文献类型：期刊文献

中文题名：基于改进麻雀搜索算法优化SVR和FLAC3D的开采地表沉降预测

英文题名：Prediction of Mining Surface Subsidence by Optimized SVR Based on Improved Sparrow Search Algorithm and FLAC3D

作者：黄鑫康 刘萍 芦庆和 高方玲 陈镇 罗畅

第一作者：黄鑫康

机构：[1]贵州大学矿业学院,贵州贵阳550025;[2]贵州理工学院矿业工程学院,贵州贵阳550003

第一机构：贵州大学矿业学院,贵州贵阳550025

年份：2023

卷号：43

期号：10

起止页码：110-118

中文期刊名：矿业研究与开发

外文期刊名：Mining Research and Development

收录：CSTPCD;;Scopus;北大核心:【北大核心2020】；

基金：贵州省科技计划资助项目(黔科合基础1Z047)。

语种：中文

中文关键词：最大下沉量;探索性因子分析;切线飞行机制;支持向量回归机;FLAC3D

外文关键词：Maximum subsidence;Exploratory factor analysis;Tangent flight mechanism;Support vector regression machine;FLAC3D

摘要：准确地预测矿山地下开采造成的地表沉降一直是矿山安全发展的难题。为了能够提高采空区地表最大下沉量预测的精度,分别利用机器学习和数值模拟的方法预测某磷矿的采空区地表沉陷,并进行对比研究。综合考虑地表最大下沉量的影响因素,参照有关文献对国内矿山相关数据的统计成果,利用探索性因子分析(EFA)实现影响因素降维,用t分布和切线飞行机制改进的麻雀搜索算法(tSSA)优化支持向量回归机(SVR),从而构建采空区地表沉降EFA-tSSA-SVR预测模型,并用该模型和FLAC3D数值模拟模型对某磷矿地下开采所造成的地表沉降进行预测。结果表明,两种模型对一步骤盘区开采后的最大下沉量预测结果相近,与一步骤开采后的实际下沉量做对比,FLAC3D数值模型预测的地表最大下沉量的相对误差是EFA-tSSA-SVR模型的3倍,表明EFA-tSSA-SVR模型预测结果较好。用EFA-tSSA-SVR模型预测该磷矿未来的二、三、四步骤盘区开采后的地表最大下沉量,预测结果分别是为22.3 cm、24.1 cm、25.9 cm。
Accurate prediction of surface subsidence caused by underground mining has always been a difficult problem in the development of mine safety.In order to improve the prediction accuracy of the maximum surface subsidence in the goaf,machine learning and numerical simulation methods were used to predict the surface subsidence in the goaf of a phosphate mine,and comparative studies were conducted.Taking the influencing factors of the maximum surface subsidence into account,referring to the statistical results of the literature on the relevant data of domestic mines,the exploratory factor analysis(EFA)was used to reduce the dimension of the influencing factors,and the sparrow search algorithm(tsSA)improved by t distribution and tangent flight mechanism was used to optimize the support vector regression machine(SVR),so as to construct the EFA-tSSA-SVR prediction model of surface subsidence in the goaf.The model and FLAC3D numerical simulation model were used to predict the surface subsidence caused by underground mining of a phosphate mine.The results show that the two models have similar prediction results for the maximum subsidence after one-step panel mining.Compared with the actual subsidence after one-step mining,the relative error of the maximum surface subsidence predicted by the FLAC3D numerical model is 3 times that of the EFA-tSSA-SVR model,indicating that the EFA-tSSA-SVR model has better prediction results.The EFA-tSSA-SVR model is used to predict the maximum surface subsidence after the second,third and fourth step panel mining of the phosphate rock in the future.The predicted results are 22.3 cm,24.1 cm and 25.9 cm respectively.