How you can get started with machine learning for CAE in 1 minute
Have you ever wanted to try design exploration & machine learning but never had the time? Then this post is for you.
Below is a video showing how to set up a design exploration study of a radioss model in 1 minute 8 seconds. Hopefully this will inspire you to give it a go!
The steps are:
- Create a Design Exploration
- Define Design Variables
- Define Responses
- Run the exploration.
Once set up and the runs have finished, it's time to reap the benefits. You can now:
- Look at linear effects of each design variable on the responses.
Which design variable influences your responses the most? - Run what-if studies
What if I change the thickness of the outter panel to 2mm? What effect will that have on my responses? - Run what if studies using the field predictor
What effect will changing design variables have on the full analysis result of the model? - Clustering
How many different types of post buckling shapes are there in my design exploration?
Have a look at all these four steps in the video below.
Hopefully you agree with me that it was one minute well spent. What problems are you looking to solve using design exploration? Please comment below.
Comments
-
Hello Mr. Ollar, can you provide the FE model please? Thank you in advance!
0 -
Alexander Melikov_22244 said:
Hello Mr. Ollar, can you provide the FE model please? Thank you in advance!
Hi Alexander,
No problem at all. I have attached it to this message. Hope you find it useful.
Jonathan0 -
Jonathan Ollar_21112 said:Alexander Melikov_22244 said:
Hello Mr. Ollar, can you provide the FE model please? Thank you in advance!
Hi Alexander,
No problem at all. I have attached it to this message. Hope you find it useful.
Jonathanthank you very much!
0 -
Hi Jonathan,
Which is the difference between this ML approach and the DOE classic one? I mean in a conceptually way, excluding the useful and easy GUI implemented.
Thanks in advance
****
0 -
Joaquín González_20468 said:
Hi Jonathan,
Which is the difference between this ML approach and the DOE classic one? I mean in a conceptually way, excluding the useful and easy GUI implemented.
Thanks in advance
****
Hi Joaquin,
Sorry for the late reply.
Do you mean what the difference is between DesignExplorer and HyperStudy? If so, you are right, they are very similar. In fact DesignExplorer uses HyperStudy technology, but is available within the HyperWorks UI for a more interactive user experience. This lets us do things like vizualise main effects and predictions directly on the model.
Thanks for the question,
Jonathan0