Measurement and Control

Engineering
Chapter 5:
Process Controllers
 Outline

5.1 Introduction

5.2 On-Off Control

5.3 Proportional Control

5.4 Derivative Control

5.5 Integral Control

5.6 PID Control

5.7 Tuning

5.8 Digital Systems

5.9 Fuzzy Logic Control

5.10 Neural Networks

Chapter 5 : Process Controllers 2

 Objective

• Introduce process controllers and their role in modifying system output.

• Explain how controllers use the error signal (difference between required and actual values).
• Describe basic on/off control and its limitations.
• Present the three-term (PID) controller combining proportional (P), integral (I), and derivative (D) actions.
• Introduce newer control methods, including fuzzy logic control and neural networks.

Chapter 5 : Process Controllers 3

 5.1 INTRODUCTION

Direct and Reverse Actions

• Direct action: when the input to a system increases, the output also increases. Example: a central
heating system where raising the thermostat setting increases the room temperature.
• Reverse action: when the input to a system increases, the output decreases. Example: an air
conditioner where supplying more energy results in a lower room temperature.
• Choosing direct or reverse action depends on the desired system behavior and the nature of the
controlled variable.
Dead Time
• Dead time refers to the delay between a change in input and the system’s response.
• Sources of dead time include sensor response time (e.g., thermocouples respond in ~0.5 s, RTDs in
several seconds) and transfer delays in the system.
• Transfer delays occur in flow systems where changes at one point take time to affect another point,
such as water flowing through a pipe or material moving along a conveyor.

Chapter 5 : Process Controllers 4

 5.1 INTRODUCTION

• Example: in a room heating system, opening a window or changing the thermostat causes a lag before the
heater restores the temperature.
• Dead time affects control performance and must be considered in controller design to avoid overshoot or
instability.
Capacitance
• Capacitance describes the ability of a system to store or buffer the controlled quantity, affecting how quickly
it responds to changes.
• In tank level control, a larger water surface area slows the system’s response to inflow changes, while a
smaller surface allows faster adjustment.
• Electrical analogy: in an RC circuit, the larger the capacitance, the slower the voltage rises to steady state;
similarly, higher system capacitance slows response, while lower capacitance speeds correction.
• Capacitance must be accounted for in control design to achieve stable and timely system response.

Chapter 5 : Process Controllers 5

 5.2 ON-OFF CONTROL

On-Off Control
• The controller acts as a switch, activated by
the error signal, supplying only on or off
output; also called a two-step controller.
• Bimetallic thermostat: two metals with different
expansion rates are joined; temperature rise
curves the strip to open/close a switch. Heater
switches off if temperature is above set value
and on if below.
• Wax thermostat: wax expands when heated,
pushing a rod to open a valve; used in car
engine cooling and central heating radiator
valves.

Chapter 5 : Process Controllers 6

 5.2 ON-OFF CONTROL

On-Off Control
• On-off control is discontinuous, causing oscillations around the set point due to system time lags; e.g.,
room temperature fluctuates above and below the desired value.
• Dead band/neutral zone : prevents frequent switching from minor temperature changes; e.g., a
thermostat set at 20°C may switch on at 19.5°C and off at 20.5°C.
• A larger dead band causes larger temperature swings; a smaller dead band increases switching
frequency.

Chapter 5 : Process Controllers 7

 5.2 ON-OFF CONTROL

On-Off Control

Chapter 5 : Process Controllers 8

 5.2 ON-OFF CONTROL

On-Off Control
• On-off control works well with
systems of large capacitance
(slow response), is simple,
inexpensive, and can be
implemented mechanically
(bimetallic strips, relays) or
electronically (thyristors,
transistors).

Chapter 5 : Process Controllers 9

 5.2 ON-OFF CONTROL

On-Off Control
• Common applications: domestic central heating systems, car radiator cooling fans using wax
thermostats.

Chapter 5 : Process Controllers 10

 5.2 ON-OFF CONTROL

Relay
• Relays use a small current to energize a
solenoid, creating a magnetic field that moves
an armature to close contacts.
• Closing the contacts switches a much larger
current to a correction element (heater,
motor).
• When the solenoid current drops below a
threshold, the armature returns to off via the
spring action.
• Relays respond to error signals, turning the
system on or off when the error reaches set
limits.

Chapter 5 : Process Controllers 11

 5.3 PROPORTIONAL CONTROL

• Unlike on-off control, the controller output is proportional to the size of the error.
• Controller output = Kp × error,
where Kp is the gain constant.
• Correction element input varies with required correction, preventing overshoot and reducing
oscillations.
• Example: heater power decreases as temperature approaches the set point to maintain stable
temperature.
• Float-type water level control:
• lever mechanism translates error into proportional output;
• output movement = (x/y) × error.
• Amplifier example:
• output signal proportional to input error, used for temperature control of tank outflow..

Chapter 5 : Process Controllers 12

 5.3 PROPORTIONAL CONTROL

Proportional Band
• Controller output is often expressed as a
percentage of its full range (0–100%).
• Fully closed correction element
corresponds to 0% output, fully open to
100%; output varies proportionally with
error.

Chapter 5 : Process Controllers 13

 5.3 PROPORTIONAL CONTROL

Proportional Band
• Range: extreme
values the system
operates between
(e.g., 4–20 mA).
• Span: difference
between extreme
operating values.

Chapter 5 : Process Controllers 14

 5.3 PROPORTIONAL CONTROL

Proportional Band
• Absolute deviation:
error = measured value – set
value.
• Fractional deviation:
absolute deviation as a
fraction of the span.
Example: 3 °C error in a
0–30 °C span → 10%
deviation.

Chapter 5 : Process Controllers 15

 5.3 PROPORTIONAL CONTROL

Proportional Band
• Proportional Band
(PB): the percentage
deviation that
produces a 100%
change in controller
output.
• %PB = (% deviation
÷ % change in
output) × 100

Chapter 5 : Process Controllers 16

 5.3 PROPORTIONAL CONTROL

Proportional Band
• Example: 50% PB means a 50% error produces full (100%) controller output; 100% PB means a
100% error produces full output.
• PB is inversely related to proportional gain Kp: higher Kp → smaller PB; lower Kp → larger PB.

Chapter 5 : Process Controllers 17

 5.3 PROPORTIONAL CONTROL

Limitations of Proportional Control

• Proportional controllers apply a
correction proportional to the error:
correcting signal = Kp × error.
• As the error decreases, the correction
becomes smaller; if the error reaches
zero, no correction is applied.
• This results in a steady-state error: the
output cannot exactly reach the set
value using pure proportional control.

Chapter 5 : Process Controllers 18

 5.3 PROPORTIONAL CONTROL

Limitations of Proportional Control

• Increasing the gain Kp can reduce steady-state error but raises
the risk of system oscillations due to time lags.
• Oscillations occur because high gain produces large corrections
for small errors, potentially overshooting the set point.
• Zero steady-state error requires adding integral action to
proportional control, forming a PI controller.

Chapter 5 : Process Controllers 19

 5.4 DERIVATIVE CONTROL

• Controller output is proportional to the rate of change of the error:

derivative output = Kd × (rate of change of error).
• Output is often expressed as a percentage of the full output range, error as a percentage of full range.
• Kd is the derivative constant, often called derivative time (units of time).
• Provides rapid corrective action because output responds to how fast the error is changing, not just its
magnitude.
• For a steadily increasing error, the derivative controller produces a constant output to counteract the
change.
• Useful for anticipating and correcting errors quickly, improving system stability.

Chapter 5 : Process Controllers 20

 5.4 DERIVATIVE CONTROL

Chapter 5 : Process Controllers 21

 5.4 DERIVATIVE CONTROL

PD Control
• Combines proportional (P) and derivative (D) actions:
PD output = Kp × error + Kd × rate of change of error.
• Proportional part responds to the magnitude of error; derivative part responds to
the rate of change of error.
• Derivative action gives a rapid initial response to changing errors, while
proportional action provides gradual correction.
• Better than proportional control alone for handling fast process changes.
• Still cannot eliminate steady-state error; a constant error is required to maintain output for changes in input
or set point.
• The derivative constant Kd is often expressed relative to proportional gain as derivative time: Td = Kd / Kp.
• PD controller output can be written as: PD output = Kp × (error + Td × rate of change of error).

Chapter 5 : Process Controllers 22

 5.5 INTERGRAL CONTROL

PI Control
• PI control combines proportional (P) and
integral (I) modes to give a controller output:
PI controller output = Kp × error + Ki × ∫ error
dt, where Kp is the proportional gain and Ki
is the integral gain.
• The proportional part responds immediately
to the current error, while the integral part
accumulates past errors over time.
• Integral action eliminates steady-state error
because it continues to produce output even
when the current error is zero, maintaining
the set point.

Chapter 5 : Process Controllers 23

 5.5 INTERGRAL CONTROL

PI Control
• PI control combines proportional (P) and
integral (I) modes to give a controller output:
PI controller output = Kp × error + Ki × ∫ error
dt, where Kp is the proportional gain and Ki
is the integral gain.
• The proportional part responds immediately
to the current error, while the integral part
accumulates past errors over time.
• Integral action eliminates steady-state error
because it continues to produce output even
when the current error is zero, maintaining
the set point.

Chapter 5 : Process Controllers 24

 5.5 INTERGRAL CONTROL

PI Control
• The integral action time Ti = Kp / Ki;
the PI output can also be expressed
as: PI output = Kp × error + (Kp / Ti) ×
∫ error dt.
• PI control is suitable for processes
with large or slow-changing errors, but
rapid changes can lead to oscillations
due to the integration lag.
• The combined action provides both
quick response to error changes
(proportional) and correction of
accumulated errors (integral).

Chapter 5 : Process Controllers 25

 5.6 PID CONTROL

PID Control
A PID controller combines proportional (P), integral (I), and derivative (D) actions to eliminate steady-state
error and reduce oscillations.
• Proportional control provides immediate response to the current error.
• Integral control eliminates offset by accumulating past errors over time.
• Derivative control reduces the effect of time lags by responding to the rate of change of error.
• PID controllers can be arranged in different forms:
Parallel form: PID output = Kp × error + Ki × ∫ error dt + Kd × d(error)/dt.
Each mode acts independently; rarely used.

Chapter 5 : Process Controllers 26

 5.6 PID CONTROL

PID Control

Chapter 5 : Process Controllers 27

 5.6 PID CONTROL

PID Control

Chapter 5 : Process Controllers 28

 5.6 PID CONTROL

PID Control

Chapter 5 : Process Controllers 29

 5.6 PID CONTROL

PID Control
• Interactive form:
PID output = Kc × [error + (1/Ti) × ∫ error dt + Td × d(error)/dt]. Derivative acts on integrator output.
• Non-interactive (ideal) form:
PID output = Kc × error + (Kc/Ti) × ∫ error dt + Kc × Td × d(error)/dt. Integral and derivative act
directly on proportional output; widely used.
• If derivative control is not used, interactive and non-interactive forms are equivalent.
• Integrator windup occurs when actuators reach limits (e.g., fully open valve or maximum motor speed), but
the integral term continues to accumulate, causing overshoot and slow recovery.

Chapter 5 : Process Controllers 30

 5.6 PID CONTROL

PID Control
• Anti-windup circuits in
commercial PID controllers
prevent this by disabling the
integral action when actuator
saturation is detected

Chapter 5 : Process Controllers 31

 5.6 PID CONTROL

PID Process Controller

• Typical PID controllers include dual displays for process value and setpoint, status indicators, and
support for thermocouple or RTD sensors.
• They provide PID control with reduced overshoot, auto-tuning functions, and the ability to switch
between manual and automatic modes.
• Advanced features often include setpoint ramping during startup, sensor error compensation, break
detection, and lockout of tuning parameters.
• The OMEGA CN63100-R1 supports PID control for both heating and cooling, on-demand auto-tuning,
dual LED displays, and optional dual alarms.
• The TEMPATRON PID330 offers configurable On/Off or PID control outputs, up to three relay or SSR
outputs, auto-tuning, alarm output, and optional RS-485 communication..

Chapter 5 : Process Controllers 32

 5.6 PID CONTROL

PID Process Controller

Chapter 5 : Process Controllers 33

 5.7 PID TURNING

• Designing a controller involves choosing the appropriate control modes (P, PI, PD, or PID) and selecting
the tuning parameters Kp, Ki, and Kd.
• These settings determine how the system reacts to disturbances, how quickly it responds, how long it
takes to settle, and whether a steady-state error remains.
• A step-input response illustrates differences between modes: proportional control responds quickly but
leaves a steady-state error; PI control removes steady-state error but may introduce more oscillations.
• PID control also removes steady-state error and, with the derivative term, reduces oscillations and
improves stability.
• Tuning refers to selecting the best settings for fast response, minimal overshoot, and rapidly decaying
oscillations.
• Many commercial controllers offer auto-tuning, where small setpoint changes are applied and the optimal
parameters are calculated automatically.

Chapter 5 : Process Controllers 34

 5.7 PID TURNING

Chapter 5 : Process Controllers 35

 5.7 PID TURNING

Process Reaction Tuning Method

The process reaction method is performed with the control loop open, usually by disconnecting the controller
from the correction unit.
• A step input signal is applied to the correction unit, expressed as a percentage change P, and the controlled
variable’s output is recorded over time.
• The resulting plot of output versus time is the process reaction curve, from which a tangent is drawn at the
point of maximum slope.
• The lag L is defined as the time between the start of the test signal and the point where this tangent
intersects the time axis.
• The maximum slope R (percentage change of the controlled variable per minute) is also measured.
• Using L and M, Ziegler–Nichols provide formulas to calculate controller settings for P, PI, or PID modes.
• These settings are designed to produce a quarter-amplitude decay response, meaning each oscillation’s
amplitude is one-quarter that of the previous one.

Chapter 5 : Process Controllers 36

 5.7 PID TURNING

Process Reaction Tuning Method

Chapter 5 : Process Controllers 37

 5.7 PID TURNING

Process Reaction Tuning Method

Chapter 5 : Process Controllers 38

 5.7 PID TURNING

Process Reaction Tuning Method

Chapter 5 : Process Controllers 39

 5.7 PID TURNING

EXAMPLE
Determine the settings of Kp, Ki and Kd required for a three-mode controller which gave a process reaction
curve when the test signal was a 10% change in the control valve position.

A tangent is drawn at the point of maximum slope on

the process reaction curve, giving a lag L = 5 min
and slope M = 8/10 = 0.8 %/min.
• Using the Ziegler–Nichols process reaction
formulas for PID control:
• Kp = 1.2 × (P / (M·L)) = 1.2 × (10 / (0.8 × 5)) = 3
• Ti = 2L = 2 × 5 = 10 min
• Td = 0.5L = 0.5 × 5 = 2.5 min
• Integral gain Ki = Kp / Ti = 3 / 10 = 0.3 min⁻¹
• Derivative gain Kd = Kp × Td = 3 × 2.5 = 7.5 min
Chapter 5 : Process Controllers 40
 5.7 PID TURNING

Ultimate Cycle Tuning Method

• In the ultimate cycle method, the controller is kept in closed-loop while integral and derivative actions are
reduced to their minimum effective values.
• The proportional gain Kp is initially set very low and then gradually increased.
• Oscillations will begin to appear in the controlled variable as Kp increases.
• The critical proportional gain Kpc is the value at which sustained oscillations first occur.
• The period of these sustained oscillations is measured and recorded as Tc.
• These two quantities, Kpc and Tc, are then used in the Ziegler–Nichols ultimate cycle formulas to compute
P, PI, or PID controller settings.

Chapter 5 : Process Controllers 41

 5.7 PID TURNING

The process Ultimate Cycle Tuning Method as follow:

1. Set the controller to manual mode and bring the plant close to normal operating conditions.
2. Disable all control modes except proportional action.
3. Set Kp to a low value (wide proportional band).
4. Switch the controller to automatic mode and apply a small setpoint change (about 5–10%).
5. Observe the response of the controlled variable.
6. Increase Kp slightly (narrow the proportional band).
7. Apply another small setpoint change of 5–10%.
8. Observe the new response.
9. Repeat the increase–test–observe cycle until the system shows sustained oscillations that neither grow
nor decay. Record the proportional gain at this point (Kpu) and the oscillation period (Tu).
10. Use Kpu and Tu with the Ziegler–Nichols tuning tables. Table 5.2 gives the quarter-amplitude decay
settings; Table 5.3 provides PID settings for systems with overshoot or no overshoot.

Chapter 5 : Process Controllers 42

 5.7 PID TURNING

Process Reaction Tuning Method

EXAMPLE
For ultimate cycle tuning, oscillations begin when Kpc = 3.3, oscillation period Tu = 500 s.
• Using the Ziegler–Nichols quarter-amplitude decay formulas for PID tuning:
Kp = 0.6 × Kpc = 0.6 × 3.3 = 1.98

Chapter 5 : Process Controllers 43

 5.7 PID TURNING

Process Reaction Tuning Method

• Ti = 0.5 × Tu = 0.5 × 500 = 250 s, giving Ki = Kp / Ti = 1.98 / 250 = 0.0099 s⁻¹
• Td = 0.125 × Tu = 0.125 × 500 = 62.5 s, giving Kd = Kp × Td = 1.98 × 62.5 ≈ 123.75 s
• The suitable tuning constants for quarter-amplitude decay are Kp ≈ 1.98, Ki ≈ 0.0099 s⁻¹, Kd ≈ 123.75 s.

Chapter 5 : Process Controllers 44

 5.7 PID TURNING

Quarter Amplitude Decay:

• A modified ultimate cycle method
tunes the controller for quarter-
amplitude decay.
• The controller is set to
proportional-only, and a step input
is applied.
• The proportional gain is adjusted
by trial and error until the output
shows quarter-amplitude decay.
• Then Ti = T/1.5 and Td = T/6,
where T is the oscillation period.

Chapter 5 : Process Controllers 45

 5.7 PID TURNING

Lambda Tuning : aims for a non-oscillatory response to setpoint changes and load disturbances.
• Procedure: switch the controller to manual mode, wait for the process variable to reach steady state, then
apply a step change (bump) in the setpoint.
• Monitor the process variable until it settles and extract the following from the response graph:
• Dead time 𝐷
• Process time constant 𝜏(lambda value)
• Process gain

change in process variable

𝐾𝑝 =
size of step input
•PID controller settings based on lambda tuning:
• Proportional gain:
𝜏
𝐾𝑝 =
𝐾𝑝 ⋅𝜆

Chapter 5 : Process Controllers 46

 5.7 PID TURNING

Lambda Tuning :
• Integral time: 𝑇𝑖 = 𝜏 + 𝐷 or 𝑇𝑖 = 𝜆 + 𝑇𝑑 depending on variant
• Derivative time: 𝑇𝑑 = 0.

Chapter 5 : Process Controllers 47

 5.7 PID TURNING

Software Tools
• Software packages are available for tuning.
• They gather data from the system, develop process models and then suggest optimal tuning values.
• An impulse can be induced in the system and then the control system’s response used to determine
optimum loop values.
Adaptive Controllers
• Adaptive controllers change the controller parameters to adapt to changes in the process that occur with
time, e.g. a change in plant load.
• If the transfer function of the plant changes then there needs to be retuning of the controller if optimum
control conditions are to occur.
• An adaptive controller thus determines the values of K and K d p , K i needed to adapt to new process
conditions and makes the necessary changes.

Chapter 5 : Process Controllers 48

 5.7 PID TURNING

Adaptive Controllers
The self-tuning, often termed auto-tuning that exists with many commercially available controllers is a form of
adaptive control.
When the operator presses a button on the controller, the controller injects a small disturbance into the
system and determines the system’s response.
This is then compared with the desired response and the control parameters adjusted to bring the actual
response closer that required.

Chapter 5 : Process Controllers 49

 5.8 DIGITAL SYSTEMS

The Direct Digital Control (DDC)

• DDC uses digital computers or microcontrollers to compute control signals applied to actuators for
controlling the plant. Samples of the plant feedback are taken at regular intervals via an ADC.
• The computer compares the measured value with the digital setpoint and computes the control action
according to a control law (e.g., PID). The computed correction signal is sent through a DAC to actuators.
• DDC programs operate on both current and previous sampled values of the process variable.
Proportional (P) Control Program
• Initialise: set required value and proportional gain Kp
• Loop:
• Sample process variable and calculate error
• Compute output: Output = Kp × error
• Send output to DAC
• Wait until next sampling period

Chapter 5 : Process Controllers 50

 5.8 DIGITAL SYSTEMS

Proportional-Derivative (PD) Control Program

• Initialize: set required value, previous error, Kp, and derivative gain Kd
• Loop:
• Sample error
• Compute proportional output: P = Kp × error
• Compute derivative output: D = Kd × (error – previous error) / T
• Compute total output: Output = P + D
• Send output to DAC and update previous error
• Wait until next sampling period

Chapter 5 : Process Controllers 51

 5.8 DIGITAL SYSTEMS

Proportional-Derivative (PD) Control Program

Chapter 5 : Process Controllers 52

 5.8 DIGITAL SYSTEMS

Proportional-Integral (PI) Control Program

• Initialize: set required value, previous output, Kp, Ki
• Loop: • Sample error. Compute proportional output: P = Kp × error
• Compute integral output: I = previous output + Ki × error × T
• Compute total output: Output = P + I
• Send output to DAC and update previous output. Wait until next sampling period

Chapter 5 : Process Controllers 53

 5.9 FUZZY LOGIC CONTROL

• Traditional PID control is widely used, especially in temperature control systems, but can produce
overshoot when the setpoint changes abruptly or gradually.
• Fuzzy logic control offers a more flexible approach by using a set of control rules rather than relying solely
on a mathematical model.
• This method can continuously adjust control actions, improving performance when the system
experiences changing setpoints.
• Commercial examples include Omega’s Platinum Series controllers, which combine auto-tune PID with
fuzzy logic adaptive control.
• Fuzzy logic controllers are increasingly applied in domestic appliances (e.g., washing machines, vacuum
cleaners) to manage processes that are difficult to model mathematically..

Chapter 5 : Process Controllers 54

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic
• Conventional controllers respond to crisp inputs
(e.g., switch on heating at exactly 15°C). fuzzy
logic controllers respond to ranges of values,
handling concepts like “cold” or “warm” rather
than precise numbers.
• Decisions are based on degrees of truth, not
strictly YES or NO, allowing for partial
membership
• Fuzzy sets allow overlapping categories; an
input can belong to multiple sets simultaneously
with different membership values.

Chapter 5 : Process Controllers 55

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic
• Example – Heights:
classical sets classify
a person as either tall
or short; fuzzy sets
allow gradual
membership. A
person 1.5 m tall
might have 0.25
membership in “tall”
and 0.75 in “short.”

Chapter 5 : Process Controllers 56

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic
• Example – Temperature:
overlapping fuzzy sets like cold
(0–30°C), warm (10–30°C), hot
(20–40°C). A temperature of
18°C can belong partially to
“cold” (0.2) and “warm” (0.8)
sets.
• Fuzzy logic better represents
real-world situations where
categories are not sharply
defined and decisions are
intuitive.

Chapter 5 : Process Controllers 57

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Control Systems

• Overview: A fuzzy logic controller handles a range of input values rather than a single crisp input, using
three main stages: fuzzification, rule-based decision making, and defuzzification.

Chapter 5 : Process Controllers 58

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Control Systems

1. Fuzzification:
• Convert crisp sensor inputs into fuzzy values using membership functions.
• Inputs can belong to multiple fuzzy sets with varying degrees of membership (0–1).
• Example: Temperature 18°C may belong to WARM (0.9) and COLD (0.2).
2. Rule-Based Decision Making:
• Use IF–THEN rules to determine control actions based on input memberships.
• Example: IF temperature is HIGH THEN action 1; IF temperature is MEDIUM THEN action 2.
• Inputs may trigger multiple rules with different strengths.

Chapter 5 : Process Controllers 59

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Control Systems

Chapter 5 : Process Controllers 60

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Control Systems

Chapter 5 : Process Controllers 61

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Control Systems

3. Defuzzification:
Convert fuzzy outputs into a single crisp value.
Mamdani method: truncate output membership functions at their membership values, aggregate
overlapping areas, then take the center of gravity of the consolidated area as the crisp output.
Alternative method: overlapping areas counted separately, center of gravity of summed areas gives
the output.
The center of gravity is the point where the area to the left equals the area to the right.
• Additional Notes:
For symmetric triangular membership functions, the center of gravity lies under the apex.
For trapezoidal cuts, area = ½ × height × (sum of top and bottom lengths).

Chapter 5 : Process Controllers 62

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Controller (FLC) Applications

• Temperature Control:
Components: temperature sensor (e.g., LM35), microcontroller, actuator (heater).
Analog sensor signal → ADC → microcontroller → fuzzy logic program → heater on/off.
Example fuzzy rules for heater on-time:
IF error negative → very long on-time
IF error small negative → long on-time
IF error zero → medium on-time
IF error small positive → short on-time
IF error large positive → very short on-time

Chapter 5 : Process Controllers 63

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Controller (FLC) Applications

• Temperature Control:

Chapter 5 : Process Controllers 64

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Controller (FLC) Applications

• Fuzzy Feedback Control:
Crisp inputs (error, rate of change) fuzzified into triangular membership functions.
Fuzzy rules map inputs to fuzzy outputs.
Defuzzification (e.g., Mamdani method) produces crisp control signal.
• Automobile Cruise Control Example (Fuzzy PD Control):
Inputs: velocity error, acceleration
Rules:
Too fast → minimum force applied
Zero error → normal force applied
Too slow → maximum force applied
Derivative term approximated using backward difference of error

Chapter 5 : Process Controllers 65

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Controller (FLC) Applications

Chapter 5 : Process Controllers 66

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Controller (FLC) Applications

• Domestic Appliances (Washing Machines):
Inputs: load weight, fabric type, dirtiness
Outputs: wash time, spin speed, water amount, temperature
Example rule: IF dirt low AND fabric less sensitive AND weight low
→ short wash, low spin, low temp
• Fuzzy Implementation of Classical Controllers (PD, PI, PID):
Linguistic terms (positive, negative) map to error and rate of change.
Rules define output based on combination of error and its derivative.
Example PD rules:
IF error positive AND error rate positive → output zero
IF error positive AND error rate negative → output positive
IF error negative → output zero

Chapter 5 : Process Controllers 67

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Controller (FLC) Applications

Chapter 5 : Process Controllers 68

 5.9 FUZZY LOGIC CONTROL

Fuzzy Logic Tuning of PID Controllers

• Fuzzy logic can be used to automatically tune PID parameters when system conditions change.
• A self-tuning PID adjusts Kp, Ki, and Kd based on fuzzy rules derived from system behavior.
• Typical rules include: large steady-state error → increase Kp;
oscillatory response → increase Kd; sluggish response → increase Kp; excessive overshoot →
decrease Kp.
• This adaptive tuning improves both transient and steady-state performance compared to a fixed-
parameter PID controller.

Chapter 5 : Process Controllers 69

 5.10 NEURAL NETWORK

• Conventional control methods rely on mathematical models, whereas fuzzy control relies on rule sets.
• Neural networks provide another modern control method, inspired by the structure and function of the
human brain.

Chapter 5 : Process Controllers 70

 5.10 NEURAL NETWORK
• An artificial neural network (ANN) consists of interconnected “neurons” that process data through weighted links.
• A typical ANN has input neurons, one hidden layer, and output neurons. Information flows from input → hidden
→ output layers.
• Each connection carries a weight; higher weights give stronger influence to the transmitting neuron.

Chapter 5 : Process Controllers 71

 5.10 NEURAL NETWORK

• Hidden neurons compute a weighted sum of their inputs and pass the result through a threshold (activation)
function.
• The challenge is determining suitable weights; this is achieved through supervised learning with known
input–output pairs.
• During training, the network compares its actual output with desired output and adjusts the weights to
reduce the error.
• Through repeated exposure to examples, the network “learns” correct pathways and can later generalize to
new inputs.

Chapter 5 : Process Controllers 72

 5.10 NEURAL NETWORK

Chapter 5 : Process Controllers 73

 5.10 NEURAL NETWORK

Chapter 5 : Process Controllers 74

 5.10 NEURAL NETWORK

• Neural networks can act directly as controllers in a control system, known as direct neural control.
• Unlike conventional on/off control with fixed switching conditions, a neural controller can learn to adapt its
switching behavior when system conditions change.
• Example: a room heating controller using a neural network may take inputs such as room temperature,
temperature change over time, outdoor temperature, and outdoor temperature change, process them through
hidden neurons, and produce an on/off output signal.
The neural controller must be trained to generate correct control actions over a wide range of input
conditions.
• Neural networks can also be used indirectly for control by modelling the plant being controlled.
• In indirect neural control, the neural network predicts the plant’s behaviour; a conventional controller then
uses these predictions to determine its parameters.
• This approach is especially useful when no mathematical model of the system exists and only performance
data is available...

Chapter 5 : Process Controllers 75

 5.10 NEURAL NETWORK

Chapter 5 : Process Controllers 76

•.

Summary: Chapter 5 Process Controllers

• Process Dynamics: Systems exhibit direct/reverse action, dead time, and capacitance, all affecting
controller performance and stability.

• Basic Control Modes: On–off control is simple but causes oscillations; proportional control reduces
oscillations but leaves steady-state error.

• P, I, D Actions: P gives immediate correction, I removes steady-state error, and D anticipates changes by
reacting to error rate.

• PID Controllers: Combine P, I, and D to balance speed, stability, and accuracy; include issues like integral
windup and use anti-windup circuits.

• Tuning Methods: Ziegler–Nichols (process reaction & ultimate cycle) and auto-tuning software determine
optimal Kp, Ki, Kd for fast, stable response.

• Modern Control: Digital control (DDC), fuzzy logic controllers, and neural networks improve adaptability
and handle nonlinear or poorly modeled systems.
Chapter 5: Process Controllers 77