多峰函数优化的改进群居蜘蛛优化算法  

Improved social spider optimization algorithm for multimodal function optimization

作　　者：王丽[1] 王晓凯[2] WANG Li;WANG Xiaokai(School of Information Technology and Engineering, Jinzhong University, Jinzhong, Shanxi 030619, China;School of Physics and Electronic Engineering, Shanxi University, Taiyuan 030006, China)

机构地区：[1]晋中学院信息技术与工程学院,山西晋中030619 [2]山西大学物理电子工程学院,太原030006

出　　处：《计算机工程与应用》2017年第3期1-6,共6页Computer Engineering and Applications

基　　金：教育部高等学校教学指导委员会项目(No.JZW-14-JW-09);山西省高等学校教学改革项目(No.J2014108);山西省科技攻关计划项目(No.20110321025-02);晋中学院教学改革项目(No.ZL2016jg04)

摘　　要：针对群居蜘蛛优化(SSO)算法求解复杂多峰函数成功率不高和收敛精度低的问题,提出了一种自适应多种群回溯群居蜘蛛优化(AMBSSO)算法。引入自适应决策半径概念,动态地将蜘蛛种群分成多个种群,种群内适应度不同的个体采取不同的更新方式,提高了种群样本多样性;提出回溯迭代进化策略,在筛选全局极值的基础上,根据进化程度执行回溯迭代更新,保证了算法全局寻优能力。高维多峰函数仿真结果表明,同SSO算法、PSO算法等优化算法相比,AMBSSO算法具有较快的收敛速度和较高的收敛精度,尤其适用复杂高维多峰函数优化问题。An Adaptive Multi-swarm Backtracking Social Spiders Optimization(AMBSSO)is proposed to solve thecomplex multimodal function optimization problems of Social Spiders Optimization(SSO)algorithm which has lowsuccess rate and convergence precision.The adaptive decision radius is introduced in SSO algorithm to improve the samplepopulation diversity.The spider population is dynamically divided into multiple populations.Individual spider takesdifferent updating ways according to its fitness.The backtracking evolution strategy is put forward to ensure globalsearching ability and it is carried out according to evolutionary level based on the selection of global extremum of function.The simulation results show that AMBSSO algorithm has faster convergence speed and higher convergence precision,especially for high-dimensional and multimodal function optimization problems,compared with SSO,PSO and otheroptimization algorithms.

关 键 词：群居蜘蛛优化算法 多种群 多峰函数优化 自适应 回溯 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]