Research Of Coal And Gas Outburst Rediction Models Based On Neural Network And PSO-SVM

China is rich in coal resources, coal mining increased year by year, coal consumption has an absolute dominant position in China’s primary energy consumption, energy security has become an important basis for coal of China’s energy security. But there are still many problems in terms of security of mining, according to statistics, coal and gas outburst accident is now the main aspects of mining accidents. Security Mining not only energy security issues, but also related to all aspects of mining and mining personnel safety equipment safety. With the expanding range of coal mining, the depth and complexity are increasing, the traditional method of coal and gas outburst prediction has been unable to meet the actual needs of current and continue to more quickly and accurately to coal and gas outburst by more modern approach prediction. The main contents of this paper include the following:1, Through the study of the mechanism of coal and gas outburst, considering the prominent law occurred, the process and the associated aspects of controlling factors such as, learn from others on the basis of outstanding academic research on the coal and gas outburst clear is a complex nonlinear process. And build a more comprehensive model of coal and gas outburst from the integrated areas.2, Coal and gas outburst is a complex nonlinear geological hazards, the focus is to establish an appropriate forecasting multi-index nonlinear model. This paper uses two are nonlinear, self-learning, self-organization and parallel processing capabilities BP and RBF artificial neural networks, respectively, coal and gas outburst prediction has been modeled. In the MATLAB environment to forecast accuracy and convergence speed of the two indicators to judge the criteria for a coal mine in Shanxi measured data for simulation tests; experimental results show that, RBF neural network prediction compared with BP neural network faster and more accurate and reliable results, more practical applications in coal and gas outburst prediction.3, In the neural network to predict, based on the search for suitable broader-based algorithms and statistical methods combined with the smallest vector machine algorithm based on the evolution of particle swarm optimization algorithm, multiple targets can be achieved in a wider range of nonlinear predictive model building. The introduction of this algorithm with respect to the neural network are greatly improved in terms of parameter optimization and generalization capability. Studies have shown that, PSO-SVM algorithm has a larger application space in coal and gas outburst prediction model aspects.