Prediction Method Study On Water Inrush Through Coal Floor Based On Support Vector Machines

Northern China coal field is the main coal production area in our country. Badly threatened by Ordovician karst water, the mine production capacity and the service years of the mine are always difficult to achieve the regular level for a long time. For continuous and steady development of the coal mine industry, it's the only way to mine the coal resources threatened by groundwater hazards. To realize safe mining it's important to study the prediction method of water inrush through coal floor (WITCF).Because the prediction of WITCF is a complicated, nonlinear and high-dimensional problem, and it's also synthetically affected by multi-factors, it's often difficult for traditional methods to solve. It's an active research direction for the coal hydrogeologist to search an effective method for prediction of WITCF.A new way to solve mining groundwater hazards is put forward in this thesis, hydrogeological characteristics and water inrush affection factors of Northern China coal field were analyzed and the latest Statistics Learning Theory——Support Vector Machines(SVM) was used, a prediction method of WITCF based on the SVM is studied and put forward.The main contributions of this thesis are as follows: the main groundwater hazard types and affection factors of WITCF were analyzed, the training and predicting processes of SVM from linear SVM to nonlinear SVM were inferred detailedly, and training algorithm was summed up, the prediction model for WITCF based on SVM was presented for the first time, the algorithm was implemented with MATLAB programming, factors as minimum combination for prediction were extracted from water inrush affection factors set by feature selection, grid search and 10-folder cross validation were adopted to search the best parameters of SVM, classifier for prediction of WITCF was determined, the results show that the method of water inrush through coal floor based on support vector machines has effectively increased the forecasting precision compared with the method based on water inrush coefficient, and then the technology approaches of applying the method of water inrush through coal floor based on SVM to the mining under safe water pressure of aquifer were discussed, Finally, the theoretical and technical problems to be studied and to be solved were put forward.