【作者】 王远辉；

【导师】 王秀坤；

【作者基本信息】 大连理工大学 ， 计算机应用技术， 2011， 硕士

【摘要】 本文以一类卫星舱布局优化设计问题为应用背景,研究改进的差分进化和协同差分进化算法及其在一类带约束的复杂布局优化问题的应用。该优化问题属于NP-hard问题。差分进化算法(DE)由Storn和Price于1995年提出,是一种基于种群个体间差异的典型进化计算类高效优化算法。近年来又出现协同差分进化算法(CCDE),用于求解复杂高维优化问题。本文的研究目的在于发展差分进化和协同差分进化算法,提高算法的计算性能,用于求解布局优化问题。本文的研究工作主要有：(1)提出了一种基于布局模式的人机结合差分进化算法(LPHCIDE,简称布局模式差分进化算法),用于布局优化。首先,依据布局模式对初始设计方案进行非同构变换形成人工方案。然后,对人工方案进行数值化编码构造人工个体加入DE算法种群,以指引种群进化方向,避免“早熟”现象并加速算法收敛。经Packing算例的实验结果表明,与文中其它文献算法相比,本文LPHCIDE具有较高的计算精度。(2)提出了一种基于对立策略的协同差分进化算法(OCCDE,简称对立协同差分进化算法),用于求解带约束的布局优化问题。首先,基于协同进化框架,采用“分而治之”的策略对问题进行分解,降低问题的求解难度。其次,在算法种群的初始化和进化过程中引入的对立策略,提高算法的收敛速度和计算精度。经标准测试函数和Packing算例的实验结果表明,与文中其它文献算法相比,本文OCCDE具有较高的收敛速度和计算精度。在OCCDE的基础上提出了一种对立扰动协同差分进化算法(OCCDEG),用于求解简化卫星舱布局问题。该算法在进化机制中加入随机扰动算子,增强算法的局部搜索能力。经简化卫星舱布局算例的实验结果表明,与文中其它文献算法相比,本文OCCDEG具有较高的收敛速度和计算精度。更多还原

【Abstract】 With the background of the layout design optimization of satellite module, the improved Differential Evolution (DE) and Cooperative Co-evolutionary DE (CCDE) algorithm and their application in the complex layout design optimization problem with constraints were studied in this paper. This optimization problem belongs to NP-hard and is difficult to solve.DE was proposed by Storn and Price in 1995 and is a typical effective evolutionary computation optimization algorithm, which is based on the individual difference. In recent years, CCDE has been proposed to solve complex high dimensional optimization problems. This paper aims to develop DE and CCDE to improve their computing performance for layout optimization problems.The main research work of this paper:(1) Layout Pattern Human-Computer Interactive DE (LPHCIDE) was proposed for layout optimization. Firstly, according to non-isomorphic layout pattern, the initial layout scheme was transformed into artificial scheme. Secondly, artificial schemes were encoded into artificial individuals and artificial individuals were added to the population of DE to guide population evolution, avoid getting into the local optimum and prompt DE convergence. The experiment results from packing problems show that, compared with other algorithms in this paper, LPHCIDE obtained the competitive computational precision.(2) Opposition-based Cooperative Co-evolutionary DE (OCCDE) was proposed for layout optimization problem with constraints. Firstly, based on Cooperative Co-Evolutionary Algorithm (CCEA) framework, decomposed the problem to reduce the difficulty in solving it. Secondly, opposition-based optimization is applied in subpopulation initialization and evolution to improve the convergence speed and computational precision. The experiment results from benchmark functions and packing problem show that, compared with other algorithms in this paper, OCCDE obtained the competitive convergence speed and computational precision.Based on OCCDE, OCCDE with Gaussian mutation (OCCDEG) was proposed for simplified satellite module layout problem. Gaussian mutation operator was added to OCCDE in order to enhance the local search of the algorithm. The experiment results from a layout design of simplified satellite module show that, compared with other algorithms in this paper, OCCDEG obtained the competitive convergence speed and computational precision.更多还原