动态信息调整且速度可控的改进型合作粒子群算法  

An improved cooperated particle swarm optimization algorithm with dynamic information adjustment and controlled speed

作　　者：付丽辉[1,2] 尹文庆[1] 

机构地区：[1]南京农业大学工学院,南京210031 [2]江苏省淮阴工学院电子与电气工程学院,江苏淮安223003

出　　处：《振动与冲击》2012年第21期120-125,共6页Journal of Vibration and Shock

基　　金：国家自然科学基金(60977038);江苏省高校重大基础研究项目(08KJA460001);江苏省普通高校研究生科研创新计划项目(CX-LX12-0277);江苏省高校自然科学基础研究面上项目(12KJD510003);淮安市科技支撑计划农业项目(SN1161)

摘　　要：针对粒子群算法中因多样性丧失引致的早熟收敛问题,提出了一种动态信息调整且速度可控的改进型合作粒子群算法。该算法通过子群划分,在粒子自身最好值、全局粒子最好值基础上,增加了子群粒子最好值对粒子飞行状态的控制作用,并利用当前寻优次数,动态调整各最好值对粒子下一次状态确定的贡献率,实现三种参考信息的有效融合,从而具有更强的寻优能力;通过子群数的调整,研究实现收敛速度控制的可能性与可行性,在保证算法搜索精度的同时,使其具有更为合适的收敛速度。最后,利用仿真实验对理论分析结果进行验证,结果表明,相对于其他PSO类算法,本算法具有更好的收敛精度,且收敛速度可控。Aiming at premature convergence problems of PSO caused by loss of diversity, an improved cooperated particle swarm optimization algorithm with dynamic information adjustment and controlled speed was proposed. On the basis of optimal values of particles and their global optimization, the optimal values of subgroups were added to control the flight states of particles dividing subgroups, the current optimization number was used to dynamically adjust the contribution rate of each best value for determination of the next state of particle, three kinds of reference information were fused effectively to seek the better ability of optimization. The possibility and feasibility of convergence speed control were realized by adjusting of the number of subgroup. The algorithm could have more appropriate convergence rate, meanwhile its convergence precision was guaranteed. Finally, the theoretical analysis results were verified with simulations, and it was shown that the new algorithm has better convergence accuracy and controlled speed compared to other PSO algorithms.

关 键 词：早熟收敛 信息共享 收敛条件 收敛速度 分组合作 粒子群算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]