Recursive Prediction Error Algorithm for the Nufzy System to Identify Nonlinear Systems
B.T. Tien, Gerrit Van Straten · Industrial and Engineering Applications of Artificial Intelligence and Expert Systems · 2022
In our early work, an integrated neural-fuzzy approach named NUFZY has been developed for nonlinear system identification. From the literature recursive methods have been applied with success to neural networks. Due to the resemblance of the NUFZY system and multi-layered neural networks recursive methods may also be suited for the neural-fuzzy approach. In this paper the recursive prediction error algorithm is adapted for on-line parameter tuning of the NUFZY system. Examples are presented to demonstrate the feasibility of this recursive identification to models of the NUFZY type.