The Study On Intelligent Algorithms For Multiobjective And Many-objective Optimization Problem

Recently,the study in multi-objective intelligent algorithms,such as multiobjective evolutionary algorithms(MOEAs)and multi-objective particl swarm optimization algorithms(MOPSOs),has become a hot topic.The aim of this dissertation is to explore the theories and mechanisms of multi-objective optimization problem(MOP)and many-objective optimization problem(MaOP),to design modified strategies to overcome the shortcomings of existing MOEAs and MOPSOs,to improve the performance of these two intelligent algorithms on all kinds of MOPs and MaOPs?The main work in this dissertation include the following three aspects:In constrained decomposition-based MOEAs,the used weight vectors with uniform distribution cannot always match the complex Pareto-optimal front,which may lead to the disequilibrium of solutions associated to the subproblems,then influence the population's diversity.To tackle the above problem,this paper suggests a novel MOEA with dynamic decomposition strategy.In the proposed algorithm,after the evolution,all the parents and offspring populations with the population size N are combined as a union population,which are associated to the preset N weight vectors using the constrained decomposition strategy.By counting the number of solutions that fall within each subproblem's feasible region,the distribution of union population can be shown and then the number of subproblems that are not associated to any solution can be calculated to indicates the number of subproblems to be regenerated.Thus,the subproblem associated with the largest number of solutions will be found and further decomposed into two new subproblems using the designed dynamic decomposition method with clustering.At last,a simple convergence indicator is used to select one solution showing the best convergence for each subproblem.The experiment results validate the effectiveness of the designed strategy.It was found that most MOPSOs perform poorly when tackling MaOPs.This is mainly because the loss of selection pressure that occurs when updating the swarm.The number of non-dominated individuals is substantially increased and the diversity maintenance mechanisms in MOPSOs always guide the particles to explore sparse regions of the search space.This behavior results in the final solutions being distributed loosely in objective space,but far away from the true Pareto-optimal front.To avoid the above scenario,this paper presents a balanceable fitness estimation method and a novel velocity update equation,to compose a novel MOPSO,which is shown to be more effective to tackle MaOPs.Moreover,an evolutionary search is further run on the external archive in order to provide another search pattern for evolution.To solve the difficulty of balancing convergence and diversity in high dimensional objective space,this paper suggests a novel clustering-based evolutionary algorithm for MaOPs.The main idea of this algorithm is to classify the population into a number of clusters,in which each individual in the same cluster should show high similarity based on the vector angle.Thus,these clusters actually help to portray the population's distribution.To obtain these clusters efficiently,partitional clustering is firstly used to separate the population into m main clusters based on the m axis vectors(m is the number of objectives),and then agglomerative hierarchical clustering is further run on the m main clusters to get N final clusters(N is the population size,and N > m).At last,in environmental selection,one individual from each of the clusters closest to the axis vectors is selected to maintain diversity,while one individual from each of the rest clusters is preferred by a simple convergence indicator to ensure convergence.The experiment results conform the effectiveness of the designed evolutionary algorithm.