Sei sulla pagina 1di 19

for Western Digital

Connecting Theory with Practice

Control Systems Design,


Tuning and Optimization with
MATLAB and Simulink

James Ang
Application Engineer
TechSource Systems Malaysia
ang.cs@techsource.com.my
© 2010 TechSource Systems

Agenda
• Brief Introduction to Simulink
• Introduction to Controls System Design
• Various approaches in Plant Modeling
– First Principles Method + demo
– Data Driven Modeling (Blackbox method) + demo
– Parameter Estimation (Graybox Method) + demo
• Compensator Design
– Design Challenges
– Controls Design, Tuning and Optimization demo
• Summary
• Question and Answers
© 2010 TechSource Systems

Simulink

Lego : Using standard bricks to construct, assemble or build models.


The pieces used can be taken apart again, used to create other objects.
© 2010 TechSource Systems

Simulink®
 Block-diagram environment
 Model, simulate, and analyze time-
varying systems
– Control systems
– Signal processing systems
– Communications systems
– Other dynamic systems
 Accurately design, implement, and test
 Platform for Model-Based Design
© 2010 TechSource Systems

Simulink® Blocksets
Simulink blocksets offers a customized set of block libraries that
let you design, implement, verify and validate various time-
varying systems.
Simulink
Signal Video & Image Control
Processing Processing Design And many
more …

Communications SimMechanics RF

Quick
Demo
© 2010 TechSource Systems

What is Control Systems design?

Controller/
+
Feedback
Controller/ Plant
Input Compensator
Feedback Plant Output
-
Compensator

Feedback loop

Servo system = Closed looped control system


© 2010 TechSource Systems

Speed Control of a DC motor

Controller Plant
+
- Model Model
Input Output

How to start?
1. (a) Getting the plant model
© 2010 TechSource Systems

Two Approaches to Modeling Dynamic Systems

First-Principles Modeling

Use an understanding of the


system’s physics to derive a
mathematical representation
© 2010 TechSource Systems

Two Approaches to Modeling Dynamic Systems

First-Principles Modeling Data-Driven Modeling

Use an understanding of the Use system test data to derive a


system’s physics to derive a mathematical representation
mathematical representation

 s1 
s3 3s2 2
H(s)  
 s 3 
2

 2 
 s s1 
© 2010 TechSource Systems

Both Approaches Have Advantages & Disadvantages

First-Principles Modeling Data-Driven Modeling

Advantages: Advantages:
 Model provides insight into the  Fast to develop a model
system’s underlying behavior  Usually highly accurate
Disadvantages: Disadvantages:
 Model does not accurately match  Requires a physical system to
model output acquire test data
 System contains unknown,  Model lacks description of
nonlinear parameters like friction physics of the system
 Time consuming to develop  Need many data sets to cover
 Highly nonlinear systems difficult range of system operation
to express in first principles
© 2010 TechSource Systems

Tools That Span Both Modeling Approaches

Complete
First-Principles Modeling Environment
Modeling Data-Driven Modeling

First-Principles Data-Driven

Simulink Simulink System


Design Identification
SimMechanics Optimization Toolbox
SimDriveline Neural Network
SimPowerSystems Toolbox

Aerospace Fuzzy Logic


Blockset Toolbox
Demo
© 2010 TechSource Systems

Controller - Speed Control of a DC motor

Controller Plant
+
- Model Model
Input
Output

How to start?
1. (a) Getting the plant model
2. (b) Design the Controller model  Controller Design and
Simulation
© 2010 TechSource Systems

Control Challenges
• Common Method : Trial and
Error tuning of feedback controllers
• Tuning challenges
– Tuning is not systematic and is
time consuming
– Tuning on a real system can be
costly
– Impossible to tell if you’re at
optimal solution
© 2010 TechSource Systems

A controller changes the natural plant response into a desired


response
A controller helps make
the natural plant response
natural plant response go from this…

desired plant response

… to become
like this
Demo
© 2010 TechSource Systems

Summary and Wrap up


Control systems design

 Building the model


– First-Principles Modeling (Simulink), Data-Driven Modeling (System
identification Toolbox), Using Root locus and bode diagram to
design the controller (SISO tool),

 Optimize your model


– specify the desired output (Simulink DesignOptimization)
© 2010 TechSource Systems

Key Takeaways
Control Systems toolbox

Design
Analysis

Modeling
© 2010 TechSource Systems

Key Takeaways
System Identification
Toolbox

 Model estimation using


time & frequency domain
data
 Data preprocessing tools
included
 Blocks for directly adding
data-driven models to
Simulink models
© 2010 TechSource Systems

Key Takeaways
Simulink Design Optimization • Tuned Parameters
• Desired Response
 Estimate and Optimize model
parameters
 Interactive data preprocessing
 Model validation using residual
plots

Rise time, settling time,


% overshoot
© 2010 TechSource Systems

Do you have any questions?

Thank you for your attendance.

Tel: 603 – 8076 1953 Fax: 603 – 8076 1954


Email: info@techsource.com.my Web: www.techsource.com.my
Tech-Support: techsupport@techsource.com.my

Potrebbero piacerti anche