ED5330
Control of Automotive
Systems
TOPIC
A Comprehensive Study of
Project Group Details PI and Model Predictive
ED25M007 – Biswayan Adhikari Control Strategies for SPS Modulated
ED25M011 – Kuldip Bag Dual ActiveBridge (DAB) Converter
ED25M014 – Naveenkumar V in Variable-Voltage
ED25M016 – Ramteke Kunal Electric Vehicle (EV) Battery Chargers
Pralhad
Faculty Details
Prof. ATRIYA BISWAS
OBJECTIVES
The objectives of the project are :
1. To perform the open loop simulation of Single-Phase Shift Modulated Dual-Active Bridge Converter
2. To perform the closed loop simulation of SPS DAB converter using PI controller
3. To perform the closed loop simulation of SPS DAB converter using MPC controller, designed using
SIMULINK blocks
4. To perform the closed loop simulation of SPS DAB converter using MPC controller, designed using
MATLAB coding
5. To perform a comparative analysis between them based on rise time for variable voltage demands
VARIABLE VOLTAGE BATTERY CHARGERS
A DC-DC converter which If a 200V battery is charged
Can be used to charge battery from a 400V battery charger, it
provides a range of output
packs of different voltage ratings will severely damage the
voltages
battery pack
Can reduce traffic in a charging
station as one single converter can
give different voltages
Topology used : Dual-Active
Bridge Converter
PRINCIPLE OF OPERATION
Equivalent Model of DAB converter Design Equations
Specifications
Power is transferred based on the phase shift
between the transformer primary and secondary
currents
Control Phase Shift Control 𝑽𝒐𝒖𝒕
Single Phase Shift Modulation
Same leg switches on either side of the transformer have a phase shift
Open Loop SIMULINK Schematic
Closed Loop SIMULINK Schematic using PI controller
The value of Output Voltage demand can be dynamically changed
during runtime by pausing the simulation (𝟐𝟎𝟎𝑽 ≤ 𝑽𝒅𝒆𝒎𝒂𝒏𝒅 ≤ 𝟒𝟎𝟎𝑽)
Closed Loop SIMULINK Schematic using MPC controller
The value of Output Voltage demand can be dynamically changed
during runtime by pausing the simulation (𝟐𝟎𝟎𝑽 ≤ 𝑽𝒅𝒆𝒎𝒂𝒏𝒅 ≤ 𝟒𝟎𝟎𝑽)
Process to Implement MPC using SIMULINK Blocks
Perform Open loop simulation and collect Phase
Use System Identification Application
Shift and 𝑽𝒐𝒖𝒕 data in workspace.
The app will read the data and provides the
transfer function.
Select Discrete time transfer function and save it
the workspace
Open a separate file and use LTI system block to
form the transfer function
Feed the transfer function and the reference Set limits of phase shift and output voltages and
output voltage to MPC controller design the MPC
Take the designed MPC and deploy it in the DAB
converter file.
Generalized Averaging Model
Modelling
GAM is a frequency-domain modeling analysis that
decomposes high-frequency transformer currents into
their fundamental Fourier components.
Hence, the first order approximation can be written as follows:
The dynamics of the system is transformed into the dynamics of
Fourier coefficients as follows:
The primary side is transformed to the secondary side such
that the turns ratio can be set to 1 for easier analysis.
Generalized Averaging Model
Linearization
Model Predictive Control using GAM
Discretization Feedback Control Law
Adding Integral state Cost Function
Model Predictive Control using GAM
LMI Problem Robust Stability
Introducing Qo, the slack variable and the decay variable To ensure robust stability, the stability condition must hold for
0 < 𝛼𝐾 ≤ 1 we may write every vertex of this uncertainty set. The performance can be
altered by changing 𝐷ഥ and μ.
Schur Complement form
Problem Formulation
Tuning such that
Simulation Results
OPEN LOOP RESULTS
Simulation Results
CLOSED LOOP USING PI CONTROLLER
Simulation Results
CLOSED LOOP USING MPC CONTROLLER IN SIMULINK
Simulation Results
CLOSED LOOP USING MPC CONTROLLER IN MATLAB
Comparative Analysis
T1 : (0V to
200V)
T2 : (200V to
300V)
T3 : (300V to
400V)
The settling time obtained using Model Predictive Controller is superior to PI and
LQR
MATLAB coding-based MPC provides better results when compared to
SIMULINK based design
MATLAB code used GAM while the SIMULINK MPC Designer uses Reduced
Order Model by default.
THANK YOU