基于Matlab的概率积分法预计参数反演方法比较

Comparative Analysis of Methods Used in Parameter Inversion of Probability Integral Method

预计参数反演是利用概率积分法预计煤矿开采后地表移动与变形的关键。总结分析了预计参数反演的最小二乘法、模式法和遗传算法,利用Matlab软件中的最优化算法工具箱,在缓倾斜煤层非充分开采条件下,对3种反演方法的计算结果进行比较。结果表明,最小二乘法计算结果准确性较好,但初始点选择不当会导致计算失败;模式法和遗传算法都无需计算函数的梯度信息,遗传算法的初值是种群,其计算结果比模式法更加准确;在各参数中,开采影响传播角θ的误差较大,通过采用遗传算法工具箱和fminbnd函数联合走向断面和倾向断面的数据进行反演,可以提高预计参数的准确性。

The predicted parameter inversion is one of the key issues in the application of predicting the surface movement and deformation after coal mining by using the probability integral method. The paper summarizes and analyzes the least square method, pattern search and genetic algorithm of predicted parameter inversion, and compares the calculation results of three inversion methods under the condition of insufficient mining in gently inclined coal seam by using the optimization algorithm toolbox of Matlab. The results show that the accuracy of the least square method is better than the others, but the unsuitable initial point may lead to divergence problem;neither the pattern search nor the genetic algorithm needs to compute the gradient information of functions, moreover, the initial value of the genetic algorithm is a population, and its calculation result is more accurate than that of the pattern search;the error of propagation angle "θ" is comparatively large, so the accuracy of the predicted parameter can be obviously improved by using the genetic algorithm toolbox and fminbnd function in Matlab to combine the data of strike section and inclined section.