摘要
文章针对多目标粒子群优化算法多样性损失和收敛性不好的问题,提出了一种自适应混合多目标粒子群优化算法。首先,使用Sobol序列映射决策变量初始值,使得初始解集在全决策空间范围有更均匀的分布。使用线性递减权重法调整粒子群算法的权重,增强算法收敛性。提出了使用基于多样性指标SP的自适应变异算子增加种群多样性的同时,还提出了在最优档案集中,使用基于改进的世代距离指标GD的自适应混沌搜索增强算法局部搜索能力。最后,将文中提出的改进算法与MOPSO(基本多目标粒子群优化算法)和NSGA2对比,结果显示出该算法能够在保持优化解收敛性的同时获得更好的多样性。
Aim. The introduction of the full paper reviews a number of papers in the open literature and then proposes AHMOPSO algorithm, which we believe is better and is explained in sections 1, 2 and 3. Section 1 briefs past research. The core of section 2 consists of： ＂Firstly, the initial solution sets are mapped by the Sobol sequence to distribute the decision variables uniformly. And the linear descending weight is utilized to enhance the convergence of the algorithm. The adaptive mutating operator based on the diversity index SP is brought to add the variety of the chromosomes. In addition, the adaptive chaos searching operator based on the improved generation distance index GD is adopted to enhance the local search ability. ＂Simulation results, presented in Tables 1 through 3 and Figs. 2 through 5, compare our AHMOPSO algorithm with three generally used algorithms; the comparison shows preliminarily that AHMOPSO can indeed obtain better convergence and diversity.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2011年第5期695-701,共7页
Journal of Northwestern Polytechnical University
基金
航空科学基金(20090753008)资助
关键词
多目标粒子群优化
Sobol序列
自适应
变异算子
混沌搜索
optimization, algorithms, simulation, convergence of numerical methods, Sobol sequence
mutation operator
chaos search
