基于GM(0,N)模型的煤自然发火期预测  

Forecasting study on spontaneous combustion period of coal based on GM(0,N) model

作　　者：马砺[1,2] 雷昌奎 张志鹏[1,2] 袁婷[1,2] 

机构地区：[1]西安科技大学西部矿井开采及灾害防治教育部重点实验室,陕西西安710054 [2]西安科技大学能源学院,陕西西安710054

出　　处：《辽宁工程技术大学学报（自然科学版）》2016年第9期902-907,共6页Journal of Liaoning Technical University (Natural Science)

基　　金：国家自然基金青年项目(51204135)

摘　　要：为科学准确预测煤自然发火期,运用灰色系统理论,基于灰色关联分析,选取煤样工业分析中的灰分、挥发分和元素分析中的C、H、O、S含量作为系统相关因素,建立了预测煤最短自然发火期的GM(0,7)模型,经后验差检验,模型精度为优;通过与多元线性回归模型预测结果对比,GM(0,7)模型预测煤自然发火期的平均相对误差为2%,多元线性回归模型预测的相对误差为10.35%.经外来数据回代检验,GM(0,7)模型预测结果的相对误差在2%左右,多元线性回归模型预测结果相对误差达26.27%,说明GM(0,7)模型预测结果优于多元线性回归模型.研究结果表明:利用灰色关联分析选取适当参数建立GM(0,N)模型能够较好预测煤最短自然发火期.In order to forecast spontaneous combustion period of coal scientifically and accurately, this paper used grey system theory, selected the ash, volatile of proximate analysis and C, H, O, S concentration of elemental analysis of coal samples to be correlative factors of system based on the grey correlation analysis, and built the GM（0,7） model to forecast the shortest spontaneous combustion period of coal. After-test residue checking shows that the model precision is excellent; compared with multiple linear regression model, the average relative error of the spontaneous combustion period of coal that GM（0,7） model forecasted is 2 %, and it is 10.35 % forecasted with multiple linear regression model. The back-substitution check on external data shows that the relative error of GM（0,7） model is about 2 %, and it is 26.27 % forecasted with multiple linear regression model, which demonstrates the results that GM（0,7） model forecasted is better than multiple linear regression model. The study results show that the GM（O,N） model with appropriate parameters selected based on grey correlation analysis can be used to perform better forecasting of the shortest spontaneous combustion period of coal.

关 键 词：煤 自燃 自然发火期 灰色理论 灰色关联度 GM(0 N)模型 多元线性回归 

分 类 号：TD75[矿业工程—矿井通风与安全]