Safety Evaluation and Prediction of Seepage Flow Rate of Huaishuguan Reservoir Dam
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 res-ervoir water level,temperature,rainfall,seepage behind the dam,and dam engineering construction and geological da-ta 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 geo-logical 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 re-sults show that,except for P3,P6,and P9,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 soft-ware 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.
dam seepage monitoringsafety evaluationAutobank 7.7 seepage verificationnetwork model prediction
NETL
NSTL
万方数据
维普
