搜索后期斥力增强型混合引斥力微粒群算法及可靠性优化应用  

LRPSO Algorithm and Applications in Reliability Optimization

作　　者：陈东宁[1,2] 姚成玉[3] 王斌[3] 张瑞星[1,2] 

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]燕山大学先进锻压成形技术与科学教育部重点实验室,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：《中国机械工程》2014年第21期2930-2936,共7页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(51405426;50905154);河北省自然科学基金资助项目(E2012203015);河北省教育厅资助科研项目(ZH2012062)

摘　　要：将利用贝叶斯网络构造的系统故障概率函数作为可靠性指标,考虑费用、质量、体积构造了资源约束函数。针对微粒群算法引斥力规则的不足,提出了搜索后期斥力增强型混合引斥力微粒群算法(LRPSO算法):在搜索前期,使微粒在其他微粒的引斥力作用下进行最优搜索,以保持种群多样性;在搜索后期,减小引力、增强斥力,利用斥力项避免微粒陷入较差位置,以提高局部搜索能力。算法测试和可靠性优化实例验证了LRPSO算法的有效性。Fault probability function constructed by Bayesian network was regarded as reliability index,and the resource constraint functions were established by considering the functions of cost,weight and volume.To overcome the shortages of attraction and repulsion rule of particle swarm optimization algorithm,a LRPSO algorithm was proposed.At the earlier-stage,each particle searched the optimum under the attraction and repulsion produced by all particles,to maintain the population diversity.At the later-stage,the effect of attraction was reduced and the effect of repulsion was enhanced using the repulsion term to avoid particles being trapped in worse searching position and improving local searching ability.The effectiveness of LRPSO algorithm was verified by algorithm tests and reliability optimization examples.

关 键 词：可靠性优化 贝叶斯网络 引斥力 LRPSO算法 

分 类 号：TB114.3[理学—概率论与数理统计] TP18[理学—数学]