Research On Evolutionary Multi-objective Optimization Algorithms And Its Application

Multi-objective optimization for scientists and engineers is undoubtedly a very important research topic because practical problems with most of the characteristics of multi-objective are often difficult to be handled. Traditional methods to solve multi-objective such as: weighting method, constraint, linear programming, etc. have great limitations. Intelligent optimization algorithm with the features of highly parallel, self-organization, self-learning and self-adaptive is established by simulating a natural phenomenon or a process, which provides a new way to solve complex problems.Such algorithms include evolutionary algorithm (EA), particle swarm optimization (PSO), artificial immune system (AIS) and ant colony optimization (ACO) and so on. This paper is to study the use of evolutionary algorithms to solve multi-objective optimization problem.Evolutionary algorithm called evolutionary algorithms is a kind of stochastic optimization methods by simulating natural evolutionary process.Group search strategy and inter-group exchange of information between individuals is that it is two major characteristics superior than the traditional optimization algorithms.With the development of evolutionary algorithms, the domestic and foreign researchers propose a variety of multi-objective genetic algorithm based on the simple genetic algorithm in which non-dominated sorting Genetic Algorithm NSGA research has confirmed the best one.however, in continue to use the process of the algorithm, people found that the algorithm has many unsatisfactory ,then improved the shortcomings of the NSGA algorithm and proposed the non-elite dominated Sorting Genetic Algorithm NSGAâ…¡.At present,the research on the NSGAâ…¡both in the theoretical area and applied area abroad are more in-depth,but less in our country. This basic principle of the algorithm are studied and researched systematically and used to solve some practical problems,in solving multi-objective optimization problems NSGAâ…¡has many strong advantages and solve well these real-life problems.The main contents are as follows:(1) First of all,we introduce the evolutionary multi-objective optimization background and its three development periods;then the concept of multi-objective optimization and related terms;we analyze and summarize its most commonly useful evaluation methods and other commonly useful classification standards and classifications; discuss the commonly useful classical multi-objective optimization methods and its the basic framework.(2)In this paper,we introduce multi-objective optimization genetic algorithm and focuse on the non-dominated sorting genetic algorithm. The basic principles and shortcomings of the non-dominated sorting genetic algorithm NSGA are discussed, and its improved algorithm NSGA-â…¡,such as, fast non-dominated sorting method ,crowding and its comparison operator elitist strategy and the main process of NSGA-â…¡is researched in detail.(3)Facility location problem as a reality of the typical multi-objective optimization problems, this paper attempts to use the NSGA-â…¡algorithm to solve the waste disposal station location problem and the largest emergency facility location problem. For the former, we analyze its multi-objective mathematical model based on the consideration of the minimum of the total cost and the negative effects ,and give the NSGA-â…¡in the specific application process of the facility location and give examples of Authentication. For the latter, by the description of the characteristics of major emergencies,we establishe a three-objective-function including the minimum of the region weighted maximum of facilities and services to emergency relief needs, the greater the maximum the weight the needs of the region by the emergency services and the minimum weighted distance of the demand point in the emergency rescue facilities and services,applicate NSGA-â…¡algorithm and give the example Authentication. Experiments show that the algorithm can get Pareto optimal solution in a short time and has a good optimization result. Simulation results show that the algorithm is a feasible and effective solution to multi-objective optimization problem. We explore the NSGA-â…¡in other applications, and the algorithm has wide application prospect in solving multi-objective optimization, hoping to offer the basis for the decision-making departments.We give the conclusion of our works and present the prospect of further investigation of multi-objective evolutionary algorithm.