Study Of Seepage Properties Of Clay Core Wall Dams Based On Monitoring Data

Due to its simple construction and easy availability of materials,earth and rock dams occupy the vast majority of dam types and play a huge role in securing irrigation,water supply,and shipping needs.However,earth and rock dams have a high incidence of seepage damage due to their own material characteristics and the complex external environment in the course of long-term service.In order to objectively assess the seepage safety state of earth and rock dams,this paper is based on the monitoring data of a reservoir and systematically investigates three parts: noise reduction processing of abnormal values in monitoring data,analysis of monitoring data and optimization of monitoring models.The main research contents are as follows:(1)In response to the problem that noise and anomalies will be mixed in the seepage monitoring data acquisition process,which affects the accuracy of monitoring data analysis and model construction,the variational modal decomposition combined with permutation entropy algorithm is used to decompose and reconstruct the monitoring data pre-processing,which improves the continuity and reliability of monitoring data.(2)The qualitative analysis(process line method,correlation analysis and potential analysis)and quantitative analysis(statistical modeling method)were used to analyze the seepage behavior of the dam using the pre-processed monitoring data,and the qualitative analysis showed that the seepage behavior of the dam was in accordance with the seepage law of the heart wall dam,and the seepage behavior was good.By establishing a multivariate regression statistical model of the pressure tube water level,the analysis results show that the model fits well,the upstream and downstream water levels are the main influencing factors on the pressure tube water level,rainfall and temperature have little effect on the seepage of the dam,and the negative time-effect factor indicates that the dam solidifies with time deposition,and the overall seepage behavior of the dam is normal and the seepage prevention effect is obvious.(3)In order to overcome the shortcomings of the inaccuracy of the seepage statistical model to characterize the nonlinear characteristics of seepage,a combined prediction model based on the deep limit learning machine and using the whale algorithm to optimize the number of neuronal nodes in each hidden layer is proposed.In order to verify the effectiveness of the above model,combined with the actual monitoring data of a reservoir in Chongqing,the whale algorithm optimization deep limit learning machine(WOA-DELM)model was compared with the traditional statistical model and the single deep limit learning machine(DELM)model.The results show that the WOA-DEMM model has higher fitting accuracy and higher prediction accuracy,and has good practical application value.