基于RBF神经网络的瓦斯含量预测研究  

Research on prediction of gas contents base on RBF nerves network

作　　者：吴观茂[1] 黄明[1] 李刚[1] 郭相坤[2] 

机构地区：[1]中国矿业大学(北京)煤炭资源与安全开采国家重点实验室 [2]中国矿业大学(北京)化学与环境工程学院,北京100083

出　　处：《煤炭科学技术》2008年第1期49-52,共4页Coal Science and Technology

基　　金：国家自然科学基金重大项目(50490271);“十五”国家科技攻关重点项目(2001BA803B0403)

摘　　要：以淮南矿区潘一矿13-1煤层为研究对象,确定了煤层埋深、煤层厚度、顶板岩性和构造是影响煤层瓦斯含量的主要因素;在分析勘探钻孔资料的基础上,利用RBF神经网络方法建立了瓦斯含量预测模型,结合实际数据,对预测模型进行训练和检验,预测结果表明,该模型比使用线性回归和BP神经网络模型预测能获得更高的精度,说明预测模型可靠。Taking the 13-1 seam in Panyi Mine of Huainan Mining Area as the research object, it is settled that the main factors affected to the seam gas content were the seam depth, seam thickness as well as the roof lithological character and tectonics. Based on the analysis on the exploration borehole information, a prediction model of the gas content was established with the RBF nerves network method. In combined with the actual data, a training and inspection was made with the prediction model. The predicted results show that the prediction model could have an accuracy higher than the liner regressive and BP nerves network model, which could show the reliability of the prediction model.

关 键 词：瓦斯含量 RBF神经网络 预测模型 

分 类 号：TD712[矿业工程—矿井通风与安全] TP183[自动化与计算机技术—控制理论与控制工程]