SUCCESSIVE IDENTIFICATION OF A FUZZY MODEL AND ITS APPLICATIONS TO PREDICTION OF A COMPLEX SYSTEM

A successive identification method of a fuzzy model is suggested. The identification mechanism consists of two levels. One is the supervisor level and the other is the adjustment level. The supervisor level determines a policy of parameter adjustment using a set of fuzzy adjustment rules. The adjustment rules are derived from the fuzzy implications of a fuzzy model and are extended to fuzzy adjustment rules by using an extended concept of Zadeh's contrast intensification. The adjustment level executes the policy of parameter adjustment determined with the fuzzy adjustment rules. The parameter adjustment consists of premise parameter adjustment and consequent parameter adjustment. Both of them are realized by the weighted recursive least square algorithm. Finally, it is shown from two examples that the method is very useful for modeling complex systems.