My controlled object is a 7×7 state space expression with three inputs and three outputs. I want to use three parallel agents in reinforcement learning to control these three

1 次查看（过去 30 天）

嘻嘻 2023-11-2

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2041891-my-controlled-object-is-a-7x7-state-space-expression-with-three-inputs-and-three-outputs-i-want-to

评论：嘻嘻 2023-11-4

My controlled object is a 7×7 state space expression with three inputs and three outputs, just like the following picture. I want to use three parallel agents in reinforcement learning to control these three channels respectively, but when training the agent, it always breaks down. The error code is as follows. Is there any effective way to solve this problem? I would be very grateful for your answer.

The error code： ''The derivative of the state in Simulink is not finite, the simulation will stop, and there may be a singularity in the solution.''

I will be very grateful for your answer.

2 个评论
显示无隐藏无

Emmanouil Tzorakoleftherakis 2023-11-3

I would first make sure the simulation works without adding RL agent block. There seems to be a modeling error somewhere. Not sure what you mean by 7x7 state space either. Do you have 7 states?

嘻嘻 2023-11-4

@Emmanouil Tzorakoleftherakis

First, my model is capable of running without adding RL. Second, my model does have seven states. Perhaps it is precisely because my model is a complex nonlinear model that training always stops because the state solution has a singularity. Thank you very much for your answer. I look forward to your suggestion again.

请先登录，再进行评论。

请先登录，再回答此问题。