计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (13): 131-136.DOI: 10.3778/j.issn.1002-8331.1702-0182

• 模式识别与人工智能 • 上一篇    下一篇

基于SQP和自适应搜索的混沌粒子群算法

郑庆新,顾晓辉,张洪铭   

  1. 南京理工大学 机械工程学院,南京 210094
  • 出版日期:2018-07-01 发布日期:2018-07-17

Chaotic particle swarm optimization algorithm based on SQP and adaptive search

ZHENG Qingxin, GU Xiaohui, ZHANG Hongming   

  1. School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
  • Online:2018-07-01 Published:2018-07-17

摘要: 针对基本粒子群优化算法(PSO)算法易陷入局部最优的缺点,提出混沌自适应粒子群-序列二次规划算法(CAPSO-SQP)。在基本PSO算法的基础上,加入混沌搜索和自适应惯性权重提高全局收敛能力,并在PSO算法每一代的迭代过程中,引入SQP策略,加快局部搜索并提高对约束优化问题的计算可靠性。测试函数仿真结果表明,CAPSO-SQP算法计算精度高,稳定性好,收敛速度快。将所提出算法应用于悬臂梁结构优化设计,求解结果表明算法在结构优化计算方面的可行性,而且相对于CPSO算法求解更加准确,具有较高的计算可靠性和实用价值。

关键词: 粒子群算法, 序列二次规划, 混沌搜索, 自适应惯性权重

Abstract: Chaotic Adaptive Particle Swarm-Sequence Quadratic Programming(CAPSO-SQP) is proposed to overcome the shortcomings of basic PSO algorithm. Based on the basic PSO algorithm, the chaos search and the adaptive inertia weight are added to improve the global convergence ability. In each iteration of the PSO algorithm, the SQP is introduced to speed up the local search and improve the whole searching effectiveness and the computational reliability of constrained optimization problems. The simulation results show that CAPSO-SQP algorithm has high accuracy, good stability and fast convergence. The cantilever structure optimization design results show the feasibility of the algorithm in structural optimization problems, and the solution with respect to CPSO is more accurate and has high reliability and practical value.

Key words: particle swarm algorithm, sequential quadratic programming, chaos search, adaptive inertia weight