Applications of PSO and data transformation technique in interval type-2 fuzzy identification

This paper proposed an interval type-2 fuzzy model identification method based on sigmoid data transformation technique and particle swarm optimization algorithm (PSO) to solve the problem of dynamic nonlinear system identification. To avoid the regulation of membership functions, this method applied the data transformation technique to transform the input data of fuzzy model to another domain to simplify the process of fuzzy modeling in the first. Then the fuzzy space is divided by grid diagonal method and the type-2 symmetric triangular membership function is used as the primary membership function. The forgetting factor recursive least square (FFRLS) method is applied to update the consequent parameters. At the same time, to achieve the goal of accurate modeling, the parameters in sigmoid function and the width of type-2 symmetric triangular membership function are optimized by PSO. Simulation experiments on three nonlinear systems verified the effectiveness and superiority of proposed approach.