Experimental Study on Improved Differential Evolution for System Identification of Hammerstein Model and Wiener Model

Abstract

For nonlinear system of the Hammerstein model and Wiener model, a method for nonlinear system identification is proposed based on differential evolution Algorithm (DE). The based idea of the method is that the problem of nonlinear system identification is changed into optimization problems in parameter space. In order to enhance the performance of the DE identification, put forward a kind of adaptive mutation differential evolution algorithm for scaling factor (MDE), and on this basis, we make an improvement on crossover to make a better performance. To make an analysis for particle swarm optimization (PSO), DE and improved DE, the improvement DE has higher accurate and recognition ability, stronger convergence.