Modified Biogeography-based Optimization Algorithms And Their Application For Nonlinear System Model Identification

Swarm Intelligence(SI)algorithm methods is prevalent in engineering community and rapidly growing in the field of research.Exploration and exploitation are two key issues that researchers care about.In SI,Biogeography-Based Optimization(BBO)algorithm has good performance in exploitation but not in exploration.In this paper,two modified algorithms are proposed based on BBO to solve this problem.The experiments show that modified algorithms has good performance under both test functions and nonlinear multivariable system identifying.Main innovations are as follows:1.In this paper,a modified BBO(MBBO)is presented based on BBO algorithm to overcome the premature drawback of BBO.A computing method of mutation ratio is added in MBBO to enhance exploration ability.Then,the MBBO algorithm is used for a Hammerstein model with a heavy-tailed perturbation.Experiment results verify that MBBO has much higher accuracy than BBO.2.Furthermore,another modified algorithm(Self-adaptive BBO)has been proposed to enhance the exploitation of MBBO so that the algorithm can perform better in both global search and local search.The Self-adaptive BBO can adjust its parameter with the convergence state adaptively,so it has faster search speed and could get satisfied solutions in early stage.The experiment under Wiener model with a heavy-tailed perturbation has been implemented and the results show that the improvement of this algorithm is feasible.3.To simulate the actual system more realistically,heavy-tailed noises was appended to Hammerstein model and Wiener model.Noises has been pretreatment before identifying with modified algorithms.Experiment results show that algorithms could find satisfied solutions and the application of evolutionary algorithm for nonlinear system identifying is feasible.