How to keep actions values at minimum before disturbance and let the agent choose different action values only after the disturbance?

Hello,

I am training reinforcement learning DDPG agent for for a control application. In my training, the agent tries to increase the agent actions from the start of each episode, I want the agent to keep minimum values of actions before the disturbance occurs and then change the action values after the disturbance. How to implement this idea ?

Thank you.

Answer 1

[Emmanouil Tzorakoleftherakis]

Please take a look here. As of R2022a you can place the RL policy block inside a triggered subsystem and only enable the subsystem for training when required.

Hope this helps