1© 2014 The MathWorks, Inc.

System Design and Simulation Using

Simulink

Prasanna Deshpande

Application Engineering,

Control Design and Automation

MathWorks India

2

Multi-domain System Design and Simulation

3

Grid

Today’s Example: Wind Turbine System

Yaw

Generator

Speed

Tower

Geartrain Generator

Pitch

Rotor

Speed

Blades

Hub

Wind

Nacelle

4

Traditional Development Process

Design and Implementation

MechanicalEmbedded

Software

Integration and Test

Requirements are

not integrated in

design process

Separate simulation

tools are difficult to

integrate

Errors are found too

late in the process

using expensive

prototypes

Control Electrical

Requirements and Specifications

5

Model-Based Design Process

Simulation Model

MechanicalEmbedded

Software

Requirements

and

Specifications

Save time by

developing in a single

simulation environment

Control Electrical

Produce better designs

by continuously

comparing design and

specification

Lower costs by using

HIL tests and fewer

hardware prototypes

PLC HIL

6

Challenges Involved in Modeling Complex

Multi-domain Systems

Modeling different components of the multi-domain

system

Integrating multi-domain components into single

simulation model

Designing and testing of control strategies at simulation

model and in real time

Managing models and data, managing versions of

multiple models, testing - analysis and reporting

7

System Level Model of Wind Turbine

8

Agenda

Modeling Dynamic Systems Using Simulink

Modeling Multi-domain Systems Using Physical

Modeling Approaches

Automatically Estimating Model Parameters Based On

Test Data

Re-using System Level Simulations for Performance

and Design Trade-off Studies

9

Data-Driven ModelingFirst Principles Modeling

Neural Networks

Physical NetworksSystem

Identification

Parameter Tuning

Programming

Block Diagram

Modeling Language

Symbolic Methods

Modeling Approaches

Different Approaches for Modeling Dynamic

Systems

Statistical Methods

(MATLAB, C)

(Simulink)

(Simscape language)

(Symbolic MathToolbox)

(Simscape and other

Physical Modeling

products)

(Neural NetworkToolbox)

(Model BasedCalibration Toolbox)

(Simulink Design Optimization)

(System Identification Toolbox)

10

Model and Simulate

Manage Design Data

Visualize and Analyse Results

Simulink

SLX

FileSLX

File

Model 1

Model 2

Model 3

SLX

File

SLDD

FileSLDD

FileSLDD

FileGlobal Data

Generate Code

Modeling Dynamic Systems With Simulink

11

Agenda

Modeling Dynamic Systems Using Simulink

Modeling Multi-domain Systems Using Physical

Modeling Approaches

Automatically Estimating Model Parameters Based On

Test Data

Re-using System Level Simulations for Performance

and Design Trade-off Studies

12

Supervisory Control

Mechanical

System

Hydraulic

System

Electrical

System

Electrical

System

Modeling Multi-domain System

Driveline

System

Control

AlgorithmControl

Algorithm

Control

Algorithm

Control+- Plant

13

Simulating Multidomain Physical Systems in

One Environment

Simscape enables modeling of multiple domains in one

environment:

– Connect domains with physical connections

Simulink integration enables modeling

of domains not yet covered in Simscape

V+

V-P TT

A B

14

Physical Systems in Simulink

Multibody mechanics (3-D) Mechanical systems (1-D)

Fluid power and controlMultidomain physical systems

Electrical power systems

Electromechanical and

electronic systems

Sim

Me

ch

an

ics

Sim

Dri

ve

lin

e

Sim

Hyd

rau

lics

Sim

Ele

ctr

on

ics

Sim

Po

werS

yste

ms

Simscape

MATLAB, Simulink

Sim

Me

ch

an

ics

Sim

Dri

ve

lin

e

Sim

Hyd

rau

lic

sS

imE

lec

tro

nic

s

Sim

Po

we

rSys

tem

s

SimscapeMechanical Hydraulic Electrical

Thermal

Liquid

Custom Domains via

Simscape Language

Pneumatic Magnetic

N S

15

Data-Driven ModelingFirst Principles Modeling

Neural Networks

Physical NetworksSystem

Identification

Parameter Tuning

Programming

Block Diagram

Modeling Language

Symbolic Methods

Modeling Approaches

Different Approaches for Modeling Dynamic

Systems

Statistical Methods

(MATLAB, C)

(Simulink)

(Simscape language)

(Symbolic MathToolbox)

(Simscape and other

Physical Modeling

products)

(Neural NetworkToolbox)

(Model BasedCalibration Toolbox)

(Simulink Design Optimization)

(System Identification Toolbox)

16

Agenda

Modeling Dynamic Systems Using Simulink

Modeling Multi-domain Systems Using Physical

Modeling Approaches

Automatically Estimating Model Parameters Based

On Test Data

Re-using System Level Simulations for Performance

and Design Trade-off Studies

17

AreaA AreaB AreaV

0.025 0.02 175

Estimating Parameters Using

Measured Data

Problem: Simulation results do not

match measured data because

parameters values are incorrect

Solution: Use Simulink Design

Optimization to automatically tune

model parameters

Model:

A B

P TT

A B

AreaA AreaB

AreaA AreaB AreaV

0.0176 0.0106 200

AreaV

18

Supervisory Control

Mechanical

System

Hydraulic

System

Electrical

System

Electrical

System

Modeling Multi-domain System

Driveline

System

Control

AlgorithmControl

Algorithm

Control

Algorithm

Control+- Plant

19

Control+-

Possibilities for Compensator Design

Linear Control Theory

– Linearize system using Simulink Control Design

– Perform linear control design with Control System Toolbox

– Retest controller in nonlinear system

A x + B u

Root Locus Bode Plot

Real Axis Frequency

Control+-

Specify System Response

– Specify response

characteristics

– Automatic tuning using

Simulink Design Optimization

20

Model the Supervisory

Control of the Wind Turbine

Problem: Create a supervisory

controller that sets the state of

various components based on Plant

conditions

Model:

Solution: Use Stateflow to

model the event-based

controller

wind > cut in speed &&

wind < cut out speed

turbine >

min speed

wind spd < min spd

|| wind spd > max spd

|| turbine spd < min spd

|| turbine spd > max spd

Turbine spd

< park spd

park brake = 0

pitch brake = 0

generator = 0

Startuppark brake = 0

pitch brake = 0

generator = 1

Generating

park brake = 0

pitch brake = 1

generator = 0

Brakepark brake = 1

pitch brake = 0

generator = 0

Park

21

System Level Model of Wind Turbine

23

Large Scale Modeling in Simulink

SubsystemsLibrariesModel

Referencing

o Componentization

24

Large Scale Modeling in Simulink

o Variant Subsystems

Model Variants,

Variant Subsystems

and

Variant Manager

System-Level Variant Management

Fewer models

Easy switching between variants

Less errors

Configurable software

with single code base

• Similar subsystems with slight

variations

• Design Choices

• Multiple Simulation workflow

options

25

Large Scale Modeling in Simulink

o Simulink Projects

• Managing project files and connecting with source control software

• Connection with Source Control Software

26

Agenda

Modeling Dynamic Systems Using Simulink

Modeling Multi-domain Systems Using Simscape and

Add-ons

Automatically Estimating Model Parameters Based On

Test Data

Re-using System Level Simulations for

Performance and Design Trade-off Studies

27

System Level Simulation Helps Detect System

Integration Issues In Simulation

Problem: Test for system

integration issues before

building hardware prototypes

Solution: Use the Simulink

environment to integrate the

separate systems in simulation

Model:

Mechanical

Hydraulic

Electrical

Controls

Park

Spin

Supervisory

Logic

LiftDrag

Wind

Aero-

dynamics

Actuator

(Ideal)Inputs

System

(Include)

Actuator

(Realistic)

System

(Ignore)

28

Summary: System Modeling and Simulation

29

Dynamic systems

Environment models

Analog behavior

Continuous-time

Summary: System Modeling and Simulation

30

Difference EquationsDSPImage/videoDigital controlSystem objects

Discrete-time

Continuous-time

Summary: System Modeling and Simulation

31

Differential Algebraic EquationsElectronicsMechanicsHydraulicsOther domains through Simscape

Physical models

Discrete-time

Continuous-time

Summary: System Modeling and Simulation

32

Control logicMode logic

Continuous-time

Discrete-time

Physical models

State machines

Summary: System Modeling and Simulation

33

LatencyResources

Continuous-time

Discrete-time

Physical models

State machines

Discrete-event

Summary: System Modeling and Simulation

34

Extended Kalman Filter

Continuous-time

Discrete-time

Physical models

State machines

Discrete-event

DesktopEmbedded

MATLAB

Summary: System Modeling and Simulation

35

Summary: System Modeling and Simulation

Models

Subsystems

Architecture

Continuous-time

Discrete-time

Physical models

State machines

Discrete-event

MATLAB

Inputs

Turbine Input 2

Wind

Inputs

Turbine Input 1

Wind

Wind

GNVa

GNVb

GNVc

10

Turbine Bank 2

Wind

GNVa

GNVb

GNVc

10

Turbine Bank 1

ElectricalSubsystemGNVa

GNVb

GNVc

GNV

Station 2

GNV

Grid

ElectricalSubsystemGNVa

GNVb

GNVc

GNV

Station 1

36

Continuous-time

Discrete-time

Physical models

State machines

Discrete-event

MATLAB

Summary: System Modeling and Simulation

• Integration of “home-grown” models,

using C, Fortran, or other language

• Co-simulation integration with

domain-specific modeling tools for

mechanical, hydraulic, electrical, etc.(over 100 of them integrate with Simulink….)

Architecture

37

Thank you for attending the session

Details Session / Demo Booth

Modeling Power Electronics, Electrical

Systems, Physical Modeling

2:30pm, Same Hall,

Vivek Raju

Control Design Techniques

Real Time Testing

4:30pm, Same Hall,

Chirag Patel

Customization of physical components

(Simscape Language)

Demo Booth on HVAC System

Dhirendra Singh

Code Generation for Embedded

Targets

Demo Booth on Programming

Embedded Targets

Antonin

Data Driven Modeling Demo Booth on Identify Plant

Dynamics and Detect Faults using

Online System Identification

Chirag Patel

SimMechanics, SimHydraulics Demo Booth of SimMechanics