Learning of rule importance for fuzzy controllers to deal with inconsistent rules and for rule elimination

Extraction of correct arid precise rules front experts is a difficult probl em, Moreover, even when the extracted rules are correct, all of them may no t, have equal importance to achieve the goal of the fuzzy system. Rule tuni ng is usually achieved through modification of membership functions. Effect of changing a member ship function is global in the sense, it influences a ll rules that involve the membership function:. Here we propose an effectiv e extension of the ordinary fuzzy controller model which incorporates an im portance factor for each rule. The importance factor allows tuning of the s ystem at the rule level. Of course, one carl still tune the membership func tions. The extended model enables us to cope with incorrect and/or incompat ible rules anti thereby enhances the robustness, flexibility anti system mo deling capability. It also helps us to eliminate redundant rules easily. Fo r the. Takagi-Sugeno frame work, we derive the learning algorithm for the r ule importance factor as well as that, for the consequent. We demonstrate t he superiority of the extended model through extensive simulation results u sing the inverted pendulum.