Rule base refactoring design for fuzzy logic controllers

Although the requirement of logic completeness is not necessary, most engineers are used to building a complete rule base, due to a lack of knowledge about the physical properties of a system. This situation leads to unnecessary computations in rule inference and the control algorithm may be intractable in real time. It is worthwhile to refine a complete rule base by means of some strategies. In this paper, a rule base refactoring method is proposed. This method, which can reduce the number of rules without having any other significant effect, is inspired by the methods for gates minimization of logic design. Experiments were conducted to study the control of an inverted pendulum and backer-upper truck due to their apparent property of non-linearity. The results of the experiments show that the control behavior of the fuzzy controller with the tuned rule base is still similar to that of the controller with the original rule base.