engine systems - eth zürich · important notes 2/3 02.10.2017 eth zurich -dmavt idsc 3 you do not...
TRANSCRIPT
Engine SystemsProblem 1: Modeling of an Intake Manifold
02.10.2017 1ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Important notes 1/3
02.10.2017 2ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Structure of the exercises – a reminder
Theory Exercises
Assigments 1-6
Matlab/Simulink Exercises
Intake Manifold
Exercise
Idle Speed Control
Exercise
Week 1-3
Week 4-13
The intake
manifold
prepares you
very well for
the much more
complex ISCS-
Exercise!
Important notes 2/3
02.10.2017 3ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
You do not have to hand in any material for the intake
manifold exercise.
Feel free to hand in a little report on October 13 which we
will proofread. This is to make sure you know what you are
doing before we start the idle speed control exercise.
We will discuss intermediate steps in each exercise class
and recommend you solve these tasks within the following
week.
Important notes 3/3
02.10.2017 4ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Theory sheet #2 is online
Complete sheet #1 by October 9th
Complete sheet #2 by October 23th
Intake Manifold Exercise Part 2
02.10.2017 5ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
So far: You should have created a working
Simulink model
Data: «Manifold Identification Data» from the webpage
Intake Manifold Exercise Part 2
So far: You should have created a working
Simulink model
Data: «Manifold Identification Data» from the webpage
02.10.2017 6ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Ugly layout Hard to debug
Nice layout
much easier to debug
Intake Manifold
Exercise Part 2
02.10.2017 7ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Helpful blocks: Goto
Simulink/Signal Routing
From
Simulink/Signal Routing
Fcn
Simulink/ User-Defined Functions
Intake Manifold Exercise Part 2
Still to come:
Parameter Identification (task of this week)
Linearization/Normalization (task of next week)
02.10.2017 8ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
General remarks There are many approaches to do a parameter identification
Different approaches often differ substantially in their efficiency
We use fminsearch, a Matlab-function that minimizes a user-defined cost function.
Very easy to use
Can be used to solve a large class of problems
Well suited for the given problem
Generally very inefficient
Parameter Identification using fminsearch
02.10.2017 9ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Task for this week
Parameter Identification using fminsearch
02.10.2017 10ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Study the files in ParIdDemo.zip on the webpage
Task for this week
Download «Manifold Identification Data» from the webpage
Normalize & Linearize the nonlinear model
02.10.2017 11ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Define nominal input
Equilibrium point: run your model with nominal inputs
until it reaches steady-state conditions and choose these
values as equilibrium around which you want to linearize
your model
Use the command linmod to perform the linearization
Theory behind normalizing/linearizing can be found in the
exercise instructions
Task for next week
Compare the system response of the linearized
and normalized model with the nonlinear model
Task for next week
A, B, C, D Matrices
02.10.2017 13ETH Zurich - DMAVT - IDSC www.idsc.ethz.ch
Just do it!