Regional Prediction Of Coal And Gas Outbursts Based On Geological Complexity Using SVM Model

Coal is the most important and basic energy of our country, and China is also the largest coal-producing country in the world. Corresponding to this,China is the worst affected country which has the largest number of coal mine disasters. Coal and gas outburst is the most serious of all disasters in coal mines. According to the requirements of " provisions for prevention and cure of coal and gas outburst ", coal and gas outburst prediction need to be done in the process of coal mine production. The factors which effect the coal and gas outburst are numerous and complex in the mechanisms. The Support Vector Machine as a new intelligent classification technique in solving nonlinear problems with small samples has a very good affection. To this end, we propose the use of Support Vector Machine classification model to achieve regional prediction of coal and gas outbursts.It analyses many factors that influence coal and gas outbursts in this paper. Moreover, the geological structure is one of the most important controlling factors. People have a preliminary understanding on the relationship between geological structure so as coal and gas outburst. However, the quantitative relationship between them is still being explored. It has been established a quantitative description method which is about basic geological structure index of fault, fold and angle through researching on quantitative evaluation of the structural complexity.Considering the Support Vector Machine modeling, this paper selects seven characteristic indexes, so as to establish a model of regional prediction of coal and gas outbursts. The best accuracy rate is the ultimate goal in the model training process. When determining the optimal kernel function, it chooses RBF kernel function for the model through continuous testing. Next, the parameters of C and g are optimized by the cross validation method. Based on the views of safety, we accept the practical value of the prediction model eventually.In this paper, using the model has set up regional prediction for a coal mine. Analyzed for the dangerous area, it chose the program of a hybrid hole arrangement for seam gas drainage. For the low permeability of coal mine, it used perforating pre-splitting technology to increase coal seam permeability which has been proved a good result of this method according to analyzing the comparison on the experimental data.