槐树关水库大坝渗流安全评价与预测  

Safety Evaluation and Prediction of Seepage Flow Rate of Huaishuguan Reservoir Dam

作　　者：焦林可 杨赟 范雪 刘崇昕 李生彬[1] JIAO Linke;YANG Yun;FAN Xue;LIU Chongxin;LI Shengbin(School of Environmental and Municipal Engineering,Lanzhou JiaotongUniversity,Lanzhou 730070,China;Linxia Hui Autonomous Prefecture Water Conservancy Research Institute,Linxia 731199,Gansu,China)

机构地区：[1]兰州交通大学环境与市政工程学院,兰州730070 [2]临夏回族自治州水利科学研究院,甘肃临夏731199

出　　处：《兰州交通大学学报》2024年第2期104-113,共10页Journal of Lanzhou Jiaotong University

基　　金：临夏州槐树关水库大坝安全运行及水土保持监测研究(2021-Z-29)。

摘　　要：为评价槐树关水库大坝的渗流稳定性,收集大坝近年来面板周边缝与坝基监测点处渗压计数据,库水位、气温、降雨量、坝后渗流的监测数据及大坝工程建设与地质资料,先将监测点处渗压计值和坝前水位进行Pearson相关性分析,再根据大坝工程建设及地质资料绘制大坝断面模型并导入Autobank7.7软件计算坝后渗流量,最后将气温、库水位、降雨量等影响因子代入PCA BP神经网络与随机森林、BP神经网络模型对坝后渗流量进行预测。结果表明:除P_(3)、P_(6)、P_(9)外的监测点渗压计值、坝后量水堰监测渗流量值与库水位相关性较强,符合大坝渗流场的一般规律,初步判定大坝较稳定;实际监测的坝后渗流量是Autobank软件模型理论计算值的8倍,判定大坝有出现渗流破坏的可能,后续应采取相应措施除险加固加强监测以保证大坝的稳定性;PCA BP神经网络与随机森林、BP神经网络3种模型最佳预测准确度分别为83.20%、79.35%、69.41%,其中PCA BP神经网络的预测效果最佳,通过模型预知渗流量防止渗流破坏对保证大坝的安全运行具有重要意义。To evaluate the seepage stability of the Huaishuguan Reservoir Dam,the seepage pressure gauge data at the panel peripheral joints and monitoring points at the dam foundation in recent years,as well as monitoring data on reservoir water level,temperature,rainfall,seepage behind the dam,and dam engineering construction and geological data were collected.Firstly,Pearson correlation analysis was conducted between the seepage pressure gauge values at the monitoring points and the water level in front of the dam,based on the construction of the dam project and geological data,a dam cross section model is drawn and imported into Autobank 7.7 software to calculate the seepage flow after the dam.Finally,factors such as temperature,reservoir water level,and rainfall are substituted into PCA BP neural network,random forest,and BP neural network models to predict the seepage flow after the dam.The results show that,except for P_(3)、P_(6) and P_(9),the monitoring points of the seepage pressure gauge and the seepage flow rate of the measuring weir behind the dam have a strong correlation with the reservoir water level,which conforms to the general law of the seepage flow field of the dam.It is preliminarily judged that the dam is relatively stable;The actual monitored seepage flow rate behind the dam is 8 times the theoretical calculation value of the Autobank software model.It is determined that there is a possibility of seepage damage to the dam,and corresponding measures should be taken to eliminate risks and streng then monitoring to ensure the stability of the dam.The best prediction accuracy of PCA BP neural network,random forest,and BP neural network models are 83.20%,79.35%,and 69.41%,respectively.Among them,PCA BP neural network has the best prediction effect.It is of significant importance to ensure the safe operation of the dam by predicting seepage flow through modeling to prevent seepage damage.

关 键 词：大坝渗流监测 安全评价 Autobank7.7渗流校核 网络模型预测 

分 类 号：TV641[水利工程—水利水电工程]