基于神经网络与灰色理论的工程岩体分级  

Engineering Rock Mass Classification Based on ANN and Grey Models

作　　者：张兆省 来光 厉从实 聂胜立 皇甫泽华 ZHANG Zhaosheng;LAI Guang;LI Congshi;NIE Shengli;HUANGFU Zehua(Qianping Reservoir Construction Administration Bureau of Henan Province,Zhengzhou 450003,China;Henan Water Conservancy Survey Co.,Ltd.,Zhengzhou 450008,China)

机构地区：[1]河南省前坪水库建设管理局,河南郑州450003 [2]河南省水利勘测有限公司,河南郑州450008

出　　处：《人民黄河》2019年第1期93-96,共4页Yellow River

基　　金：河南省水利科技攻关计划项目(GG201652)

摘　　要：建立岩体分级结果与岩石强度、岩体完整度、地下水分布等影响因素间的非线性映射关系,对大型水利水电工程的岩体质量分级工作具有重要意义。以前坪水库坝址区工程岩体为例,采用灰色理论对影响因素及对应结果进行聚类划分,构建灰色理论岩体质量分级体系;以类似工程岩体数据作为输入样本对BP神经网络进行训练,拟合各影响因素与分级结果之间的函数关系,并构造特定网络,最后将研究区岩体数据作为检验样本进行分级。与比传统工程岩体质量分级方法比较表明:新的模型能最大限度利用勘察数据库,且分级结果与传统方法基本一致,少数岩组偏向于经济性。Establishing the relationshipamong the results and factors such as rock strength,rock mass integrity and groundwater distribution is of great significance to rock mass classification in large-scale Hydropower projects. The engineering rock mass of Qianping Reservoir was taken as an example,the grey models was used to cluster the influence factors and the corresponding results and the rock mass quality classification system was established. The BP neural network was trained by using similar engineering rock mass data as input samples,fitting the functional relationship between the influence factors and the results and constructing a specific network. Finally,the rock mass data in the study area were taken as test samples for classification. Compared with the traditional engineering rock mass classification method,the results show that the new model can make the maximum use of survey database and the results of classification are basically the same as those of traditional methods. A few groups are biased towards the more economical. It can be a reference for quality classification of engineering rock mass.

关 键 词：岩体分级 灰色理论 BP神经网络 不确定性分析 

分 类 号：TV221[水利工程—水工结构工程]