Study On Spontaneous Combustion Of Coal Based On BP Neural Network Forecasting System

Spontaneous combustion of coal seam is a kind of major disasters in coal mine production, which seriously affects the normal production of coal mine. It damages the equipment, and brings huge losses to the production of roadway and the coal resources. Coal spontaneous combustion prediction is the basis of mine fire prevention and treatment. Spontaneous combustion of coal based on BP neural network forecasting system is advanced.Spontaneous combustion of coal based on BP neural network forecasting system uses the beam tube system collects underground gas. Put the gas into chromatograph and get the data of gas concentration. Then use the BP neural network processing the data and make a judgment on whether coal spontaneous combustion.Through studying the data of Tangshan Donghuantuo coal samples oxidation experiment, choose the ratio of CH4 to CO and O2 to CO2 as the network’s input. It overcomes the shortcomings of traditional index gas method which easily effected by the draft conditions. In the case of original hardware system it greatly improves the accuracy and anti-jamming ability of forecasting coal spontaneous combustion. To simple the network output sets 0 or 1. 0 means not fire and 1 means fire. Directly establish the connection between the concentrations of index gases and whether fire. Establish the BP neural network in the MATLAB. Input the training sample to the network. After 36 times training, error reaches the requirement(<0.001). Research shows that using the BP neural network processing the data of index gas is feasible.Spontaneous combustion of coal based on BP neural network forecasting system solves the problem of modeling. It overcomes the influence of ventilation tube system defects, and improves the forecasting accuracy.