粒子群优化算法在多参数拟合中的应用  

Application of particle swarm optimization to multiparameters fitting

作　　者：杨朝霞[1] 方健文[1] 李佳蓉[1] 曾胜财[1] 

机构地区：[1]浙江师范大学数理与信息工程学院,浙江金华321004

出　　处：《浙江师范大学学报（自然科学版）》2008年第2期173-177,共5页Journal of Zhejiang Normal University:Natural Sciences

基　　金：国家自然科学基金资助项目(60477039);浙江省自然科学基金资助项目(Y104574)

摘　　要：提出一种新的自适应粒子群优化算法,以解决梯度法为基础的算法在进行多参数拟合时因各参数之间相关性较高而带来的拟合上的问题.该粒子群优化算法采用自适应变异和动态自适应调整搜索范围、惯性权重相结合的改进策略,数值模拟了将该算法应用于测量薄膜热物性时的多参数拟合,结果表明该算法是可行和有效的.In order to make up the deficit of fitting strong correlation parameters with gradient-based methods, a new adaptive particle swarm algorithm was proposed. A modified strategy was developed by combining a new adaptive mutation and an adaptive adjustment inertia weight, searching regions in the algorithm. The new algo- rithm was used to simulate numerically the multiparameter fitting in the process of characterizing thermal properties of thin films. It was showed that the new algorithm was feasible and efficient

关 键 词：粒子群 多参数拟合 相关性 惯性权重 自适应变异 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]