关键词: BP神经网络, 煤自燃, 温度预测, 早期预报

Abstract: As the time change ,coal in the process of oxidation temperature will desorption to produce CO, CO2, CH4, C2H6, C2H4, and other gas. For a long time ,coal spontaneous combustion prediction only considers the internal factors and external factors ,and has not yet to consider a self-heating effect on coal spontaneous combustion feedback. Based on zhao lou coal mine as the background, use gas component analysis and neural network algorithm to establish the BP neural network prediction model, select CO, CO2, CH4, C2H6, C2H4 gas concentration as neural network input layer, the coal temperature as output layer, set up eight hidden layer neurons in the forecast of coal spontaneous combustion. The result shows that after training, the predicted temperature and real temperature error control between 0 to 0.00065. The establishment of the prediction model for coal mine spontaneous combustion has an important guiding significance.