Optimize Simulink Model with valid parameter combination and invalid solutions/errors | Skip solutions with errors in optimization

I am trying to run a global opimization on my simulink model and within the range of my parameters are combinations, that creates invalid solutions/ erros.

Is there a Way to skip these combinations?

I guess I could create constraints that could prevent some of them, but I would also hinder valid solutions...

Answer 1

[Esha Chakraborty]

Hi Tobias,

I understand that you want to optimize a Simulink Model while skipping the invalid solution sets.

You can define a parameter-only constraint that identifies such design variable sets which result in model simulation error. Then the solver can use the constraint to skip only such sets and prevent the model from being evaluated in an invalid solution space.

You may check out a similar example of Thermostat Model here.

[Tobias Held]

Catching the errors/ invalid solutions and giving NaN back works. I am wondering how the optimizer interpret NaN solutions and if making it inf or any other very high number could be an advantage, but I have to do some optimization read up to do any way. Thank you!