Training an agent as a motor's controller using reinforcement learning in Matlab, but it doesn't train at all?

Question

I'm trying to use reinforcement learning to train an agent as a controller of a motor.

After reading this example , I only made some minor modifications:

1. Replace the watertank model with my motor's transfer function 69.4/(s+69.4), like this: .
2. Set the sample time Ts and simulation time Tf to 1/8000 and 10, respectively.

However, the result is:

It didn't do training at all :( The reward is always zero.

Could anybody give me advices on how to deal with this problem?

Very thankful in advance.

Answer 1

You maybe should check the output values that your transfer-function generates.

In the watertank-example you have the "stop simulation"-block, which stops the simulation if your "height"-value exeeds 20 or is lower than 0.

This would explain that your step-per-episode count is 0, because no step was taken due to incompatible values.