent studies.

28
 the CNN and the Recurrent
Neural Network (RNN).

[130] AI --- The authors proposed a novel The outcomes showed that several The outcomes from this 2023
method for smart healthcare challenges and constraints are meet research depended on strong
systems by employing the 5G the 5G methods such as performance well-known mathematical
along with the Industry 4.0 tradeoffs. equations in the field instead
standards. The proposed method of using the software.
used the network simulator N2S.

3. Statistical Outcomes
Upon conducting a thorough review, it was discerned that the application of AI, ML, and DL techniques
has exhibited commendable performance across various engineering domains. Figure 15 visually
encapsulates the distribution percentages of each technique, drawing insights from the compilation of
studies delineated in the literature. The depicted percentages provide a comprehensive overview of the
prevalence and effectiveness of AI, ML, and DL methodologies within the context of the examined
engineering applications. This representation serves as a valuable synthesis, encapsulating the collective
impact and efficacy observed in the diverse studies encompassing these advanced technological approaches.

Percentages of Al, ML, and DL application in

Engineering Fields

23.81%

23.81%
73.02%

ML AI DL

Figure 15: How wide are the AI, M L, and DL in engineering fields?

ML stands out as the most influential technique in engineering applications when juxtaposed with
its counterparts. This prominence is justifiably attributed to the relative ease with which ML algorithms can
be implemented. The pragmatic advantage of ML lies in its ability to deliver commendable performance
without necessitating the intricate complexities and substantial computational resources demanded by DL.
It is worth noting that the selection between Artificial AI, ML, and DL hinges on several factors, including

29
the availability of data, the specific requirements of the task at hand, and the computational resources at
one's disposal. The decision-making process, therefore, becomes a dynamic interplay of these
considerations. And finally, it is important to underscore that the percentages presented were derived from
a comprehensive analysis of studies documented in Table 3, and it is acknowledged that these figures may
vary from one year to the next based on evolving research and technological advancements.
ML techniques are widely favored in research papers for several reasons. Firstly, ML offers a
versatile toolkit of algorithms that can be applied across various domains and research problems,
accommodating a wide range of applications. Secondly, these techniques have been developed and refined
over several decades, providing researchers with a mature set of tools to tackle complex problems.
Additionally, ML algorithms often offer better interpretability compared to DL, which is crucial for
understanding model decisions in fields like healthcare or law. Moreover, ML models can sometimes
achieve good performance with smaller datasets compared to DL, making them more practical in scenarios
where data collection is limited or expensive. Overall, the combination of versatility, interpretability, and
historical development makes Machine Learning a prominent choice for researchers exploring new
applications and solving real-world problems.
Future research in AI, ML, and DL applications in engineering presents several promising avenues
for exploration. One key direction involves advancing interpretability and explainability of AI models,
crucial for enhancing trust and adoption in engineering applications. Additionally, exploring novel
architectures and algorithms tailored to engineering-specific challenges, such as real-time processing and
robustness to noisy data, will be pivotal. Integrating AI with emerging technologies like IoT and edge
computing holds potential for decentralized and efficient engineering solutions. Moreover, ethical
considerations surrounding AI deployment, including bias mitigation and privacy preservation, will require
ongoing attention. Finally, leveraging AI for automated design optimization and predictive maintenance
could revolutionize engineering practices, offering substantial efficiency gains and cost savings.

Conclusion

In conclusion, this extensive investigation has illuminated the extensive applications of AI, ML, and
DL across 14 distinct engineering disciplines. The research commenced with a precise definition of AI,
ML, and DL, facilitating a thorough exploration of their varied applications. A meticulous systematic
review of pertinent studies was conducted, examining critical aspects such as the specific technique
employed (AI, ML, or DL), tools or software utilized, adopted methodologies, and providing concise
summaries of the outcomes. The culmination of these efforts indicates that ML is the predominant
technique, accounting for nearly 73% of the studies in the engineering research domain. Each study
underwent rigorous examination, accompanied by insightful commentary, offering readers a nuanced
understanding of the methodologies and outcomes within each specific field.

This research not only presents a comprehensive overview of AI, ML, and DL applications in
engineering but also highlights the prevailing dominance of ML as the most commonly adopted technique
in contemporary research endeavors. The findings enhance our understanding of the current landscape and
provide valuable insights for future research directions at the intersection of artificial intelligence and
engineering disciplines. Moreover, it was found that studies combining multiple techniques yielded superior
outcomes.

30
 References
[1] Chaudhari, S. A. (2023). TRENDS IN ARTIFICIAL INTELLIGENCE, M ACHINE LEARNING AND DEEP LEARNING.
Artificial Intelligence, 52(4).

[2] H. Wang et al., “Scientific discovery in the age of artificial intelligence,” Nature, vol. 620, no. 7972, pp. 47–60, Aug. 2023,
doi: [Link]

[3] M . Khaleel, A. A. Ahmed, and A. Alsharif, “Artificial Intelligence in Engineering,” Brilliance: Research of Artificial
Intelligence, vol. 3, no. 1, pp. 32–42, M ar. 2023, doi: [Link]

[4] C. Surianarayanan, J. J. Lawrence, P. R. Chelliah, E. Prakash, and C. Hewage, “A Survey on Optimization Techniques for Edge
Artificial Intelligence (AI),” Sensors, vol. 23, no. 3, p. 1279, Jan. 2023, doi: [Link]

[5] U. T. Jagadale, C. B. Nayak, A. M ankar, S. B. Thakare, and W. N. Deulkar, “An experimental-based python programming for
structural health monitoring of non-engineered RC frame,” Innovative Infrastructure Solutions, vol. 5, no. 1, Jan. 2020, doi:
[Link]

[6] OpenAI, “ChatGPT,” [Link], 2023. [Link]

[7] T. Wu et al., “A Brief Overview of ChatGPT: The History, Status Quo and Potential Future Development,” IEEE/CAA Journal
of Automatica Sinica, vol. 10, no. 5, pp. 1122–1136, M ay 2023, doi: [Link]

[8] J. Kocoń et al., “ChatGPT: Jack of All Trades, M aster of none,” Information Fusion, vol. 99, pp. 101861–101861, Nov. 2023,
doi: [Link]

[9] Q. Zhong, L. Ding, J. Liu, B. Du, and D. Tao, “Can ChatGPT Understand Too? A Comparative Study on ChatGPT and Fine-
tuned BERT,” arXiv (Cornell University), Feb. 2023, doi: [Link]

[10] Ibrahim Adeshola and Adeola Praise Adepoju, “The opportunities and challenges of ChatGPT in education,” Interactive
Learning Environments, pp. 1–14, Sep. 2023, doi: [Link]

[11] Jordan, M . I., & M itchell, T. M . (2015). M achine learning: Trends, perspectives, and prospects. Science, 349(6245), 255-260.

[12] D. Kreuzberger, N. Kühl, and S. Hirschl, “M achine Learning Operations (M LOps): Overview, Definition, and Architecture,”
IEEE Access, pp. 1–1, 2023, doi: [Link]

[13] A. Kumar and N. M or, “An Approach-Driven: Use of Artificial Intelligence and Its Applications in Civil Engineering,” Studies
in Big Data, pp. 201–221, 2021, doi: [Link]

[14] A. Aldahiri, B. Alrashed, and W. Hussain, “Trends in Using IoT with M achine Learning in Health Prediction System,”
Forecasting, vol. 3, no. 1, pp. 181–206, M ar. 2021, doi: [Link]

[15] S. Rajbanshi, “M achine Learning Algorithms | Introduction to M achine Learning,” Analytics Vidhya, M ar. 25, 2021.
[Link]

[16] D. Solyali, “A Comparative Analysis of M achine Learning Approaches for Short-/Long-Term Electricity Load Forecasting in
Cyprus,” Sustainability, vol. 12, no. 9, p. 3612, Apr. 2020, doi: [Link]

[17] Osisanwo, F.; Akinsola, J.; Awodele, O.; Hinmikaiye, J.; Olakanmi, O.; Akinjobi, J. Supervised machine learning algorithms:
Classification and comparison. Int. J. Comput. Trends Technol. 2017, 48, 128–138.

[18] Praveena, M .; Jaiganesh, V. A Literature Review on Supervised M achine Learning Algorithms and Boosting Process. Int. J.
Comput. Appl. 2017, 169, 32–35, doi:10.5120/ijca2017914816.

[19] Friedman, J.; Hastie, T.; Tibshirani, R. The Elements of Statistical Learning (no. 10); Springer Series in Statistics: N ew York,
NY, USA, 2001.

[20] A. M . Rahmani et al., “Artificial intelligence approaches and mechanisms for big data analytics: a systematic study,” PeerJ
Computer Science, vol. 7, p. e488, Apr. 2021, doi: [Link]

[21] Dike, H.U.; Zhou, Y.; Deveerasetty, K.K.; Wu, Q. Unsupervised Learning Based on Artificial Neural Network: A Review. In
Proceedings of the 2018 IEEE International Conference on Cyborg and Bionic Systems (CBS), Shenzhen, China, 25–27
October 2018; pp. 322–327.

31
[22] Tambuskar, D.; Narkhede, B.; M ahapatra, S.S. A flexible clustering approach for virtual cell formation considering real-life
production factors using Kohonen self-organising map. Int. J. Ind. Syst. Eng. 2018, 28, 193–215.

[23] Ben Ali, J.; Saidi, L.; Harrath, S.; Bechhoefer, E.; Benbouzid, M . Online automatic diagnosis of wind turbine bearings
progressive degradations under real experimental conditions based on unsupervised machine learning. Appl. Acoust. 2018,
132, 167–181, doi:10.1016/[Link].2017.11.021.

[24] S. K. Sahu, A. M okhade, and N. D. Bokde, “An Overview of M achine Learning, Deep Learning, and Reinforcement Learning-
Based Techniques in Quantitative Finance: Recent Progress and Challenges,” Applied Sciences, vol. 13, no. 3, p. 1956, Feb.
2023, doi: [Link]

[25] “Top 10 Algorithms every M achine Learning Engineer should know - GeeksforGeeks,” GeeksforGeeks, M ay 22, 2019.
[Link] -10-algorithms-every-machine-learning-engineer-should-know/

[26] Saadaldeen Rashid Ahmed, Aso Kurdo Ahmed, and Swran Jawamir Jwmaa, “Analyzing The Employee Turnover by Using
Decision Tree Algorithm,” Jun. 2023, doi: [Link]

[27] H. Blockeel, L. Devos, B. Frénay, G. Nanfack, and S. Nijssen, “Decision trees: from efficient prediction to responsible AI,”
Frontiers in artificial intelligence, vol. 6, Jul. 2023, doi: [Link]

[28] “Predicting financial performance for listed companies in Thailand during the transition period: A class-based approach using
logistic regression and random forest algorithm,” Journal of Open Innovation: Technology, M arket, and Complexity, vol.
9, no. 3, p. 100130, Sep. 2023, doi: [Link]

[29] A. M ehta, “Why Is Logistic Regression Called‘Regression’ If It Is A Classification Algorithm?,” M edium, Nov. 30, 2020.
[Link] -is-logistic-regression-called-regression-if-it-is-a-classification-algorithm-9c2a166e7b74

[30] M . P. Behera, A. Sarangi, D. M ishra, and S. K. Sarangi, “A Hybrid M achine Learning algorithm for Heart and Liver Disease
Prediction Using M odified Particle Swarm Optimization with Support Vector M achine,” Procedia Computer Science, vol.
218, pp. 818–827, 2023, doi: [Link]

[31] Y. Xu, B. Klein, G. Li, and B. Gopaluni, “Evaluation of logistic regression and support vector machine approaches for XRF
based particle sorting for a copper ore,” M inerals Engineering, vol. 192, p. 108003, Feb. 2023, doi:
[Link]

[32] Sharifani, K., & Amini, M . (2023). M achine Learning and Deep Learning: A Review of M ethods and Applications. World
Information Technology and Engineering Journal, 10(07), 3897-3904.

[33] Amini, M ahyar and Ali Rahmani. "Achieving Financial Success by Pursuing Environmental and Social Goals: A
Comprehensive Literature Review and Research Agenda for Sustainable Investment." World Information Technology and
Engineering Journal 10.04 (2023): 1286-1293.

[34] Zhou, Teng, Zhen Song, and Kai Sundmacher. "Big data creates new opportunities for materials research: a review on methods
and applications of machine learning for materials design." Engineering 5.6 (2019): 1017-1026.

[35] Amini, M ahyar, and Zavareh Bozorgasl. "A Game Theory M ethod to Cyber-Threat Information Sharing in Cloud Computing
Technology ." International Journal of Computer Science and Engineering Research 11.4 (2023): 549-560.

[36] Deng, Li, and Dong Yu. "Deep learning: methods and applications." Foundations and trends® in signal processing 7.3–4
(2014): 197-387.

[37] Nazari Enjedani, Somayeh, and M ahyar Amini. "The role of traffic impact effect on transportation planning and sustainable
traffic management in metropolitan regions." International Journal of Smart City Planning Research 12, no. 2023 (2023): 688-
700.

[38] Chai, Junyi, et al. "Deep learning in computer vision: A critical review of emerging techniques and application scenarios."
M achine Learning with Applications 6 (2021): 100134.

[39] Jahanbakhsh Javidi, Negar, and M ahyar Amini. "Evaluating the effect of supply chain management practice on implementation
of halal agroindustry and competitive advantage for small and medium enterprises ." International Journal of Computer
Science and Information Technology 15.6 (2023): 8997-9008

[40] Pouyanfar, Samira, et al. "A survey on deep learning: Algorithms, techniques, and applications." ACM Computing Surveys
(CSUR) 51.5 (2018): 1-36.

32
[41] Amini, M ahyar, and Negar Jahanbakhsh Javidi. "A M ulti-Perspective Framework Established on Diffusion of Innovation
(DOI) Theory and Technology, Organization and Environment (TOE) Framework Toward Supply Chain M anagement System
Based on Cloud Computing Technology for Small and M edium Enterprises ." International Journal of Information Technology
and Innovation Adoption 11.8 (2023): 12171234

[42] Amini, M ahyar, et al. "Development of an instrument for assessing the impact of environmental context on adoption of cloud
computing for small and medium enterprises." Australian Journal of Basic and Applied Sciences (AJBAS) 8.10 (2014): 129-
135.

[43] M ahrishi, M ehul, et al., eds. M achine Learning and Deep Learning in Real-Time Applications. IGI global, 2020.

[44] Amini, M ahyar, et al. "The role of top manager behaviours on adoption of cloud computing for small and medium enterprises."
Australian Journal of Basic and Applied Sciences (AJBAS) 8.1 (2014): 490-498.

[45] M athew, Amitha, P. Amudha, and S. Sivakumari. "Deep learning techniques: an overview." Advanced M achine Learning
Technologies and Applications: Proceedings of AM LTA 2020 (2021): 599-608.

[46] Amini, M ahyar, and Nazli Sadat Safavi. "A Dynamic SLA Aware Solution For IaaS Cloud Placement Problem Using
Simulated Annealing." International Journal of Computer Science and Information Technologies 6.11 (2014): 52-57.

[47] Deng, Li, and Dong Yu. "Foundations and Trends in Signal Processing: DEEP LEARNING–M ethods and Applications."
(2014).

[48] Sadat Safavi, Nazli, Nor Hidayati Zakaria, and M ahyar Amini. "The risk analysis of system selection and business process re-
engineering towards the success of enterprise resource planning project for small and medium enterprise." World Applied
Sciences Journal (WASJ) 31.9 (2014): 1669-1676.

[49] B. Kenwright, “Is It the End of Undergraduate Dissertations?,” Advances in educational technologies and instructional design
book series, pp. 46–65, Dec. 2023, doi: [Link]

[50] Y. (Gina) Cui, P. van Esch, and S. Phelan, “How to build a competitive advantage for your brand using generative AI,”
Business Horizons, M ay 2024, doi: [Link]

[51] A. M ilinković, D. Vuleta, and i Tihana Babić, “Student Perception of Using Generative AI Tools in Relation to Academic
Integrity and Their Advantages and Disadvantages,” M ay 2024, doi: [Link]

[52] A. Konya and P. Nematzadeh, “Recent applications of AI to environmental disciplines: A review,” Science of The Total
Environment, vol. 906, p. 167705, Jan. 2024, doi: [Link]

[53] E. Osadchaya et al., “To ChatGPT, or not to ChatGPT: Navigating the paradoxes of generative AI in the advertising industry,”
Business horizons, M ay 2024, doi: [Link]

[54] P. Dave, “How AI Can Revolutionize the Pharmaceutical Industry,” Journal of drug delivery and therapeutics, vol. 14, no. 6,
pp. 179–183, Jun. 2024, doi: [Link]

[55] Imdad Ali Shah, Noor Zaman Jhanjhi, and Sarfraz Nawaz Brohi, “Use of AI-Based Drones in Smart Cities,” Advances in
information security, privacy, and ethics book series, pp. 362–380, Jan. 2024, doi: [Link]
8.ch015.

[56] M ini Han Wang et al., “AI-Based Advanced Approaches and Dry Eye Disease Detection Based on M ulti-Source Evidence:
Cases, Applications, Issues, and Future Directions,” Big data mining and analytics, vol. 7, no. 2, pp. 445–484, Jun. 2024, doi:
[Link]

[57] E. C. S. Ku and C.-D. Chen, “Artificial intelligence innovation of tourism businesses: From satisfied tourists to continued
service usage intention,” International Journal of Information M anagement, p. 102757, Jan. 2024, doi:
[Link]

[58] Amer F.A.H. Alnuaimi and Tasnim H.K. Albaldawi, “An overview of machine learning classification techniques,” Bio web
of conferences/BIO web of conferences, vol. 97, pp. 00133–00133, Jan. 2024, doi:
[Link]

[59] D. Theng and K. K. Bhoyar, “Feature selection techniques for machine learning: a survey of more than two decades of
research,” Knowledge and Information Systems, Dec. 2023, doi: [Link]

[60] R. Shang et al., “M achine Learning‐Enhanced Triboelectric Sensing Application,” Advanced M aterials Technologies, Feb.
2024, doi: [Link]

33
[61] Gorapalli Srinivasa Rao and G M uneeswari, “A Review: M achine Learning and Data M ining Approaches for Cardiovascular
Disease Diagnosis and Prediction,” EAI endorsed transactions on pervasive health and technology, vol. 10, M ar. 2024, doi:
[Link]

[62] Faisal Dharma Adhinata, None Wahyono, and Raden Sumiharto, “A comprehensive survey on weed and crop classification
using machine learning and deep learning,” Artificial intelligence in agriculture, Jun. 2024, doi:
[Link]

[63] A. Aramburu, C. Guido, P. Bares, B. Pla, P. Napolitano, and C. Beatrice, “Knock detection in spark ignited heavy duty engines:
An application of machine learning techniques with various knock sensor locations,” M easurement, vol. 224, pp. 113860–
113860, Jan. 2024, doi: [Link]

[64] L. Yang, M . Tian, D. Xin, Q. Cheng, and J. Zheng, “AI-Driven Anonymization: Protecting Personal Data Privacy While
Leveraging M achine Learning,” [Link], Feb. 26, 2024. [Link]

[65] Yasir Afaq and A. M anocha, “Blockchain and Deep Learning Integration for Various Application: A Review,” ˜The œjournal
of computer information systems/˜The œJournal of computer information systems, vol. 64, no. 1, pp. 92–105, Feb. 2023, doi:
[Link]

[66] J. Zhang, L. Xiao, Y. Zhang, J. Lai, and Y. Yang, “Optimization and Performance Evaluation of Deep Learning Algorithm in
M edical Image Processing,” Frontiers in computing and intelligent systems, vol. 7, no. 3, pp. 67–71, Apr. 2024, doi:
[Link]

[67] C. Zhang, N. Nur, and R. Ibrahim, “Deep learning models for price forecasting of financial time series: A review of recent
advancements: 2020–2022,” WIREs Data M ining and Knowledge Discovery, Sep. 2023, doi:
[Link]

[68] R. Bai et al., “Analytics and machine learning in vehicle routing research,” International Journal of Production Research, pp.
1–27, Dec. 2021, doi: [Link]

[69] N. Tudoroiu, M . Zaheeruddin, R.-E. Tudoroiu, M . S. Radu, and H. Chammas, “Intelligent Deep Learning Estimators of a
Lithium-Ion Battery State of Charge Design and M ATLAB Implementation—A Case Study,” Vehicles, vol. 5, no. 2, pp. 535–
564, Jun. 2023, doi: [Link]

[70] N. Nabi, B. Ray, F. Rashid, W. K. Hussam, and S. M . M uyeen, “Parametric analysis and prediction of energy consumption of
electric vehicles using machine learning,” Journal of Energy Storage, vol. 72, pp. 108226–108226, Nov. 2023, doi:
[Link]

[71] I. Ullah, K. Liu, T. Yamamoto, M . Shafiullah, and A. Jamal, “Grey wolf optimizer-based machine learning algorithm to predict
electric vehicle charging duration time,” Transportation Letters, pp. 1–18, Aug. 2022, doi:
[Link]

[72] T. M azhar et al., “Electric Vehicle Charging System in the Smart Grid Using Different M achine Learning M ethods,”
Sustainability, vol. 15, no. 3, p. 2603, Feb. 2023, doi: [Link]
[73] M aduranga, M . W. P., & Abeysekera, R. (2020). M achine learning applications in IoT based agriculture and smart farming:
A review. Int. J. Eng. Appl. Sci. Technol, 4(12), 24-27.

[74] Amini, M ., & Rahmani, A. (2023). Agricultural databases evaluation with machine learning procedure. Australian Journal of
Engineering and Applied Science, 8(2023), 39-50.

[75] H. Pallathadka, M . M ustafa, D. T. Sanchez, G. Sekhar Sajja, S. Gour, and M . Naved, “IM PACT OF M ACHINE learning ON
M anagement, healthcare AND AGRICULTURE,” M aterials Today: Proceedings, Jul. 2021, doi:
[Link]

[76] M . Javaid, A. Haleem, I. H. Khan, and R. Suman, “Understanding the potential applications of Artificial Intelligence in
Agriculture Sector,” Advanced Agrochem, vol. 2, no. 1, pp. 15–30, Oct. 2022, doi: [Link]

[77] SSL, D. A., Praveenkumar, R., & Balaji, V. (2023). An Intelligent Crop Recommendation System using Deep Learning.
International Journal of Intelligent Systems and Applications in Engineering, 11(10s), 423-428.

[78] Y. Wang, J. Ge, S. M iao, T. Jiang, and X. Shen, “Application of hybrid artificial bee colony algorithm based on load balancing
in aerospace composite material manufacturing,” Expert Systems with Applications, vol. 215, pp. 119375–119375, Apr. 2023,
doi: [Link]

34
[79] I. Ben-Yelun, M . Díaz-Lago, L. Saucedo-M ora, del M ar, R. Callado, and F. J. M ontáns, “Self-learning locally-optimal
hypertuning using maximum entrop y, and comparison of machine learning approaches for estimating fatigue life in composite
materials of the aerospace industry,” Engineering Structures, vol. 283, pp. 115829–115829, M ay 2023, doi:
[Link]

[80] I. Shafi et al., “Deep Learning-Based Real Time Defect Detection for Optimization of Aircraft M anufacturing and Control
Performance,” Drones, vol. 7, no. 1, p. 31, Jan. 2023, doi: [Link]

[81] R. P. R. Nogueira, R. M elicio, D. Valério, and L. F. F. M . Santos, “Learning M ethods and Predictive M odeling to Identify
Failure by Human Factors in the Aviation Industry,” App lied Sciences, vol. 13, no. 6, p. 4069, Jan. 2023, doi:
[Link]

[82] U. Karaoğlu, O. M bah, and Q. Zeeshan, “Applications of M achine Learning in Aircraft M aintenance,” [Link],
M ar. 2023, doi: [Link]

[83] S. Li, C. Li, Y. Huang, and C. Zhai, “Real-Time Seismic Damage Simulation for Urban Building Portfolio Based on Basic
Building Information and M achine Learning,” Jan. 2023, doi: [Link]

[84] N. Rane, S. Choudhary, and J. Rane, “Artificial Intelligence (AI) and Internet of Things (IoT) - based sensors for monitoring
and controlling in architecture, engineering, and construction: applications, challenges, and opportunities,” Social Science
Research Network, Jan. 2023, doi: [Link]

[85] Joern Ploennigs and M . Berger, “AI art in architecture,” AI in Civil Engineering, vol. 2, no. 1, Aug. 2023, doi:
[Link]

[86] N. Rane, “Integrating Leading-Edge Artificial Intelligence (AI), Internet of Things (IoT), and Big Data Technologies for Smart
and Sustainable Architecture, Engineering and Construction (AEC) Industry: Challenges and Future Directions,” Social
Science Research Network, Sep. 24, 2023. [Link]

[87] A. Heena, N. Biradar, N. M . M aroof, S. Bhatia, R. Agarwal, and K. Prasad, “M achine learning based biomedical image
processing for echocardiographic images,” M ultimedia Tools and Applications, Jul. 2022, doi:
[Link]

[88] J. A. Sandhiya, S. Preetha, A. Sarumathi, D. Umadevi, G. Yogeshwari, and K. Shanmugam, “Identification and Alertness of
Cardiovascular Disease Using M ATLAB With IoT,” vol. 1, no. 1, pp. 31–36, Jan. 2023, doi:
[Link]

[89] A. Khanna, P. Selvaraj, D. Gupta, T. H. Sheikh, P. K. Pareek, and V. Shankar, “Internet of things and deep learning enabled
healthcare disease diagnosis using biomedical electrocardiogram signals,” Expert Systems, Nov. 2021, doi:
[Link]

[90] S. Aminizadeh et al., “The applications of machine learning techniques in medical data processing based on distributed
computing and the Internet of Things,” Computer M ethods and Programs in Biomedicine, vol. 241, p. 107745, N ov. 2023,
doi: [Link]
[91] M . Rana and M . Bhushan, “M achine learning and deep learning approach for medical image analysis: diagnosis to detection,”
M ultimedia Tools and Applications, Dec. 2022, doi: [Link]

[92] S. Ghani, N. Kumar, M . Gupta, and Sunil Saharan, “M achine learning approaches for real-time prediction of compressive
strength in self-compacting concrete,” Asian Journal of Civil Engineering, Dec. 2023, doi: [Link]
023-00942-5.

[93] F. Parisi, S. Ruggieri, Ruggiero Lovreglio, M aria Pia Fanti, and G. Uva, “On the use of mechanics-informed models to
structural engineering systems: Application of graph neural networks for structural analysis,” Structures, vol. 59, pp. 105712–
105712, Jan. 2024, doi: [Link]

[94] W. Teo, “Image-based structural performance prediction for reinforced concrete exterior beam-column joints,” [Link],
2023, Accessed: Jan. 02, 2024. [Online]. Available: [Link]

[95] A. M ikami, “Fundamental Study on Damage Detection of Civil Structures M odeled as M DOF System Based on M achine
Learning,” STRUCTURAL HEALTH MONITORING 2023, vol. 0, no. 0, 2023, Accessed: Jan. 02, 2024. [Online]. Available:
[Link]

[96] A. M elnikov, M . Kordzanganeh, A. Alodjants, and R.-K. Lee, “Quantum M achine Learning: from physics to software
engineering,” Advances in Physics: X, vol. 8, no. 1, p. 2165452, Dec. 2023, doi:
[Link]

35
[97] Georgiev, D. (2023, September). Exploring knowledge management from a software engineering perspective. In European
Conference on Knowledge M anagement (Vol. 24, No. 2, pp. 1571-1578).

[98] G. Uddin, M . A. A. Alamin, and A. Das, “An Empirical Study of Deep Learning Sentiment Detection Tools for Software
Engineering in Cross-Platform Settings,” [Link], Jan. 16, 2023. [Link] (accessed Nov. 27, 2023).

[99] U. A. Butt, R. Amin, H. Aldabbas, S. M ohan, B. Alouffi, and A. Ahmadian, “Cloud-based email phishing attack using machine
and deep learning algorithm,” Complex & Intelligent Systems, Jun. 2022, doi: [Link]

[100] Z. Zhang et al., “M achine learning aided high-throughput prediction of ionic liquid@M OF composites for membrane-
based CO2 capture,” Journal of Membrane Science, vol. 650, pp. 120399–120399, M ay 2022, doi:
[Link]

[101] Z. Shi, H. Liang, W. Yang, J. Liu, Z. Liu, and Z. Qiao, “M achine learning and in silico discovery of metal-organic
frameworks: M ethanol as a working fluid in adsorption-driven heat pumps and chillers,” Chemical Engineering Science,
vol. 214, p. 115430, M ar. 2020, doi: [Link]

[102] Q. Huang et al., “M achine learning and molecular fingerprint screening of high-performance 2D/3D M OF membranes
for Kr/Xe separation,” Chemical Engineering Science, vol. 280, pp. 119031–119031, Oct. 2023, doi:
[Link]

[103] W. Wei, S. Nan, H. Wang, S. Xu, H. Liu, and R. He, “Design and preparation of sulfonated polymer membranes for
Zn/M nO2 flow batteries with assistance of machine learning,” Journal of Membrane Science, vol. 672, pp. 121453–121453,
Apr. 2023, doi: [Link]

[104] X. Bai et al., “M achine-Learning-Assisted High-Throughput computational screening of M etal–Organic framework

membranes for hydrogen separation,” Chemical Engineering Journal, vol. 446, pp. 136783–136783, Oct. 2022, doi:
[Link]

[105] S. D. M ilić, Ž. Đurović, and M . D. Stojanović, “Data science and machine learning in the IIoT concepts of power plants,”
International Journal of Electrical Power & Energy Systems, vol. 145, p. 108711, Feb. 2023, doi:
[Link]

[106] Faten Khalid Karim, Doaa Sami Khafaga, M . M . Eid, S. K. Towfek, and Hend Khalid Alkahtani, “A Novel Bio-Inspired
Optimization Algorithm Design for Wind Power Engineering Applications Time-Series Forecasting,” Biomimetics, vol. 8,
no. 3, pp. 321–321, Jul. 2023, doi: [Link]

[107] A. Presekal, A. Stefanov, V. S. Rajkumar, and P. Palensky, “Attack Graph M odel for Cyber-Physical Power Systems using
Hybrid Deep Learning,” IEEE Transactions on Smart Grid, pp. 1–1, 2023, doi: [Link]

[108] R. A. de Oliveira and M . H. J. Bollen, “Deep learning for power quality,” Electric Power Systems Research, vol. 214, p.
108887, Jan. 2023, doi: [Link]

[109] N. E. Benti, M . D. Chaka, and A. G. Semie, “Forecasting Renewable Energy Generation with M achine Learning and Deep
Learning: Current Advances and Future Prospects,” Sustainability, vol. 15, no. 9, p. 7087, Jan. 2023, doi:
[Link]

[110] H. H. Darwish and A. Al-Quraan, “M achine Learning Classification and Prediction of Wind Estimation Using Artificial
Intelligence Techniques and Normal PDF,” Sustainability, vol. 15, no. 4, p. 3270, Feb. 2023, doi:
[Link]

[111] T.-C. Chen et al., “Application of machine learning in rapid analysis of solder joint geometry and type on thermomechanical
useful lifetime of electronic components,” M echanics of Advanced M aterials and Structures, vol. 30, no. 2, pp. 373–381, Dec.
2021, doi: [Link]

[112] Y. Yuan et al., “Insight into the Electronic Properties of Semiconductor Heterostructure Based on M achine Learning and
First-Principles,” ACS Applied M aterials & Interfaces, vol. 15, no. 9, pp. 12462–12472, Feb. 2023, doi:
[Link]

[113] B. M amandipoor, M . M ajd, S. Sheikhalishahi, C. M odena, and V. Osmani, “M onitoring and detecting faults in
wastewater treatment plants using deep learning,” Environmental Monitoring and Assessment, vol. 192, no. 2, Jan. 2020,
doi: [Link]

[114] J. Huang and D. D. Koroteev, “Artificial intelligence for planning of energy and waste management,” Sustainable
Energy Technologies and Assessments, vol. 47, p. 101426, Oct. 2021, doi: [Link]

36
 [115] Y. Xie et al., “Enhancing Real-Time Prediction of Effluent Water Quality of Wastewater Treatment Plant Based on
Improved Feedforward Neural Network Coupled with Optimization Algorithm,” Water, vol. 14, no. 7, p. 1053, M ar. 2022,
doi: [Link]

[116] A. Giwa, S. Daer, I. Ahmed, P. R. M arpu, and S. W. Hasan, “Experimental investigation and artificial neural networks
ANNs modeling of electrically -enhanced membrane bioreactor for wastewater treatment,” Journal of Water Process
Engineering, vol. 11, pp. 88–97, Jun. 2016, doi: [Link]

[117] M . M . Hamed, M . G. Khalafallah, and E. A. Hassanien, “Prediction of wastewater treatment plant performance using
artificial neural networks,” Environmental Modelling & Software, vol. 19, no. 10, pp. 919–928, Oct. 2004, doi:
[Link]

[118] Y. Wang et al., “M achine learning framework for intelligent aeration control in wastewater treatment plants: automatic
feature engineering based on variation sliding layer,” Water Research, vol. 246, pp. 120676–120676, Nov. 2023, doi:
[Link]

[119] C. A. Escobar and R. M orales-M enendez, “M achine learning techniques for quality control in high conformance
manufacturing environment,” Advances in Mechanical Engineering, vol. 10, no. 2, p. 168781401875551, Feb. 2018, doi:
[Link]

[120] S. Sankhye and G. Hu, “M achine Learning M ethods for Quality Prediction in Production,” Logistics, vol. 4, no. 4, p. 35,
Dec. 2020, doi: [Link]

[121] M . Amini, K. Sharifani, and A. Rahmani, “M achine Learning M odel Towards Evaluating Data Gathering M ethods in
M anufacturing and M echanical Engineering,” Social Science Research Network, Jan. 01, 2023.
[Link]

[122] J. M oore, J. Stammers, and J. Dominguez-Caballero, “The application of machine learning to sensor signals for
machine tool and process health assessment,” Proceedings of the Institution of Mechanical Engineers, Part B: Journal of
Engineering Manufacture, p. 095440542096089, Sep. 2020, doi: [Link]

[123] N. Saravanan, V. N. S. K. Siddabattuni, and K. I. Ramachandran, “Fault diagnosis of spur bevel gear box using
artificial neural network (ANN), and proximal support vector machine (PSVM ),” Applied Soft Computing, vol. 10, no. 1, pp.
344–360, Jan. 2010, doi: [Link]

[124] J. Rafiee, F. Arvani, A. Harifi, and M . H. Sadeghi, “Intelligent condition monitoring of a gearbox using artificial neural
network,” Mechanical Systems and Signal Processing, vol. 21, no. 4, pp. 1746–1754, M ay 2007, doi:
[Link]

[125] F. Jia, Y. Lei, L. Guo, J. Lin, and S. Xing, “A neural network constructed by deep learning technique and its
application to intelligent fault diagnosis of machines,” Neurocomputing, vol. 272, pp. 619–628, Jan. 2018, doi:
[Link]

[126] Z. Zhen and J. Gao, “CDR2IM G: A Bridge from Text to Image in Telecommunication Fraud Detection,” Computer systems
science and engineering, vol. 47, no. 1, pp. 955–973, Jan. 2023, doi: [Link]

[127] J. Pablo, Luis, and F. J. Rico-Novella, “A machine learning based Distributed Congestion Control Protocol for multi-hop
wireless networks,” vol. 231, pp. 109813–109813, Jul. 2023, doi: [Link]

[128] Heyn, H. M ., Knauss, E., M alleswaran, I., & Dinakaran, S. (2023, April). An investigation of challenges encountered when
specifying training data and runtime monitors for safety critical M L applications. In International Working Conference on
Requirements Engineering: Foundation for Software Quality (pp. 206-222). Cham: Springer Nature Switzerland.

[129] M . Ashwin, A. S. Alqahtani, A. M ubarakali, and B. Sivakumar, “Efficient resource management in 6G communication
networks using hybrid quantum deep learning model,” Computers and Electrical Engineering, vol. 106, p. 108565, M ar. 2023,
doi: [Link]

[130] B. Alhayani et al., “5G standards for the Industry 4.0 enabled communication systems using artificial intelligence: perspective
of smart healthcare system,” Applied Nanoscience, Jan. 2022, doi: [Link]

37

View publication stats