概率积分法参数反演的文化-随机粒子群优化算法  

Random PSO embedded cultural framework for parameters inversion of probability-integral method

作　　者：王正帅[1,2] 邓喀中[1] 康建荣[2] 

机构地区：[1]中国矿业大学国土环境与灾害监测国家测绘局重点实验室,江苏徐州221116 [2]江苏师范大学测绘学院,江苏徐州221116

出　　处：《辽宁工程技术大学学报（自然科学版）》2013年第3期311-315,共5页Journal of Liaoning Technical University (Natural Science)

基　　金：国家自然科学基金资助项目(40772191);国土环境与灾害监测国家测绘局重点实验室开放基金资助项目(LEDM2011B10)

摘　　要：为解决传统方法在概率积分法参数反演中存在的反演过程发散问题.将粒子群优化(PSO)算法纳入到文化算法(CA)框架中,提出了概率积分法参数反演的文化-随机粒子群优化(CA-rPSO)算法.以随机粒子群优化(rPSO)算法作为信念空间的进化算法,并将PSO作为群体空间的进化算法,形成两者独立并行进化的"双演化双促进"机制,按照误差平方和最小化准则构建适应度函数,反演出概率积分法参数.研究结果表明:对于常规布设形式的地表移动观测站,基于CA-rPSO反演概率积分法参数的收敛成功率为1.该结论具有较高的实用价值,且对矿山其它复杂参数寻优问题有着一定的指导意义.To solve the divergence problem of parameters inversion in probability-integral method,a novel algorithm called random PSO embedded cultural framework（CA-rPSO） is proposed in this study through integrating PSO into the framework of cultural algorithm（CA）.In CA-rPSO,the evolving algorithms of belief space and population space are represented with random PSO and PSO respectively,forming independent and parallel ＂dual evolution-dual promotion＂ mechanism.Subsequently,selecting the least sum square of errors as inversion criterion,the fitness function is then established so as to inverse the parameters of probability-integral method.The case study results show that in a traditional surface movement observation station,the parameters inversions of probability-integral method based on CA-rPSO acheive a convergence rate of 1,which indicates that the algorithm has a high practicability.The study is of significance for other complex mining problems with parameters optimization.

关 键 词：开采沉陷 地表移动观测站 概率积分法 地表移动参数 参数反演 粒子群优化 文化算法 智能优化 

分 类 号：TD325[矿业工程—矿井建设]