基于AHP-RBF神经网络模型的废弃矿区泥石流危险性评价研究  

Risk assessment of debris flow in abandoned mining area based on AHP-RBF neural network model

作　　者：蔡捷 周金文 黎敏刚 张燕云 CAI Jie;ZHOU Jinwen;LI Min’gang;ZHANG Yanyun(Geological Bureau of Shenzhen,Shenzhen 518023,China;Shenzhen Geology&Construction Company,Shenzhen 518040,China;Meizhou Institute for Geo-Environment Monitoring,Meizhou 514071,China)

机构地区：[1]深圳市地质局,广东深圳518023 [2]深圳地质建设工程公司,广东深圳518040 [3]梅州市地质环境监测站,广东梅州514071

出　　处：《中国矿业》2023年第S02期55-60,共6页China Mining Magazine

摘　　要：以广东省粤东地区某废弃铁矿区域三条泥石流沟为工程背景,提出基于层次分析法的径向基神经网络模型(AHP-RBF神经网络模型),基于场地岩土工程勘察成果对废弃矿区泥石流沟的危险性评价因子进行赋值及神经网络归一化训练,在此基础上,借助MATLAB平台对废弃矿区范围开展泥石流危险性研究。结果表明:AHP-RBF神经网络评价模型的计算精准度高,能够较好地把专家的经验认识以权值的方式绑定在网络节点上,成功地模拟了专家思维模式,实现了人机交互,同时避免了AHP的人为主观性对危险性评价的过度影响,使结果准确客观。Based on the engineering background of three debris flow gullies in an abandoned iron mine area in East Guangdong Province,a radial basis neural network model(AHP-RBF neural network model for short)based on analytic hierarchy process is proposed.Based on the geotechnical investigation results of the site,the risk assessment factors of debris flow gullies in abandoned mining areas are assigned and neural network normalization training is conducted.Using MATLAB platform to carry out debris flow risk research in abandoned mining area.The results show that the evaluation model of AHP-RBF neural network has high computational accuracy,and can better bind the experience knowledge of experts to the network nodes in the way of weights,successfully simulate the thinking mode of experts,realize human-computer interaction,and avoid the excessive influence of the subjectivity of AHP on the risk assessment,so that the results are accurate and objective.

关 键 词：矿山泥石流 层次分析法 神经网络 数值分析 危险性评价 

分 类 号：TD12[矿业工程—矿山地质测量]