Documente Academic
Documente Profesional
Documente Cultură
CATIA Training
Foils
Product Engineering
Optimizer
Copyright DASSAULT SYSTEMES
Version 5 Release 17
September 2006
EDU-CAT-EN-PEO-FI-V5R17
Instructor Notes:
Course Presentation
Targeted audience
CATIA V5 Users
Instructor Notes:
Table of Contents
1. Workbench Presentation p.6
Accessing the workbench p.7
PEO User Interface p.8
Checking User Settings p.9
Terminology p.13
Understanding the general process p.19
Instructor Notes:
Table of Contents
Running the optimization p.36
Exercises Presentation p.38
Introduction p.60
Creating constraints p.61
Specifying the number of solutions p.64
Setting the options p.65
Instructor Notes:
Planning
Day 1
MORNING Workbench Presentation p.6
Optimization Problem Formulation p.20
Optimization Problem Resolution p.32
Solution Analysis p.39
Best Practices for Optimization with PEO p.45
Instructor Notes:
Workbench Presentation
In this lesson, you will learn about the Product Engineering Optimizer by
Instructor Notes:
3-
Instructor Notes:
Optimization
Optimizations Design of
node under Experiments
Relations
node Constraint
Satisfaction
tool
Copyright DASSAULT SYSTEMES
Instructor Notes:
User Settings
Instructor Notes:
Access the General/Parameters and Measure node. In the Knowledge tab check the
corresponding option if you want …
(1) the parameter values to be displayed in the specification tree
(2) the formulas driving the parameters to be displayed in the specification tree
2
Copyright DASSAULT SYSTEMES
Instructor Notes:
Access the Infrastructure/Part Infrastructure node. In the Display tab, check the
corresponding option if you want :
(1) the Part Parameters to be displayed in the specification tree
(2) the Part Relations (Rules, Formulas, Design Tables, etc) to be displayed in the
specification tree
2
Copyright DASSAULT SYSTEMES
Instructor Notes:
Access the Infrastructure/Product Structure node. In the Tree Customization tab, switch the
Activated status to “Yes” if you want :
(1) the Product Parameters to be displayed in the specification tree
(2) the Product Relations (Rules, Formulas, Design Tables, etc) to be displayed in the
specification tree
2
Copyright DASSAULT SYSTEMES
Instructor Notes:
Terminology (1/6)
1-
2-
1- An Optimization Problem is a
computational problem that formulates the
goal to find the best of all possible values of a
function with respect to some constraints.
Example : Find the value R0 of the radius R of
a cone between 12mm and 40mm such that
the value V of the cone volume calculated by
function V=πR2h/3 has its maximal value
when R= R0, is a volume optimization problem
of type Maximization.
Instructor Notes:
Terminology (2/6)
Instructor Notes:
Terminology (3/6)
Instructor Notes:
Terminology (4/6)
Instructor Notes:
Terminology (5/6)
Instructor Notes:
Terminology (6/6)
Instructor Notes:
General Process
3
Choose an optimization
algorithm
4
Instructor Notes:
Instructor Notes:
2-
1-
Copyright DASSAULT SYSTEMES
Instructor Notes:
Instructor Notes:
1
Access the
Optimization
Problem editor
Instructor Notes:
1 2
Access the Click on Select button. In the panel that appears, select the
Optimization objective function. Click on OK
Problem editor
Copyright DASSAULT SYSTEMES
3 The information on the objective function appears in the Optimization problem editor.
Instructor Notes:
The information on free parameters indicating the current value of each parameter appears in
3
the Optimization problem editor.
Copyright DASSAULT SYSTEMES
Instructor Notes:
The information on the range and / or the step of the free parameter appears in the
3
Optimization problem editor.
Copyright DASSAULT SYSTEMES
Instructor Notes:
Instructor Notes:
The text of the constraint relation appears in the list of constraints in the
Instructor Notes:
Instructor Notes:
Editing a Constraint
1 Choose the constraint to edit in the list of the constraints in the optimization
problem editor. Click on Edit button.
Instructor Notes:
Instructor Notes:
Instructor Notes:
We advise you to use Conjugate Gradient algorithm when the objective function is continuous and
differentiable at all point, and Simulated Annealing algorithm in all other cases.
If you want to obtain a fine grain optimization, start with a Simulated Annealing algorithm then refine
the results with the Conjugate Gradient algorithm. This approach is slow but works for a large amount
Copyright DASSAULT SYSTEMES
of functions.
For both algorithms, the result can be refined by removing one or several parameters from the list of
free parameters and restarting the optimization.
Instructor Notes:
Gradient algorithm
Copyright DASSAULT SYSTEMES
This algorithm should be used first to perform a local search. Based on the calculation of a local slope of
the objective function, this algorithm will use a parabolic approximation and jump to its minimum, or
use an iterated exponential step descent in the direction of the minimum.
If the properties of the objective function are known (continuous, differentiable at all point), the
gradient can be used straight on. It is usually faster than the Simulated Annealing algorithm.
Instructor Notes:
1
Access the Optimization Problem editor
automatically.
Instructor Notes:
2
Click on Run Optimization button. If the option
Save optimization data is set, the panel asking you
to enter the name of the file where you want to save
the optimization results appears. Enter a file name
then click Open button.
Instructor Notes:
1
If you want to stop the optimization
process before one of the termination
criteria is reached, click on the Stop
button.
Instructor Notes:
Exercises Presentation
Now open the CATIA Training exercise book and practice on:
To learn how to :
Copyright DASSAULT SYSTEMES
Instructor Notes:
Solution Analysis
In this Step, you will see how to analyze the optimization results
Instructor Notes:
Optimization curves
1
Check that the option Save optimization data is set
in the Optimization Problem editor. When the
optimization problem calculation is finished, click
on Show curves button. A panel with the curves
presenting optimization results is displayed.
Copyright DASSAULT SYSTEMES
Instructor Notes:
In the left window you can see the graphical representation of values of
the free parameters and of the objective function for every evaluation
step. In the right window you can see what colour corresponds to what
parameter. When you click on a parameter in the right window, the name
Instructor Notes:
best value for the objective function value of the objective function and
corresponding values of the free parameters
evaluation
number
In this file you can find the following optimization results for every evaluation step : the values of
Copyright DASSAULT SYSTEMES
the free parameters and of the objective function, distances to the values satisfying the constraints.
Since they are presented in the Excel format, you can easily reuse these data for other calculations
or for a report.
Instructor Notes:
Instructor Notes:
Exercises Presentation
Now open the CATIA Training exercise book and practice on:
To learn how to :
Analyze the optimization results
Instructor Notes:
In this lesson you will learn some advices to perform a better optimization
process
Instructor Notes:
When you define an optimization problem, try to understand the character of the
objective function. If it is derivable (I.e. its derivative exists), it is better to apply
Conjugate Gradient algorithm to optimize the function. If it is not, use Simulated
Annealing algorithm.
Conjugate Gradient algorithm gives very good results when all the constraints of
the optimization problem are satisfied when starting the optimization
computation. We advise you to start with the Simulated Annealing algorithm to
satisfy all the constraints then to stop the optimization computation and switch
to Conjugate Gradient algorithm to obtain the best value of the objective
function.
Copyright DASSAULT SYSTEMES
Instructor Notes:
When you specify the termination criteria for an optimization problem, set for the
first computation :
Try to analyse the behavior of the objective function during the computation to
modify the termination criteria (if need) after this first test. If you want to speed
up the improvement of the objective function during the computation, decrease
the Consecutive updates without improvements parameter (but keep it greather
than 10).
Copyright DASSAULT SYSTEMES
Instructor Notes:
Defining constraints
When you define the constraints of an optimization problem, keep in the mind
that the more constraints you will formulate, the more precise your computation
will be.
Write down all the constraints on the paper. Set the priorities (if you can), keep at
the beginning of the constraint list the most important ones. Give them
comprehensive and significant names.
Add the constraints one by one to the set of constraints in CATIA PEO (starting
at the beginning of your list, of course). Perform the computation with the
Simulated Annealing algorithm for every added constraint. If all the constraints
are satisfied and the objective function is derivable, stop the optimization and
switch to Conjugate Gradient algorithm. Run the optimization to get the best
value for the objective function.
Copyright DASSAULT SYSTEMES
Instructor Notes:
In this lesson, you will learn about the Design of Experiments tool:
Introduction
Selecting input parameters
Selecting output parameters
Analysing the results
Predicting an output value
Copyright DASSAULT SYSTEMES
Instructor Notes:
Introduction
???
Controlled
input System Responses
factors
Copyright DASSAULT SYSTEMES
Instructor Notes:
1 2
Click the Design of Experiments
icon to open the Design of Click the Edit list button to select the input
experiment window. parameters that will be taken into account
when performing the analysis. In the
opening dialog box, select the input
parameters by using the arrow button and
click OK.
Copyright DASSAULT SYSTEMES
Instructor Notes:
Instructor Notes:
3
Click on Run DOE button to launch the analysis.
Enter the name of the output file and click Save.
Copyright DASSAULT SYSTEMES
Instructor Notes:
Results tab
1
Once the analysis is finished, select the Results tab.
The matrix displayed in the Table of experiments section is
the result of computations for each node: the number of
evaluations presented matches the number of updates
displayed in the Settings tab.
2
You can apply the values of the table to the
parameters. To do so, select one line in
the Table of Experiments and click the Apply
these values button.
Copyright DASSAULT SYSTEMES
Instructor Notes:
Generated curves
Instructor Notes:
Instructor Notes:
1
Select the Prediction tab. This tab presents a
mathematical model of the system and is used
to get a theoretical value of the output
parameter considering a specific configuration
of the input parameters.
2
Click the parameter to select it. Change its value
in the Selected parameter' s value field (hit the
Enter key or click anywhere to validate the input).
Repeat this operation for each parameter you
want to use.
Note: the parameter values must be contained
between their lower and upper ranges.
3
Click the Run Prediction button: the result displays
Copyright DASSAULT SYSTEMES
Instructor Notes:
Exercises Presentation
Now open the CATIA Training exercise book and practice on:
To learn how to :
Select the Free Parameters
Specify a Range and a Step for a Free Parameter
Access the Constraint Editor
Formulate a New Constraint
Specify algorithm
Run optimization
Copyright DASSAULT SYSTEMES
Instructor Notes:
In this lesson, you will learn about the Constraint Satisfaction tool:
Introduction
Creating constraints
Specifying the number of solutions
Setting the options
Copyright DASSAULT SYSTEMES
Instructor Notes:
Introduction
The user is able to drive the problem from any parameters (inputs or
computed ones) without changing the definition of the problem.
He can ask for one or several solutions. In the case of several solutions, he
can specify which must be the minimal distance between 2 solutions.
Note : This distance is calculated as an euclidean distance between 2 elements of the solution
Copyright DASSAULT SYSTEMES
set.
Instructor Notes:
Example:
Volume ==smartVolume(PartBody\Pad.1 );
PartBody\Pad.1\FirstLimit\Length >=1mm;
PartBody\Pad.1\FirstLimit\Length <=100mm;
Instructor Notes:
5
Set the value of the input parameters to the
desired values. To do so, click twice (slowly)
inside the Value cell and set the value.
Copyright DASSAULT SYSTEMES
Instructor Notes:
7
This process can be reversed: The input
parameters can be transferred to outputs
and vice-versa using the Switch input/output
arrow.
Copyright DASSAULT SYSTEMES
Instructor Notes:
Instructor Notes:
Termination criteria
2
Maximal computation time (sec.): Enables
you to indicate the computation time. If the
indicated time is equal to 0, the computation
will last until a solution is found.
Generate expanded error description:
Enables you to get detailed information
about errors if they occur.
Show 'Stop' dialog: If checked, displays a
"Stop" dialog box that will enable you to
interrupt the computation.
Instructor Notes:
Black box refers to a measure function. For instance, smartVolume is a measure that
calculates the volume of a body.
Instructor Notes:
Instructor Notes:
To Sum Up ...
In this Course you have seen the functionalities of the CATIA V5R14 Product
Engineering Optimizer :
How to solve it
How to Determine a cause-effect relationship between the response(s) and the input
factors of a system (Design of Experiments)
How to manage complex and hybrid engineering problems using the Constraint
Satisfaction feature
Copyright DASSAULT SYSTEMES
Instructor Notes: