Pages 277-282
A new self-organizing fuzzy controller design algorithm suitable for multi-objective control problem is proposed based on the orthogonal array. The proposed algorithm uses a 3-level or the combined orthogonal array that has 3-level columns to adopt nonlinear search algorithm. The search algorithm is based on a region approach incorporated with the orthogonal array, not requiring any derivatives. The essence of the proposed algorithm is to introduce a multi-objective decision table (MODT) that contains the contribution ratio defined from its F-value, optimum level, sub-decision and decision for each rule. Rules of thumb are also proposed to enhance the rule decision process, although part of the guidelines still requires experts' judgment. It is shown that the proposed algorithm can be successfully applied to the fuzzy controller design of a rigid rotor-active magnetic bearing system with the strong non-linearity in dynamics and model uncertainty.