Real time control of a direct drive motor by a learning neuro-fuzzy controller

A neuro-fuzzy controller is presented which uses neural networks to modify the parameters of an adaptive fuzzy logic controller. The adaptiveness of the fuzzy controller is derived from a rule generation mechanism and changing the scaling factor or the shape of the membership functions. The neural network functions as a classifier of the system´s temporal responses. A multilayer perceptron is used to classify the temporal response of the system into different patterns. Depending on the type of pattern such as “response with overshoot”, “damped response”, “oscillating response”, etc. the scaling factor of the input and output membership functions are adjusted to make the system respond in a desired manner. The rule generation mechanism also utilizes the temporal response of the system to evaluate new fuzzy rules. The non-redundant rules are appended to the existing rule base during the tuning cycles. This controller architecture is used in real-time to control a direct drive motor. The control system hardware utilizes a digital signal processor and a PC to implement the controller architecture. Experimental results are illustrated