基于MARS的地表下沉系数求取方法研究

打开文本图片集

中图分类号：TP39；TD325.2 文献标识码：A 文章编号：2096-4706（2025）16-0159-05

Research on the Calculation Method of Surface Subsidence Coefficient Based on MARS

YU Dan （Guangzhou Natural Resources Surveying and Mapping Co.，Ltd.， Gunagzhou 51oo70， China）

Abstract： In order to solve the problem that the surface subsidence coeficient is dificult to obtain accurately，the Multivariate Adaptive Regression Spline （MARS）is introduced to establishthe prediction model.Byanalyzing the factors afectingthesurfacesubsidencecoeffcientandverifying itbasedonthe measureddataof66typicalworkingfacesinChina，the relationshipbetweenminingfactorsndsurfacesubsidencecoeffcientisdisussedndtheresultsarecomparedwithtoseof BPNeuralNetwork model.Theresultsshowthatthe MARSmodelhashigh predictionaccuracy，the maximumrelativeerror is 3.76% ，and the average relative error is 1.24% .The thickness of loose layer，average mining depth and width-depth ratio of strikeareidentifedas themaininfluencingfactors.The predictionresultsofthemodelcanprovidereferenceforthepredictionof surface subsidence coefficient parameters.

Keywords： mining subsidence; MARS; surface subsidence coeficient; regression model

0 引言

煤炭资源作为我国重要的能源燃料，在很长一段时间内将难以被替代[]。（剩余5123字）