The Research Of Multi-objective Optimization Based On Evolutionary Algorithm

In this paper, the multi-objective optimization is discussed. Evolutionary algorithmhas overcome the shortages of complexity computing, inaccurate solutions andcomplicated procedures in traditional algorithm to solve multi-objective. As a model ofevolutionary algorithm to solve multi-objective optimization problems, the NSGA-IIalgorithm has been widely used and researched because of its low computationalcomplexity, the accurate solutions and population diversity.This paper expounds the background and the present situation of evolutionaryalgorithm and assignment problem from the point of significance of the studying.Multi-objective genetic algorithm are introduced some commonly used, of which theNSGA-II algorithm is the most widespread. Then the pros and cons of the NSGA-IIalgorithm are discussed from diversity, efficiency and local optimum, with theimprovement measures put forward. Optimization algorithm has been improved frompopulation diversity and locally more accurate solutions in the same time. Theimprovement of crossover operator based on Gaussian distribution which has localconcentration and uniform volatility. In this way, most of crossover operator stable in acertain area but increase vast search space. In this paper we use poisson variation not onlyhas local escape rapidly but also has accurate search ability. Many test functions have beenused to testify the effectiveness of the algorithm. Comparison results showed thatimproved method is better than the previous in the population diversity and locally moreaccurate solutions. Then apply the NSGA-II to specific problems. Establish the expectedassignment model under the assignment problem of utility matrix of the fuzzy information,with credibility theory dealing with uncertain information in the third part of this articleusing improved the NSGA-II algorithm. Finally, summarize the full text system, and pointout the advantages of work and the deficiency in this paper.