Studies On Evolutionary Multi-objective Optimization Theory,Algorithm Design And Application

Multi-objective optimization problems are fairly common in many practical and a-cademic areas.Traditional mathematical methods often prove inadequate to solve these problems.The development of computer technology in the past two decades has made possible the current rise of computational intelligence.As an important branch of com-putational intelligence,evolutionary computation for multi-objective optimization has become an important issue.The widespread application of evolutionary multi-objective algorithms is having a profound effect on scientific research and practical application.However,the theoretical research of evolutionary multi-objective algorithms has lagged far behind a.lgorithm design and application.This thesis mainly focuses on the theoretical study of evolutionary multi-objective algorithms and the according algorithm design based on the resulting theoretical conclu-sions.The applications of the proposed algorithms in practical problems are also studied.The main contributions of this thesis are summarized as follows:(1)We defined a class of multi-objective optimization search imbalance problems,and then theoretically analyzed the properties of those search imbalance problems.Based on the theoretical results,we produced three main types of search imbalance problems and a set of search imbalance benchmark problems were constructed.(2)We studied the theory of explicit control of implicit parallelism in decomposition based evolutionary many-objective optimization algorithms.(3)We studied the effect of objective normalization and penalty param-eter on Penalty Boundary Intersection-based decomposition based evolutionary many-objective optimization algorithms.(4)We theoretically studied the role played by the non-dominated sorting in decomposition based NSGA-? algorithm,and experimentally proved that non-dominated sorting is necessary in many-objective optimization.(5)We studied the dynamic search resource allocation theory and proposed a dynamic search resource allocation strategy for degenerated many-objective optimization problems with complex Pareto-Front(PF).(6)We studied the theoretical connection between the three commonly used decomposition methods.(7)We applied the population decomposition strategy to a new multi-objective search imbalance tracking area planning problem mod-el.(8)We studied a multi-objective triclustering model and developed a new population decomposition-based evolutionary multi-objective optimization algorithm to solve this model.