基于跳跃基因的多目标差分进化算法研究

发布时间：2018-05-28 02:25

本文选题：多目标优化算法 + 差分进化算法　；参考：《暨南大学》2016年硕士论文

【摘要】：在科学研究和工程设计过程中,很多具体问题都可以归纳为参数优化问题,而现实当中,这些优化问题往往有多个设计目标,这些目标互相矛盾,彼此制约,一个目标的性能优化往往会导致其它至少一个目标的性能退化,即多个目标很难同时达到最优。因此,多目标优化算法的研究成为当今科学和工程设计当中的一个热点研究方向。进化算法是一种受自然界生物和系统的启发而发展起来的启发式搜索和优化算法的总称,用进化算法求解多目标优化问题得到了广泛的应用。作为进化算法的重要组成部分,差分进化算法是一类容易理解、结构简单、可调参数少、鲁棒性强的智能优化方法。多目标差分算法中,较著名的算法有DEMO(Differential evolution algorithm for multi-objective optimization)和MODEA(Multi-objective differential evolution algorithm)等。然而,DEMO使用的快速父代替换和MODEA使用贪婪变异策略使其在求解复杂问题时存在明显的不足,容易陷入局部最优。为解决现有的多目标差分进化算法容易陷入局部最优的问题,本文引入跳跃基因算子,提出基于跳跃基因的多目标差分进化(JGMODE)算法。不同于现有的多目标差分算法如DEMO和MODEA,JGMODE算法在传统交叉算子之后执行跳跃基因操作,以保持种群多样性,提高算法的勘探能力。数值实验结果表明提出的算法能很好地解决局部最优问题,在ZDT和DTLZ测试函数集的困难问题上展现了显著优于现有算法的性能。
[Abstract]:In the process of scientific research and engineering design, many specific problems can be summed up as parameter optimization problems, but in reality, these optimization problems often have multiple design objectives, which are contradictory and restrict each other. The performance optimization of one target often leads to the degradation of the performance of at least one other target, that is, it is difficult to achieve the optimization of multiple targets at the same time. Therefore, the research of multi-objective optimization algorithm has become a hot research direction in science and engineering design. Evolutionary algorithm is a kind of heuristic search and optimization algorithm inspired by natural organisms and systems. It is widely used to solve multi-objective optimization problems with evolutionary algorithm. As an important part of evolutionary algorithm, differential evolutionary algorithm is a kind of intelligent optimization method, which is easy to understand, simple in structure, with few adjustable parameters and strong robustness. Among the multi-objective difference algorithms, the better known algorithms are DEMO(Differential evolution algorithm for multi-objective optimization and MODEA(Multi-objective differential evolution algorithm). However, the rapid-parent method used by demo and the greedy mutation strategy used by MODEA make it easy to fall into local optimum because of its obvious deficiency in solving complex problems. In order to solve the problem that the existing multi-objective differential evolution algorithm is easy to fall into local optimum, a jump gene operator is introduced in this paper, and a multi-objective differential evolution algorithm based on jump gene is proposed in this paper. Different from the existing multi-objective differential algorithms such as DEMO and MODEAN JGMODE, the jump gene operation is performed after the traditional crossover operator to maintain population diversity and improve the exploration ability of the algorithm. Numerical results show that the proposed algorithm can solve the local optimal problem well, and the performance of the proposed algorithm is significantly better than that of the existing algorithms on the difficult problem of ZDT and DTLZ test function sets.
【学位授予单位】：暨南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP18

【相似文献】