Research On Gas Prediction Of Thin Coal Seam In Qinglong Mine Based On Machine Learning

At present,there are relatively few researches on gas prediction in thin coal seams.The regulations require that gas outburst hazard evaluation should be carried out for thin coal seams with a thickness of 0.3m and above.However,there are very few measured data of gas in thin coal seams.Therefore,the gas properties of thin coal seams are carried out.Forecast research is very necessary.The coal seam of Qinglong Mine has high gas content and thin coal seams are very developed.The distribution of gas has regional differences,which affects the safety of coal mine production.Therefore,it is particularly necessary to study the characteristics of gas occurrence in the thin coal seams of Qinglong Mine.This article takes Qinglong Mine 16~#(thick)coal seam and 2~#(thin)coal seam as the research object.Through the collection of relevant data and data statistics,mercury intrusion experiments on coal samples collected underground,and micro-joints of structural coal samples are carried out.Observation,researched the basic physical characteristics of coal seams in the minefield;discussed the main geological factors affecting gas distribution;because thin coal seam gas is not easy to measure,the coal seam depth,coal seam thickness,structural curvature value,and apparent resistivity value of the 16~#coal seam are used As well as gas content and gas pressure as training set samples,the BP neural network model is established by Python programming and the Tensor Flow machine learning library is used to predict the gas content and gas pressure of the 2~#thin coal seam.The gas content and gas pressure data of 2~#thin coal seam predicted by neural network are clustered and analyzed by K-means clustering algorithm,and then the gas high value area of 2~#thin coal seam is predicted.The coal seams in the study area are dominated by micropores and mesopores.The phase mercury intake curve of coal samples has a"double peak"phenomenon,that is,higher peaks appear in the two pore sizes less than 10nm and nearly 1000nm.The dominant orientations of coal and rock fractures are the NWW and NNW directions,indicating that the study area has experienced the compression of NW-SE structural principal stress,which has caused the development of coal and rock fractures and destroyed the integrity.The gas distribution in the 16~#coal seam in the study area presents the characteristics of“high in the southwest and low in the northeast”.The main factors affecting the occurrence of gas are the depth of the coal seam,the thickness of the coal seam,the geological structure,and the degree of coal body fragmentation.The study found that all influencing factors are related to the gas content And gas pressure show a linear relationship.The distribution of gas content and gas pressure in the 2~#thin coal seam predicted by neural network shows the characteristics of"low in southwest high and middle part".According to the analysis of K-means clustering algorithm,the occurrence of gas in the 2~#thin coal seam is mainly divided into four categories.Among them,the average value of gas content and gas pressure in the second category is the highest,which are 17.39cm3/g and 2.34MPa,respectively,which are mainly distributed The southwest and southern borders of the mine.