Research On Evolutionary Algorithm For Large Scale Function Optimization Problem And Application

Large scale function optimization problem is widely used, that lots of useful engineeringapplications can be abstracted into this model and find solutions. Evolutionary algorithmshave been considered as an effective way in solving function optimization problem. Andbecause of simpleness, accuracy and feasibility, Evolutionary algorithms have been used inmany fields. Aiming at Large scale function optimization problem, DifferentialEvolution(DE)and cooperative co-evolution framework(CC)was respectively studied.First, after fully understanding prime concepts and contents of DE, aiming at adjustmentsof control parameters in this algorithms,based on Cloud Model,especially making full useof the outstanding characteristics of this model, a self-adaptive differential evolutionalgorithm based on cloud model(CMSaDE) was proposed, which can dynamically changemutation strategies and control parameters instead of manual settings. The performance ofCMSaDE is investigated and compared with other well-known approaches and the resultsshow that CMSaDE generally outperforms other DE algorithms nearly in all the benchmarkfunctions.Then, for how to decomposing decision variables for large scale function optimizationproblem in cooperative co-evolution framework, the classic cooperative co-evolutionframework is extended. At first, a mix cooperative co-evolution framework was proposed.And then, a correlation identification computation was introduced into this framework.Putting these together, a mixed cooperative co-evolution algorithm based on correlationidentification grouping for Large scale function optimization(DEMCC-CIG)was proposed.The performance of DEMCC-CIG is investigated and compared with other well-knownapproaches on the benchmark functions CEC08.At last, combining CMSaDE with DEMCC-CIG, a preliminary research was exploredabout multi-robot path planning problem with an experiment.