A fuzzy controller that adjusts its structure in relation to changes in system or disturbance dynamics is proposed. The controller output membership functions are tuned for optimum performance. The tuning is implemented by a control supervisor that makes its decisions depending on the reliability of the current process model. If the model is reliable, the prediction error is used to tune the controller, otherwise, the tuning is realized by a measure of past performance. To prove the feasibility of the controller, the proposed technique is applied to the speed control of a DC motor.

This content is only available via PDF.
You do not currently have access to this content.