相关性的组合预测模型在开采沉陷中的应用  

A combined model for mining subsidence forecasting based on correlativity

作　　者：王金涛 郭广礼 郭庆彪 赵自强[2] 徐孟强 郭凯凯 

机构地区：[1]中国矿业大学环境与测绘学院/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 [2]上海岩土工程勘察设计研究院有限公司,上海200000

出　　处：《测绘科学》2017年第10期181-185,共5页Science of Surveying and Mapping

基　　金：国家"十二五"科技支撑计划项目(2012BAB13B03);江苏省高校优势学科建设工程资助项目(SZBF2011-6-B35)

摘　　要：针对当前开采沉陷不同阶段的单点沉陷变形预计精度不高的问题,该文首次从基于相关性的组合预测模型的思想出发,建立了面向沉陷不同阶段的基于相关系数的组合预测模型。实例验证结果表明,基于相关系数的组合预测模型在开采沉陷预测中的应用克服了单一模型的局限性,能够有效利用不同单一模型的信息,有较高的精度和可信度;能够满足不同阶段的开采沉陷预测,更适用于对沉陷变形数据要求高的场合。Aiming at the low forecasting accuracy of the point subsidence at different stages of coal seam mining, this paper first from the angle of the combination forecast model and the combination forecasts model based on correlativity was established according to different instances of mining subsidence. Experimental results showed that the proposed model could overcome the limitations of single model, which is able to effectively use different information of a single model and have higher accuracy and reliabil- ity; it could satisfy the different stages of the mining subsidence and is more suitable for the occasions with higher requirements on subsidence deformation data.

关 键 词：开采沉陷 相关系数 ARMA模型 BP神经网络 组合预测 

分 类 号：P208[天文地球—地图制图学与地理信息工程]