Resetting S-Function environment for different reinforement learning model

I am trying to design a reinforcement learning controller for my model which is defined by a Simulink S-Function block. My model runs for a small time interval [0 tf] for one computation of the reward function. How can I reset the S-Function block so it runs again from the same initial state as often as possible for training the RL controller? Thanks for your help.

Answer 1

[Anurag]

Hello Mlaudu,

You can use the “InitializeConditions” method. The "InitializeConditions" method is used to reset values that need reinitialization each time an enabled subsystem, containing the S-function, is executed.

You can refer to following MATLAB Documentation links to get better understanding

• https://www.mathworks.com/help/releases/R2020b/simulink/sfg/writing-level-2-matlab-s-functions.html#brgtux6:~:text=Initialize%20the%20values,function%20is%20reenabled.
• https://www.mathworks.com/help/releases/R2020b/simulink/sfg/initializeconditions.html

[Musa]

Thank you very much. We've since worked out a work-around for the problem.