智能优化算法在概率积分参数反演中的比较  

Comparation of Intelligent Optimization Algorithms in the Inversion of Probability Integral Parameters

作　　者：陈元非 高彦涛[3] 査剑锋 徐孟强[1,2] 张正华[1,2] 李晋龙[1,2] 

机构地区：[1]中国矿业大学环境与测绘学院,江苏徐州221116 [2]国土环境与灾害监测国家测绘地理信息局重点实验室,江苏徐州221116 [3]河南省地矿局测绘地理信息院,河南郑州450006

出　　处：《金属矿山》2017年第4期162-168,共7页Metal Mine

摘　　要：概率积分法是我国矿山开采沉陷预计的主要方法,其预计的精度直接取决于参数准确性。采用智能优化算法对实测地表沉陷数据反演是获取概率积分法参数的主要方法。为研究优化算法在开采沉陷概率积分参数反演中的应用效果,采用VB语言编程实现了模矢法、遗传算法、粒子群算法、模拟退火算法等常见概率积分参数反演算法,通过构造理论数据分析和比较了这4种算法参数反演的效果,并从运行时间、求参稳定性、搜索性能、抗局部解能力等方面对4种算法进行综合评价。研究结果表明:4种算法参数反演结果精度较高,参数相对误差小于2%,且对观测站中的观测值随机误差、粗差问题具有较强的抗干扰能力。模矢法运行效率高但容易陷入局部解,粒子群算法效率较低,遗传算法和退火算法全局能力强但后期收敛能力较弱。Probability integral method is the main method of mining subsidence prediction in China,the accuracy of the estimated precision is directly determined by the results of the parameter inversion accuracy. Intelligent optimization algorithms are the main methods to obtain the probability integral parameters. In order to study the application effect of the optimization algorithm in the inversion of mining subsidence probability integral parameters,based on the VB programming language compiler to compile the vector method,genetic algorithm,particle swarm optimization algorithm,simulated annealing algorithm of the probability integral parameter inversion procedure. The parameters inversion effects of the above four algorithms is analyzed and compared with the theoretical data,and the comprehensive evaluation of the four algorithms is done from the aspects of running time,stability of parameter seeking,search performance and the ability of anti local solutions. The results show that the four algorithms have higher precision,the relative error of the parameters is less than 2%,and has strong anti-interference ability when there are random errors existed in the observation station and the gross error. The running efficient of vector method is high but easy to fall into local solution,the running efficiency of particle swarm algorithm is low,the global capability of genetic algorithm and simulated annealing algorithm is relatively strong,but the late convergence ability is weak.

关 键 词：开采沉陷预计 概率积分法 参数反演 智能优化算法比较 

分 类 号：TD172[矿业工程—矿山地质测量]