进化算法解决多目标优化问题

发布时间：2018-01-12 15:16

本文关键词：进化算法解决多目标优化问题　出处：《中原工学院》2017年硕士论文　论文类型：学位论文

【摘要】：进化算法(Evolutionary Algorithm,EA)在处理多目标优化问题(Multi-objective optimization problem)和在实际中应用是当前研究的热门问题,随着待优化问题的维度越来越高、目标变得更多,算法也随之变得复杂,传统的粒子群、差分进化等算法已不能有效处理此类问题。多目标进化算法(Multi-objective Evolutionary Algorithm,MOEA)随之出现。它在处理上述问题中表现杰出。本文在传统优化算法的基础上,重点研究介绍了多目标粒子群算法(MOPSO,Multi-Objective Particle Swarm optimization algorithm)来处理多目标优化问题。将差分进化算法(Differential evolution algorithm,DE)在解决环境经济调度(Environment/Economic Dispatch,EED)中应用。具体内容如下:首先介绍了粒子群算法(PSO)的概念和多种建立在PSO基础上的改进算法。将它们在测试函数中的表现进行对比并分析。对深入研究MOPSO算法提供了理论基础。其次利用PSO算法的先验知识。针对传统PSO算法不能处理多目标问题。提出了加入外部存档和局部扰动策略的MOPSO算法。从实验结果看出,本文介绍的MOPSO比之NSGA2可得到更好的Pareto前沿。之后详细介绍DE算法的基本概念。列举多种扩展模式并比较。同时介绍多种改进的DE算法,并在测试函数中测试比较。本章对后期研究DE算法在电力系统EED应用打下理论基础。最后介绍电力系统EED问题。此问题是多目标、多约束并且是高维的优化问题,传统算法没有办法处理。为此,利用DE算法的先验知识,用启发式策略解决多个约束条件并加入优先列表方法。实验结果表明。本文的算法能使耗能低的发电机拥有更高优先级去更多的输出电力。从而得到更好的解决方案。
[Abstract]:Evolutionary algorithms (Evolutionary Algorithm, EA) in dealing with multi-objective optimization problems (Multi-objective optimization problem) and the application in practice is a hot issue of the current study, with more and more dimensions for the optimization problem, the goal becomes more, algorithm becomes complicated, the traditional particle swarm optimization, differential evolution algorithm has not effectively deal with such problems. The multi-objective evolutionary algorithm (Multi-objective Evolutionary Algorithm, MOEA) appears. Its performance in dealing with these issues outstanding. Based on the traditional optimization algorithm, the key research introduces multi-objective particle swarm algorithm (MOPSO, Multi-Objective Particle Swarm optimization algorithm) to solve multi-objective optimization problem. The difference the genetic algorithm (Differential evolution algorithm, DE) in solving the environmental economic dispatch (Environment/ Economic Dispatch, EED) in the application. The specific contents are as follows: firstly introduces the particle swarm optimization (PSO) algorithm and the concept of variety based on PSO. Their performance in the test function were compared and analyzed. Provide a theoretical basis for further study of the MOPSO algorithm. Then PSO algorithm using prior knowledge. According to the traditional PSO algorithm cannot handle the multi-objective problem is proposed. The addition of external archive and local perturbation strategy of MOPSO algorithm. The experimental results show that the MOPSO NSGA2 can be better than Pareto. After the detailed introduction to the basic concept of advanced DE algorithm. A list expansion mode and compared. At the same time, a variety of improved DE algorithm is introduced, and the comparison test in the test function. This chapter of the late DE algorithm to lay a theoretical foundation in power system EED power system EED application. At the end of the paper. This problem is a multi-objective, multi constraint and high dimension The optimization problem, the traditional algorithm can not deal with. Therefore, DE algorithm using the prior knowledge, the heuristic strategy to solve the multiple constraints and to priority list method. Experimental results show that this algorithm can make. The generator has low energy consumption of higher priority to more output power. In order to get a better solution.

【学位授予单位】：中原工学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】