Study On Coal Spontaneous Combustion Forecasting Based On Grey Markov Chain Theory

In China, such spontaneous combustion happens to 56% of the important mines, seriously threat to the safety of production. Therefore, early detection of spontaneous combustion signs, and precise prediction on spontaneous combustion fire is of great significance to coal production.This paper discusses the relationship between indicator gases concentration and coal temperature, introduces common gas analysis methods, and determines the prediction method of spontaneous combustion. This paper introduces the artificial neural network prediction, support vector machine(SVM) prediction, gray prediction and markov prediction four prediction methods, contrastively analysis the advantages and disadvantages of the four prediction methods. According to the analysis result, the paper puts forward a spontaneous combustion prediction method that combining grey prediction and markov prediction. The combination of grey theory and markov chain theory, not only can improve the prediction accuracy on the basis of grey prediction, but also can expand the scope of markov prediction objects by making the forecast objects have the characteristics of markov chains’ stationary and mean through using the grey theory to fit the prediction object’s.Based on grey theory and markov chain theory, the grey markov prediction model of spontaneous combustion is constructed, and the average residual errors of every state are used to improve the prediction results. Take the CO gas detected in the wind flow near 2313 working face as examples, establishe the grey prediction model and grey markov prediction model. Matlab simulation results show that the prediction accuracy of grey markov is 1.11% higher than grey model, the prediction accuracy of revised grey markov is 2.24% higher than grey model.Research shows: the gray markov model combines the advantages of the grey model and markov model, can accurately predict system’ long-term development tendency and can play the advantage of accurate prediction for volatile data, is suitable for coal mine spontaneous combustion prediction.