A new approach for on-line optimisation of a fuzzy controller
Joseph Haggège, Mohamed Benrejeb, Pierre Borne · 2002
In this paper, a new method for on-line optimisation of a fuzzy controller is presented. It is based on a Sugeno fuzzy controller which is on-line tuned to adjust its parameters, using a gradient method. The purpose of the proposed approach is to give an algorithmic way to determine the parameters of such a controller, without using human knowledge, as it is usual in the design of fuzzy controllers, while preserving a human understanding of the fuzzy controller's action. An illustrative example is given by the application of the developed method to the control of a second order linear system.