The Effective Application of BP Neural Networks Prediction Model for Gas Content in Binchang Mining

Abstract

In order to predict gas content of coal seam accurately in binchang mining, we use core data to build the BP neural network. We select the important controlling factors which impacted gas content of coal seam, coal bed thickness, ash and max vitrinite reflectance as the basic features of the BP neural network model, and establish the BP neural network prediction model between coal bed methane content and the main controlling factors. The testing results show that the BP neural network model could truly reflect the non-linear relationship between the gas content and the controlling factors, and obtain minimal error between the predicted results and the measured ones. This method provides the probability for using geological, logging and seismic information to predict gas content of coal seam.