Research On Optimization Of Elitist Non-dominated Sorting Genetic Algorithm

In the field of practical engineering and research,there are large numbers of problems which need to be optimized for more than one goal simultaneously,such problems are usually called multi-objective optimization problems.The optimal solution of multi-objective optimization problems is usually not unique,and there are usually conflicting relationships between multiple optimization goals,the optimization of a goal usually deteriorates the other goals,it is difficult for each goal to achieve its own optimal value at the same time,therefore,the solution of this kind of problem is to achieve a set of solutions which have balanced optimization performance among multiple objectives.It has important practical and theoretical significance to solve the multi-objective optimization problem.The solution of multi-objective optimization problems has received extensive attention of engineering and academia.The non-dominated sorting genetic algorithm with elitist strategy is one of the most popular multi-objective optimization algorithms,it has higher solving efficiency,can produce a number of high quality solutions in a single run,so it has become a benchmark algorithm for performance comparison of the other multi-objective optimization algorithms.With the popularization and application of NSGA-II,some problems have emerged.First of all,the running parameters of NSGA-II remain unchanged during the whole operation,it can not be adjusted adaptively according to the change of environment,so it is difficult to search efficiently for the solution space.Secondly,NSGA-II has a strong global search capability,but its local search ability is weak,which leads to the poor convergence performance of the algorithm,and it can not find the global optimal solution set,or it takes a long time to find the global optimal solution set.In view of the existing problems of NSGA-II,this paper makes a concrete research from the following two aspects:In the first aspect,an adaptive strategy based on non-dominated sorting layer isproposed in this paper to solve the problem that NSGA-II can not adjust the running parameters adaptively according to the environment.The strategy proposed in this paper dynamically adjusts the running parameters of evolutionary individuals according to the running phase of the algorithm,the evolution time,and the number of non-dominated individuals in the current temporary population.The strategy proposed in this papser can improve the self-adaptive abilities and the global search ability of the original algorithm,thus suppress premature,accelerate convergence speed and increase population diversity.In the second aspect,an optimization improvement of NSGA-II which is based on simulated annealing algorithm is proposed to solve the problem that weak local optimization ability of NSGA-II.The improvement proposed in this paper first adjusts the traditional simulated annealing algorithm,that is,the dominance relation is used as the evaluation criterion of individual,only a new solution that is superior to the original solution is accepted in the search process,and the search range is dynamically adjusted according to the annealing temperature corresponding to the current evolution.Then,the simulated annealing algorithm is introduced into NSGA-II to form a synthesis algorithm,that is,an adaptive non-dominated sorting genetic algorithm with local search,each generation of parent and offspring populations are searched by annealing algorithm in synthesis algorithm.Finally,the non-dominated sorting genetic algorithm with local search is combined with the adaptive strategy given in the firsh stage,that is,an adaptive non-dominated sorting genetic alogrithm with local search is proposed.The proposed alogrithm improve the evolutionary adaptability and local optimization ability of the algorithm,so that the ability of the original algorithm to slove the golbal optimal solution is improved,thus,it can suppress premature,accelerate convergence speed and increase population diversity,realize the optimization and improvement of NSGA-II.Analog simulation experiment is performed to validate the improments in this paper.The experiment is consisted of three parts: The first part of the experiment shows that the adaptive strategy based on non-dominated sorting layer which is proposed in this paper can effecively improve the global search ability of the original algorithm.The second part of the experiment shows that the non-dominatedsorting genetic algorithms proposed in this paper can effectively improve the local search ability of the original algorithm.The third part of the experiment shows that the adaptive non-dominated sorting genetic algorithm with local search which integrated the adaptive strategy based on non-dominated sorting layer and the non-dominated sorting genetic algorithms proposed can improve the ability of the original algorithm,it shows that the two points optimization of NSGA-II can be effectively combined,in the two aspects of convergence and diversity indicate the performance of the algorithm which proposed in this paper can improve the ability of the original algorithm.The validity of the research work on NSGA-II optimization in this paper is proved.