Matlab gets stuck when the swarm intelligence algorithm optimizes the control parameters...

Dear all,

I am working on a swarm intelligence algorithm to optimize the parameters of ADRC control, which uses m-code and simulink model. This swarm intelligence algorithm is based on iteration. I write an iterative algorithm in matlab with m-code, and then use the sim function to run ADRC mdl file and return the output value. The controlled object is a relatively complex and huge industrial system.

However, during the running process, matlab often get stuck, that is, it does not report an error or runs down, and it will keep this state, and the computer will also report that the temperature is too high.

I want to ask if it is because the controlled system is too huge or there are too many iterations, or the running time of each iteration is too long to cause matlab to get stuck? Is there any workaround?

Thank you for your help.

[Sam Chak]

Hi @Joe Jones

Without testing the code and the ADRC.mdl, we can only make wild guesses.

Does the optimization of the ADRC thing works on a simple system such a Double Integrator?

Does any non-ADRC thing such as state-feedback works on the huge industrial system?

Have you tested the optimization using any non-swarm intelligence algorithm?

Doing so may help you to identify and pinpoint one of the 3 suspects {PSO, ADRC, Huge System} that might be causing the issue at the macro level.

[Joe Jones]

Hi@Sam Chak

I have received your suggestions.

I will test the matlab file according to your suggestions and reply to you soon. I hope you can help me again then. Thank you for your valuable advice!

Best Regards!

[Joe Jones]

Hi，I'm back! @Sam Chak

I have already tested my code and mdl file according to your suggestions.

I tested the optimization of the ADRC on a simple system such a Double Integrator, however, the result was same as before.

Let me explain my model to you. My research is trajectory tracking. I used ADRC as a controller to generate control force to the controlled object, and then the controlled object will feedback the actual position to the controller to form a closed loop. Because the controller has many parameters, I use the SSA algorithm to optimize the parameters of the controller. As shown in the picture. In this model, I set the trajectory as y=x+5, and design controllers for x, y and z directions respectively. The parameters to be adjusted are beta1, beta2 and beta3 in ESO. The vessel modelof the controlled system should be fine.

I have attached my model and code in " test.zip ". Open all file and run the m-file " main ". I hope you can help me check it.

Thank you for your time and effort on my question.

Best Regards!