Research On Multi-Population Based Evolutionary Algorithms For Multi-Modal Optimization Problems

Many real-world optimization problems have multi-modal optima(called multi-modal optimization problems),which require an optimization algorithm to find as many as possible optima.In recent years,investigating EAs for multi-modal optimization problems attracted a growing interest,and multi-population methods is one of the most commonly used approaches in multi-modal optimization.The advantages and disadvantages of the multi-population methods are comprehensively analyzed by a set of experimental studies in different multi-modal optimization environments.This paper proposes a more effective multi-population method to enhance the capacity of EAs in multi-modal optimization problems.Firstly,this paper summarizes researches on stationary multi-modal evolutionary algorithms and dynamic multi-modal evolutionary algorithms and give a brief introduction of particle swarm optimization.Secondly,this paper analyses the differences between SPSO algorithm and CPSO algorithm,proposes a new algorithm by combining Species mechanism and Cluster mechanism and tests the efficiency of the proposed algorithm for multi-modal optimization problems.Lastly,this paper tests the influence of parameters on dynamic multi-modal optimization algorithms and investigates a novel multi-population method which is effective by testing on dynamic optimization problems.