摘要
主要影响角正切(tanβ)是开采沉陷预计的一个重要参数,对于准确界定下沉盆地边界具有重要作用。为快速精确地计算tanβ,进而有效提高开采沉陷预计精度,首先讨论了影响tanβ大小的地质采矿因素,确定了5个基本变量,即开采厚度、煤层倾角、开采深度、工作面斜长、岩性影响系数;然后详细分析了随机森林算法(Random forest,RF)的基本原理及基本实现流程;最后构建了一种计算tanβ的随机森林回归模型,用于训练和测试该回归模型的样本数据来源于国内部分主要矿区建立的典型地表位移观测站的实测资料。对训练后的回归模型采用测试样本进行检验分析,结果表明:1利用该模型计算的tanβ与实测值的最小相对误差为0.381%,最大相对误差为2.563%。2该模型具有较强的泛化能力,在计算tanβ时不仅速度快,而且具有较高的精度,对于高精度计算tanβ有一定的参考价值。
The tangent of major influence angle（ tanβ） is an important parameter for mining subsidence prediction,and it plays a key role to determine the subsidence basin boundary. In order to calculate the tanβ value quickly and accurately to improve the accuracy of mining subsidence prediction effectively. Firstly,the geological and mining factors that are affecting the value change of tanβ are discussed in detail,the five basic variables（ mining thickness,dip angle of coal seam,mining depth,working face slanting length,lithology influence coefficient） are determined; then,the basic principles and basic implementation process of random forest algorithm（ RF） are analyzed in depth; finally,the random forest regression model is established to calculate the tanβ value,the actual measured data of the typical surface movement observation stations that are located in the part of the main mining areas in China are regarded as the training and test samples of the regression model established in this paper. The test samples are used to conduct inspection and analysis of the random forest regression model after training,the results show that： 1the minimum relative error and maximum relation error between the tanβ calculated by the regression model established in this paper and the actual measured data are 0. 381% and 2. 563% respectively. 2the regression model established in this paper has stronger generalization ability,it has the characteristics of fast calculation speed and high accuracy in the calculation process of tanβ,besides that,it also has a good application prospect in the practical engineering,it can provide reference for calculating the tanβ value with high accuracy.
出处
《金属矿山》
CAS
北大核心
2016年第3期172-175,共4页
Metal Mine
基金
国家自然科学基金项目(编号:41272389)
江苏高校优势学科建设工程项目(编号:SZBF2011-6-B35)
关键词
随机森林算法
主要影响角正切
开采沉陷
下沉盆地
回归模型
Random forest algorithm
Tangent of major influence angle
Mining subsidence
Subsidence basin
Regression model
