多阶段自适应蝙蝠-蚁群混合群智能算法  

Multi-stage Adaptive BA-ACO Hybrid Swarm Intelligence Algorithm

作　　者：陈东宁[1,2] 刘一丹[1,2] 姚成玉[3] 杨晓荣 CHEN Dongning;LIU Yidan;YAO Chengyu;YANG Xiaorong(Hebei Provincial Key Laboratory of Heavy Machinery Fluid Power Transmission and Control,Yanshan University,Qinhuangdao 066004;Key Laboratory of Advanced Forging&Stamping Technology and Science(Yanshan University),Ministry of Education of China,Qinhuangdao 066004;Key Laboratory of Industrial Computer Control Engineering of Hebei Province,Yanshan University,Qinhuangdao 066004)

机构地区：[1]燕山大学河北省重型机械流体动力传输与控制重点实验室,秦皇岛066004 [2]先进锻压成形技术与科学教育部重点实验室(燕山大学),秦皇岛066004 [3]燕山大学河北省工业计算机控制工程重点实验室,秦皇岛066004

出　　处：《机械工程学报》2021年第6期236-248,共13页Journal of Mechanical Engineering

基　　金：国家自然科学基金(51975508,51675460);中国博士后科学基金(2017M621101)资助项目。

摘　　要：针对蝙蝠算法在优化过程中未充分利用蝙蝠间搜索信息交互影响的不足,借鉴拟态物理学中的作用力规则,基于阶段性搜索策略将搜索过程分为两个阶段,分别构造符合算法阶段性搜索特点的作用力规则,提出多形态作用力蝙蝠算法,并利用Benchmark函数对所提算法与标准蝙蝠算法、变异蝙蝠算法、标准微粒群算法、两阶段微粒群算法进行性能对比测试,结果表明,所提算法具有更好的寻优能力。针对标准蚁群算法在离散空间优化时信息素更新机制单一、容易早熟收敛的不足,结合蚁群的实际社会活动提出多阶段自适应信息素机制蚁群优化算法,并在算法出现长时间停滞时,引入混沌算子使算法跳出早熟收敛,更好地发挥蚁群算法的优势,相对于标准蚁群算法、引入差分进化算法交叉变异机制的混合微粒群算法、基于动态局部搜索蚁群算法,所提算法在旅行商问题中具有更高的寻优精度、更好的稳定性。为综合不同群智能算法的优势,针对多形态作用力蝙蝠算法全局搜索能力强、收敛速度快,多阶段自适应信息素机制蚁群优化算法局部精细化能力强的特点,将两种算法串行混合,提出了多阶段自适应蝙蝠-蚁群混合群智能算法。最后,通过液压系统可靠性优化和串-并联多态系统可靠性优化实例,验证了所提混合群智能算法的有效性。A multi-form force bat algorithm(MFBA)is proposed and searching process is divided into two stages based on periodical searching strategy drawing lessons from acting force rules in physics for dealing with the deficiency that interacting information among bats is not made full use of the bat algorithm(BA)in searching process,in addition,Benchmark functions are used to compare the performance of the proposed algorithm with standard BA,variation BA,standard particle swarm optimization(PSO)algorithm and two-stage force PSO(TFPSO)algorithm,the results show that better search ability of optimal solution can be obtained by the proposed algorithm.Aiming at the defection that pheromone updating mechanism is single and is easy to fall into premature convergence in the discrete space optimization of the standard ant colony optimization(ACO)algorithm,so a multi-stage adaptive pheromone ACO(MAPACO)algorithm combining with the actual ant social activities is proposed,when the proposed algorithm appears to be stagnant for a long time,the chaotic operator is introduced to help the algorithm to jump out of the premature convergence,and to give full play to the advantages of ACO algorithm,compared with the hybrid PSO based on cross variation mechanism of differential evolution algorithm,standard ACO and ACO based on dynamic local search,the proposed algorithm is proved to have higher search accuracy and better stability in traveling salesman problem.Aiming at the two methods’advantages that the MFBA algorithm have stronger global searching ability and faster convergence rate,while the MAPACO algorithm can deal with local problems more elaborately,a multi-stage adaptive MFBA-MAPACO hybrid swarm intelligence algorithm is proposed.Finally,the proposed hybrid swarm intelligence algorithm is applied in the reliability optimization of hydraulic system and reliability optimization of series parallel multi-state system,the effectiveness of the proposed MFBA-MAPACO hybrid swarm intelligence algorithm is further verified.

关 键 词：蝙蝠算法 蚁群算法 混合群智能 可靠性优化 

分 类 号：TB114[理学—概率论与数理统计] TB18[理学—数学]