부산대학교 도서관

Hammerstein model can finely express nonlinear characteristics of a practical process. But there is no general analytical method for the identification of Multiple-Input and Multiple-Output (MIMO) Hammerstein model, especially under cases with heavy tail noise. Meanwhile BA algorithm, as a novel heuristic optimization method, has a strong computing power to solve nonlinear complex problems. In this paper, firstly we propose an Opposite-based learning mean and Cauchy mutation BA algorithm (OCBA) to strengthen global and local exploration ability of standard BA algorithm. Then we transform the identification problem into an optimization problem and apply the proposed OCBA. Numerical simulations are carried out to verify the superiority of the OCBA for MIMO Hammerstein model identification.