基于改进BP神经网络的煤体瓦斯渗透率预测模型

Model of coal gas permeability prediction based on improved BP neural network

分析总结了煤体渗透率的3个主要影响因素——有效应力、温度和瓦斯压力,并结合煤体的力学特性建立了一个预测煤层瓦斯渗透率的BP神经网络模型。根据不同有效应力、不同温度和不同瓦斯压力条件下大量具有代表性的煤样渗透率数据来建立学习样本,并对该模型的精度进行了检验。该BP神经网络经过11 986次学习后精度满足要求,训练后BP神经网络模型所得预测结果的最大绝对误差为0.049×10-15m2,最大相对误差为4.298%。根据所建立的BP神经网络模型得到的预测值与实测值吻合较好。

Three main influential factors affecting coal seam gas permeability were analyzed and summarized in this study, which were gas pressure, temperature and effective stress. In addition, a BP neural network model of coal seam gas permeability was built based on the coal mechanical properties. A large amount of representative coal gas permea- bility data under different conditions were used for building training samples and the accuracy of the model was tested. After 11 986 times training, the BP neural network model satisfies the requirements. The results obtained by the BP neural network model show that the maximum absolute error was 0. 049x10-s m2 and the maximum relative error was 4. 298%. The results obtained by the BP neural network model match the measured data well