TSK Function Approximator Design Using GA and PSO with Minimum Membership Function and Guaranteed Approximation Error

Fuzzy approximators have a lot of application in the field of applied science, such as system model extraction from input/output data and simplifying the mathematical complex functions. The main aspect in the design of such an approximator is achieving a good level of approximation error. In this area there are a few works which have some problems such as huge number of fuzzy membership functions, large rules database and calculation consumption load that make them useless for real-time applications. Also there are a few methods with no specific algorithm for finding the rules database. This paper proposes a new method to design a fuzzy function approximator using combination of GA and PSO algorithms. The proposed approximator not only guarantees the desiredapproximation error but also minimizes the rules database, so make it useful at real-time applications. The simulation results show the performance of this method

Keywords:

Function Approximation , Fuzzy Takagi Sugeno-Kang (TSK) model , Genetic Algorithm , Particle Swarm Optimization