Cache file errors during parallel computing using simulink model

I am trying to use PSO (particle swarm optimization) where the objectivefunction runs a simulink model and evalutes some parameters. However, the simulink file sometimes does not run and I get errors in the simulink simulation that look like the following:

1. 'Cached model compile data may be stale. Delete any cached information such as 'slprj' directory and Simulink cache files. Suggested Actions: • Clear cached model compile data - Open'
2. 'Simulink code generation folder in the current folder was created for a different release. The 'slprj' subfolder is not compatible with the current release. Manually remove the 'slprj' folder and generated code files that the folder contains.'

I tried using rmdir to remove this folder after each iteration of PSO, however, even that is not working as my simulation model is still throwing these errors. Kindly provide me with the resolution to this issue. Thank you.

[Zhihua]

Hi,When I run Parallel Calculation by 'parfor',also occur this problem same with you.

[Ayush]

If you are using older versions of MATLAB then try using temp directory to build the model to avoid error in current MATLAB.

[Suyash Agrawal]

Thank you for your response. I am using Matlab 2023a, but I understand the reasoning behind your response.

Answer 1

[VINAYAK LUHA]

Hi Suyash,

As per my understanding of the issue, the “rmdir” function is unable to delete the cached files in Simulink project(slprj) directory successfully, hence the persistence of the issue.

Follow the below steps to clear Simulink cache files-

• Remove “slprj” directory and its subfolders irrespective of their writing permission using the following command-

     rmdir('myfiles', 's');

• Update generated cache files using the “rehash toolboxcache” command and restore the default MATLAB search path using “restoredefaultpath” command.

Find attached the MathWorks documentation for "rehash toolboxcache" command for your reference.

I hope it helps!

[Suyash Agrawal]

Thank you for your response. This definitely helped. I used mrdir for each instance of mu iteration and I haven't faced this error again.