Research On Multiobjective Evolutionary Computation And Its Application On Constrained Optimization

Many optimization problems encountered in the disciplines of science and engineering are multi-objective problems (MOPs), therefore, to do research on MOPs has a great value. Evolutionary algorithms of Artificial Intelligence were initially applied to those problems from the mid-eighties. Thus a popular area of research has formed recently. Research on multiobjective evolutionary algorithms for multiobjective optimization problems is still very importantThis dissertation is starts with the discussions of general definition of multi-objective optimal problem, Pareto optimal Solutions, Pareto optimal Front and reviewing the advances of MOEAs(Multi-Objective Optimization Evolutionary Algorithms) at home and abroad. Then, two MOEAs and one COPs(Constrained Optimization Problems) based on Multi-objective optimization are proposed by author. The main contribution and work are described as following:1,Proposed a new mutiobjective evolutionary algorithm. based on ELECTRE method. The proposed algorithm uses a secondary population in order to retain the non-dominated solutions found during the evolutionary process and adopts the same fitness assignment strategy as SPEA-II to get well distributed solutions. Additionally, a novel outranking relationship is constructed, and proved to be weaker than Pareto dominance relation. Experiment results show that this algorithm can converge to true Pareto Front well and effectively maintain diversity of the solutions.2,Investigate a new multi-objective evolutionary algorithm based on differential evolution. The proposed approach adopts a secondary population in order to retain the non-dominated solutions found during the evolutionary process. Additionally, the approach also incorporates the concept ofε-dominance to get a good distribution of the solutions retained. We adopted standard test functions and performance measures reported in the specialized literature to validate our proposal. Our results are compared with respect to two approaches that are representative of the state-of-the-art in the area: the NSGA-II andε-MOEA. 3,Incorporated the idea of mulitobjective optimization into Constrained Optimization problems. Proposed the non-dominated individual replacement scheme. Additionally an infeasible solutions archiving and replacement mechanism is introduced to effectively exploit infeasible solutions. The algorithm is tested on six benchmark functions, and the result shows that this algorithm is outperforms others compared with some other state-of-the-art algorithms.Finally, this dissertation points out some directions that are worthy to be researched further in this area.