I want to train anfis using a wide range of input/output data. That data comes from conventional controller like PID. PID controller can take one input at a time. I have tried different method to give a range of inputs to PID controller as a refrence

I want to train anfis (adaptive neuro fuzzy inference system) using a wide range of input/output data. That data comes from conventional controller like PID. PID controller can take one input at a time (like step input or ramp) and gives you output. I have tried different method to give a range of inputs to PID controller as a reference (like array input). ANFIS works well for the input, it has been trained like in my case step input (not for all step input of all values, just one input of value 1), for step of different value like step input = 2, it starts to oscillates .For the time being, I am learning how to use anfis for simple plant model that is (1/(s^2 +1)). For future use, I will use anfis as fuzzy controller to control the trajectory of mobile robot.

[AYOUB Frias]

Hello i have the same problem if anyone did find a solution to the problem please share it with me