基于指标气体的煤自燃预报人工神经网络专家系统研究  

Study on Artificial Neural Network Expert System for Forecasting Spontaneous Combustion of Coal Based on Gas Indices

作　　者：张辛亥[1] 孙久政[1,2] 陈晓坤[1] 程方明[1] 周金生[2] 李庆军[2] 

机构地区：[1]西安科技大学能源学院,陕西西安710054 [2]双鸭山矿业集团,黑龙江双鸭山155100

出　　处：《煤矿安全》2010年第3期10-12,共3页Safety in Coal Mines

基　　金："十一五"国家科技支撑计划资助项目(2007BAK29B03)

摘　　要：煤层自燃温度与气体产物之间存在对应关系,据此可预报煤层自燃。但是,温度与气体产物之间的关系为非线性的,非常复杂,不能用简单的数学公式表达。用前向3层人工神经网络模型表征这一对应关系,并用大量实验数据对网络进行训练以获得神经元间的连接强度。据此建立了人工神经网络专家系统,可以根据现场监测的煤自燃指标气体等参数,判断煤自燃现状,并预测自燃发展趋势。实践表明,该神经网络专家系统能够较准确地预报煤层自燃。Spontaneous combustion of coal can be forecasted based on corresponding relation between coal temperature and its gaseous products＇ concentration.Nevertheless,the corresponding relation between gaseous products and temperature is non-linear and very complicated,and can＇t be depicted with simple mathematical formula.A forward feeding 3-layer artificial neural network model is employed to token the corresponding relation,and a large amount of experimental data were used to train the network to gain the connection strength between nerve cells.An expert system to determine proceeding and predict tendency of spontaneous combustion of coal based on gas indices was set up based on the artificial neural network.It proved in real coal production that the artificial neural network expert system can forecast coal spontaneous combustion precisely.

关 键 词：煤自燃 预报 人工神经网络 专家系统 

分 类 号：TD752.2[矿业工程—矿井通风与安全]