摘要
为了探测顶板岩层组合状况从而更好地满足锚杆支护设计要求,采用人工神经网络方法研究了煤巷顶板岩层岩性识别的实际问题,并根据打钻过程的特点,采用BP网络研究了人工神经网络的结构和输出方式,分析了影响人工神经网络应用效果的各因素,在人工神经网络的优化设计方面作了较深入的研究.研究表明:人工神经网络用于识别所钻岩层误差仅有5%,达到了很好的效果;人工神经网络的参数,如学习率、训练步长、动量系数、隐含层单元数和数据处理方式等对于人工神经网络的应用效果有很大影响.
In order to detect the combining situation of roof strata so as to better meet the requirement of designing the bolt support parameters, the artificial neural network （ANN） method was used to solve the problem of identifying the strata lithology of roadway roof strata. According to the drilling features, the structure and output form of the BP ANN were designed, various factors affecting the application of the ANN were analyzed, and a further research into the optimum design of the ANN was carried out＇ The result shows that the ANN matches the identification of the strata perfectly with an error of only 5%. Various ANN parameters such as the training efficiency, learning factor, momentum, the hidden-layer unit number and the method of data processing may affect the application of the ANN.
出处
《采矿与安全工程学报》
EI
北大核心
2006年第2期182-186,共5页
Journal of Mining & Safety Engineering
关键词
人工神经网络
岩性识别
煤巷顶板
artificial neural network
identification of lithology
mining roadway roof
