A Method for Fuzzy System Identification Based on Clustering Analysis
George E. Tsekouras, Haralambos Sarimveis, George Bafas · Systems Analysis Modelling Simulation · 2002
In this paper a methodology which identifies fuzzy systems is developed. In the first place, an algorithm that employs a distance function to automatically generate fuzzy rules is proposed. In addition to that, this algorithm gives an estimation of the system parameters, which are used as initial values for the iterative parameter optimization that follows. A clustering analysis is adopted to optimize the premise parameters and the least-squares method is used to optimize the consequent parameters. The number of rules is controlled by the performance of the system. Finally, simulations show the validity of the proposed method.