基于云模型的果蝇优化算法及应用研究

发布时间：2018-03-17 22:40

本文选题：群体智能　切入点：果蝇优化算法　出处：《湖南科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：群体智能算法以其实现简单灵活以及能有效解决高维、复杂的优化问题等优点,在计算智能领域获得了越来越多的关注。果蝇优化算法是一种模拟果蝇觅食行为的群体智能优化算法。本文结合云模型对果蝇优化算法提出了一些改进策略,以提高算法的收敛性能,并进一步拓展到多目标优化和实际工程应用。本文主要工作包括以下几个方面。论文首先介绍了果蝇优化算法的研究背景及意义,回顾和总结了果蝇优化算法的国内外研究现状和进展,对果蝇优化算法的原理和步骤进行了详细的介绍,并给出了果蝇优化算法的伪代码。然后,对云模型的定义和正态云发生器的实现进行了相关介绍。为了提高果蝇优化算法的全局收敛能力和收敛精度,提出了一种基于云学习的双态果蝇优化算法。算法借鉴自然界群体分工的特性,在寻优过程中将果蝇群体分为“搜索”和“捕食”两种状态的种群,平衡算法的全局搜索与局部开采能力。另外,利用云模型描述觅食过程中的随机性和模糊性,增强逃离局部最优的能力。利用23个Benchmarks测试函数对所提算法进行了测试,实验结果表明,所提方法能显著提高算法的全局收敛能力和收敛精度。鉴于基于浓度判定值计算的候选解产生机制存在易陷入早熟收敛以及不能优化最优解为负值的优化问题,在新的候选解产生机制的基础上结合正态云模型,提出了一种基于正态云模型的果蝇优化算法。算法利用正态云模型对果蝇觅食过程中的随机性和模糊性进行描述,提高搜索效率。提出了一种正态云模型参数自适应策略,使算法前期具有较强的全局收敛能力,后期拥有良好的收敛精度。利用33个Benchmarks测试函数对算法进行了测试,实验结果表明,所提算法能获得良好的收敛性能。结合Pareto占优概念以及外部精英存档策略,提出了一种基于云模型的多目标果蝇优化算法,将果蝇优化算法拓展到多目标问题的优化。采用基于归一化最近邻域多样性测量方法来保持非占优解集的分布性和多样性。利用WFG和CEC2009多目标测试问题组对所提算法进行了测试,实验结果表明,所提算法获得的非占优解集能够较好的趋近于真实Pareto前沿,并且能保持良好的散布性。为了进一步验证所提方法在实际的工程优化设计中的有效性,将基于正态云模型的果蝇优化算法应用于永磁同步电机参数辨识,将基于云模型的多目标果蝇优化算法应用于飞行器热导管参数优化设计以及减速器优化设计。实际应用系统模型的实验结果证实了所提方法的有效性。
[Abstract]:Swarm intelligence algorithm has the advantages of simple and flexible implementation and can effectively solve high-dimensional and complex optimization problems. More and more attention has been paid in the field of computational intelligence. Drosophila optimization algorithm is a swarm intelligence optimization algorithm which simulates the foraging behavior of Drosophila melanogaster. In this paper, some improved strategies are proposed based on cloud model for Drosophila optimization algorithm. In order to improve the convergence performance of the algorithm, and further expand to multi-objective optimization and practical engineering applications. The main work of this paper includes the following aspects. Firstly, the background and significance of Drosophila optimization algorithm are introduced. This paper reviews and summarizes the research status and progress of Drosophila optimization algorithm at home and abroad, introduces the principle and steps of Drosophila optimization algorithm in detail, and gives the pseudo code of Drosophila optimization algorithm. The definition of cloud model and the implementation of normal cloud generator are introduced. In order to improve the global convergence ability and convergence accuracy of Drosophila optimization algorithm, A two-state optimization algorithm for Drosophila fly based on cloud learning is proposed. The algorithm uses the characteristics of natural population division for reference and divides Drosophila population into "searching" and "predatory" populations in the process of optimization. In addition, the cloud model is used to describe the randomness and fuzziness in the foraging process to enhance the ability to escape from the local optimum. 23 Benchmarks test functions are used to test the proposed algorithm. Experimental results show that the proposed method can significantly improve the global convergence ability and convergence accuracy of the algorithm. In view of the problem that the candidate solution generation mechanism based on the concentration decision value calculation is prone to premature convergence and can not optimize the optimal solution to negative value. Based on the new candidate solution generation mechanism, an optimization algorithm based on normal cloud model is proposed to describe the randomness and fuzziness of the foraging process of Drosophila melanogaster. In order to improve search efficiency, a parameter adaptive strategy for normal cloud model is proposed, which makes the algorithm have strong global convergence ability in the early stage and good convergence accuracy in the latter stage. 33 Benchmarks test functions are used to test the algorithm. Experimental results show that the proposed algorithm can achieve good convergence performance. Combined with the concept of Pareto dominance and the external elite archiving strategy, a multi-objective Drosophila optimization algorithm based on cloud model is proposed. The optimization algorithm of Drosophila is extended to the optimization of multi-objective problem. Based on the normalized nearest neighborhood diversity measurement method, the distribution and diversity of non-dominant solution sets are maintained. WFG and CEC2009 multiobjective test problem groups are used to solve the problem. The algorithm is tested, The experimental results show that the set of non-dominant solutions obtained by the proposed algorithm can reach the real Pareto front and maintain good dispersion. In order to further verify the effectiveness of the proposed method in the practical engineering optimization design. The optimal algorithm of Drosophila based on normal cloud model is applied to parameter identification of permanent magnet synchronous motor (PMSM). The multi-objective Drosophila optimization algorithm based on the cloud model is applied to the optimization design of the thermal duct parameters and the reducer of the aircraft. The experimental results of the practical application system model show the effectiveness of the proposed method.
【学位授予单位】：湖南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】