Research On Identification Of Nonlinear System Based On PSO Method

This paper makes a research on methods how to make use of Particle Swarm Optimization(PSO) method to identify nonlinear system with constrained conditions and the ways to improve the identification accuracy with improving identification performance by designing new objective function and PSO algorithm.First, the definition with the principal and practical meaning is introduced briefly, followed by the history of system identification. Then the basic principal and common nonlinear system models are introduced. A detailed introduction of several kinds of traditional identification methods is introduced and the requirements for identification methods are given and analyzed. Then the principal and process of basic PSO method are given and the performance of PSO is analyzed. The construction and mathematical expression of traditional penalty function are introduced. Based on adaptive penalty function method an improved adaptive penalty function is put forward to increase the identification accuracy. Finally, in order to decrease the probability of"premature"problem and increase the speed of convergence,an improved PSO algorithm named Double-PSO algorithm is provided. All the algorithms in this paper are simulated and compared in Matlab and the simulation results show that the proposed penalty function performs well in improving identification accuracy and Double-PSO algorithm decreases the probability of"premature"problem with increasing the speed of convergence effectively.