A Unified Design for the Membership Functions in Genetic Fuzzy Systems

This paper introduces a unified design for the membership functions in genetic fuzzy systems (GFSs). The membership functions can get one of the symmetric or asymmetric forms of triangles, exponential Gauss, trapezoid and so on in the general unified form. With these unified forms, the meanings of the parameters of the membership functions become more clearly and more understandable. Moreover, these unified forms permit our proposed genetic algorithm (GA) to tune the parameters of the membership functions one by one. This makes our method different because other methods always simultaneously tune the parameters. With this new approach, the better parameters are easier to find out via parameter learning process. In addition, appropriate inferences are defined in the systems. As a result, the obtained genetic fuzzy systems are more effective and more compact.