Optimal recursive fuzzy model identification approach based on particle swarm optimization

In this paper a method for recursive Takagi Sugeno fuzzy model identification using weighted recursive least squares with optimal initial condition of the parameters based on particle swarm optimization (PSO) applied to static nonlinear and time-variant delay systems is proposed. For this approach, it is used a priori knowledge of the nonlinearities of interest in the system to improve the quality of the identified TS fuzzy model, characterizing an gray box identification procedure. The methodology consists of two distinct steps: In the first, the initial condition is obtained by batch weighted least-squares identification, optimized by PSO based on nonlinear static characteristic curves; In the second step, the optimal model is used as initial condition for weighted recursive least squares identification. Experimental results show the efficiency of the proposed methodology for real time identification of a thermal plant.