Forecast of small fault based on SVM in Zhaoguan mine

Because of its fitness for small sample modeling, a model based on SVM was established. The model used for forecasting the horizontal extending length is combined with the actual measured data of 7# coal seam. The tendency, throw and dip angle of fault was the impact actors of the forecasting model. The software package called libsvm based on Matlab was the platform of the model. Taking advantage of the model, three samples in Zhaoguan mine and three samples in Qiuji mine were perfectly predicted. Compared with the result from multiple regression analysis,SVM has a better outcome when the sample is small.