Research on Chaotic Forecasting Method for Gas Emission from Working Face Based on SVM Theory

Aimed at the deficiencies in traditional linear prediction method for gas emission,a chaotic time series prediction model for gas emission based on support vector machine(SVM) theory is built by using the phase space delay coordinates reconstruction theory of chaotic dynamical systems with consideration on the inherent determinacy and nonlinear nature of the time series of gas emission. After a simulation calculation to the time series of gas emission inⅡ1024 working face,the results show that the prediction model possesses higher generalization ability than the traditional regression methods,and has high prediction accuracy. Moreover,the model has the merits of easily being built and high training-speed,etc.. Due to a full consideration on the chaos of time series of gas emission and a good utilization of the excellent prediction features of SVM,the prediction turns to be more scientific.