煤矿火区启封后复燃预测的BP神经网络模型

Prediction Model for Reignition of Fire Zone After Unsealing Based on BP Neural Networks

煤矿井下火区启封时需要预测是否发生复燃现象,目前复燃的预测手段仅仅依赖于人工经验分析,存在智能性不足的问题,预测结果的可靠性较低.为此,结合煤自燃的内因(煤自燃倾向性)和自燃指标气体CO,CO2,O2,C2H4,C2H2,提出了一种基于BP网络的多参数火区复燃预测方法,并建立了预测模型.应用表明,该网络模型收敛速度快,预测效果理想,可以准确预测火区复燃的可能性,将其应用于复燃预测是可行的,具有较好的应用前景.

The prediction of reignition phenomenon need to be done when the fire zone is unsealed.The present prediction methods just depend on gas analyses with experience which have the problems of insufficient intelligence and low prediction reliability.Therefore,combined with the internal cause of coal spontaneous combustion,the coal spontaneous combustion tendency,and spontaneous index gases such as CO,CO2,O2,C2H4 and C2H2,a multi-parameter prediction model with reignition in fire zone based on BP neural networks was put forward and built.The application showed that this network model with a satisfactory prediction results had a fast convergence speed.The possibility of reignition in fire zone could be predicted accurately.The model was feasible to be applied to reignition prediction and had a good application prospects.