基于进化神经网络模型的面板堆石坝沉降和面板挠度预测  

Settlement and Slab Deflection Prediction of Concrete Face Rockfill Dam Based on Evolutionary Neural Network Model

作　　者：赵新瑞[1] 吕晓曼 黄耀英[1] 左全裕 刘钰[1] ZHAO Xinrui LU Xiaoman HUANG Yaoying ZUO Quanyu LIU Yu(College of Hydraulic ＆ Environmental Engineering, China Three Gorges University, Yichang 443002, Hubei, China Tianjin Standard Letter Detection Technology Development Co., Ltd., Tianjin 300000, China Hunan Centian River Construction Investment Co., Ltd., Yongzhou 425500, Hunan, China)

机构地区：[1]三峡大学水利与环境学院,湖北宜昌443002 [2]天津标信检测技术发展有限公司,天津300000 [3]湖南涔天河工程建设投资有限责任公司,湖南永州425500

出　　处：《水力发电》2017年第3期68-71,105,共5页Water Power

基　　金：国家自然科学基金资助项目(51209124)

摘　　要：堆石坝变形和面板挠度的预估对指导堆石坝设计和施工具有重要意义。针对堆石坝室内试验参数和实际参数存在差异,而待建堆石坝缺乏实测变形进行参数反馈,难以较准确预估堆石坝沉降和面板挠度,为此较广泛搜集了类似面板堆石坝工程的监测数据,将遗传算法和神经网络模型相结合,建立了堆石坝工程变形预测的进化神经网络模型。由待建面板堆石坝工程的坝高、宽高比和干密度等作为控制参数,结合训练好的进化神经网络模型,预测得到待建面板堆石坝的变形及面板挠度。实例分析表明,该方法可行。The estimation of the deformation of rockfill dam and slab deflection has vital significance to instruct the design and construction of rockfill dam. In view of the differences of rockfill dam＇s parameter between indoor test and practical engineering and the rockfill dam that preparing for construction lack of measured deformation data to parameter feedback, the monitoring data of similar constructed concrete face rockfill dam are widely collected, and then the genetic algorithm and neural network model are combined to establish evolutionary neural network model for deformation prediction of rockfill dam engineering.Taking dam height, aspect ratio and dry density of concrete face rockfill dam that preparing for construction as control parameters, the deformation and slab deflection can be obtained through the evolutionary neural network model which being trained. The example analysis shows that the method is feasible.

关 键 词：遗传算法 进化神经网络模型 沉降变形 面板挠度 预测 

分 类 号：TV641.43[水利工程—水利水电工程]