Application of Support Vector Machine in establishing the quantitative relationship of mining subsidence and the coupling of joints with the tectonic stress

In order to obtain the features of coal mining subsidence under complicated geological conditions, and to improve the forecasting accuracy of mining subsidence, software FLAC3D is adopted to simulate excavating 55 models in which tectonic stress and joints is taken into account. The effect of the coupling of joints with the tectonic stress on mining subsidence is investigated. And then the quantitative relationship between mining subsidence and the coupling of joint with the tectonic stress is established by using Support Vector Machine (short for SVM). The results show that: SVM has a higher accuracy and faster calculating speed in establishing the quantitative relationship compared with using BP artificial neural network method and multiple linear regression analysis method. SVM should be adopted in searching the quantitative relationship between coal mining subsidence and geological factors.