Extracting fuzzy rules for system modeling using a hybrid of genetic algorithms and Kalman filter

This paper proposes a hybrid algorithm for extracting important fuzzy rules from a given rule base to construct a "parsimonious" fuzzy model with a high generalization ability. This algorithm combines the advantages of genetic algorithms' strong search capacity and Kalman filter's fast convergence merit. Each random combination of the rules in the rule base is coded into a binary string and treated as a chromosome in genetic algorithms. The binary string indicates the structure of a fuzzy model. The parameters of the model are then estimated using the Kalman filter. In order to achieve a trade-off between the accuracy and the complexity of a fuzzy model, the Schwarz-Rissanen Criterion is used as an evaluation function in the hybrid algorithm. The practical applicability of the proposed algorithm is examined by computer simulations on a human operator modeling problem and a nonlinear system modeling problem.