神经网络技术在煤层瓦斯含量预测中的应用  

Application of neural network technology to prediction of seam gas content

作　　者：马玉姣[1] 程根银[2] 赵东云[1] 

机构地区：[1]中国矿业大学安全工程学院,江苏徐州221008 [2]清华大学环境科学与工程系,北京100084

出　　处：《煤炭工程》2010年第1期117-119,共3页Coal Engineering

基　　金：国家重点基础研究"973"项目(2005cb221504);国家"十一五"科技攻关项目(2006BAK03B01)

摘　　要：将瓦斯含量预测技术与神经网络原理结合,利用Matlab强大的神经网络工具箱,采用BP模型对钱家营矿区域瓦斯含量进行预测。根据对井田地质条件的分析研究,选取了9个反应本矿瓦斯含量的特征指标,用13个学习样本对网络进行训练,得到了影响因素与瓦斯含量之间的关系。实验表明预测结果能较好的与实际相吻合,证明神经网络技术可以用来准确预测煤层瓦斯含量。In combination with the gas content protection technology and the neural network principle, with the strong neural network tools box of the Matlab, the BP model was applied to predict the gas content of Qianjiaying Mining Area. According to the analysis and research on the geological conditions of the mine coalfield, there were 9 feature indexes reflected the mine gas contents colleted. There were 13 leaning samples applied to the training and the relationship between the influence factors and gas contents obtained. The experiments showed that the prediction results could be well fitted with the actual results and the neural network technology was approved to correctly predict the seam gas content.

关 键 词：瓦斯含量 预测 特征指标 神经网络 MATLAB 

分 类 号：TD712.3[矿业工程—矿井通风与安全]