Study on the prediction of gas content based on grey relational analysis and BP Neural Network

Abstract

The paper mainly illustrated a kind of BP Neural Network model in MATLAB Neural Network Toolbox to predict gas content of the coal seam based on analyzing grey relational degree. This model constructed a method of gas content prediction by choosing four dominate effect factors (Coal seam buried depth, Geologic structure, Roof lithology, Coal seam thickness) as the input parameters. It has been established for training and testing, and forecasting the gas content of coal seam by using the learning samples which were collected from the instances of typical exploited borehole data of Panyi East coal mine in Huainan coal mining area. The results show that the model is an efficient prediction method for gas content, and its prediction accuracy and feasibility are better than the traditional predicting methods. It also can better meet the practical requirements of safety production in coal mine including provide some references for mine gas disaster prevention. © Springer-Verlag Berlin Heidelberg 2013.