具有启发式探测及自学习特征的降维对称微粒群算法.pdfVIP

37 5 V o l. 37 No. 5 20 10 5 Computer Science May 2010 ( 山东师范大学信息科学与工程学院 济南250014 提出对称微粒群算法Sy mPSO_ H D, 用以提高PSO 算法的搜索能力引入种群分布熵以保证种群的分布 性; 引入具有探测特 的启发式粒子, 用以影响普通粒子的位置; 提出邻域内的克隆变异选择策略及全局范围内的降 维对称粒子策略, 用以增强粒子的局部及全局学习能力仿真实验及分析结果表明, SymP SO_ H D 算法搜索能力稳 定, 适应性强, 能以较大概率收敛到全局最优 降维对称微粒群算法, 种群分布熵, 启发式探测, 克隆变异 T P 18 A Dimensionality Reduction Symmetrical PSO Algorithm Characterized by Heuristic Detection and Selflearning SH A O Z engzhen WA N G H ongg uo L IU Ho ng ( In st itu t e of In format ion Science and Engineering , S handong N orm al U ni versit y , Jin an 2500 14, China Abstract A no vel Sy mmetrical PSO alg or ithm ( SymP SOH D w as pr oposed to impro ve t he sear ch ability of PSO algo r ithm . T o init ialize the clu ster eff ectiv ely , po pulat ion scatter entro py str ategy w as int roduced . In o rder to improv e the par ticle s position vector , a special kind of par ticle characterized by detectio n w as proposed to o. And to enhance the par t icles learning ability bo th in local and global doma in, w e put for w a rd the clo nemutatio nselection str at eg y in neig h bor hood and the dimensionalit y reduction sy mmet rical st rateg y in global area. Simulatio n and analysis sho w that SymP SOHD a lg or ithm has stable search ability and strong adaptability , and can co nv erg e to the g lobal optimum w ith lar ge pr obability. Keywords Dimensio nality reductio n symmetr ical P SO , P opulatio n scatter entro py , H eur istic detectio n, Clo ne and muta t ion PSO 1995 Eberhar t[ 1, 2] , Sym PSO_ HD, , : , P SO ; , [ 3]