Study On Seepage Characteristics Early Warning Models Of Earth-rock Dam Of Fenhe Reservoir

Global water resources exploitation and utilization degree increases,whichmake the expansion of scale and number of dams and other types of hydraulicstructures become inevitable,but it also brings along dam frequent crash. Thecrash of France Marr Barcelona hyperbolic arch dam,Italy Vai Aung arch damand Henan itabashi,Qinghai hougou dam in China caused immeasurable lossesto the national economic development and people’s life safety. According todomestic and foreign statistical analysis of dam safety incidents,in the world thenumber of embankment dam is largest compared to of other types dam amongthe crashing dams,which is related to the largest number of embankment dam.The main reasons of crash of embankment are flood overtopping embankmentdam, infiltration sabotage and pipeline leaks,which perform as dam and dam base seepage,piping,flow of soil,pits collapse and other phenomena. Sopredicting abnormal situation and future operation traits of embankment damsby building an appropriate model to analyze seepage evolution of embankmentdam according to the seepage data has a big significance.This paper takes seepage characteristics early warning models research ofearth-rock dam of Fenhe reservoir in Shanxi province,the main contents of thispaper include:(1) seepage has periodic fluctuation,random fluctuations and trendscharacteristics. Model combined with time series has high forecasting accuracy,advanced statistic advantage. Unitary time series model of seepage is established；(2)Empirical Mode Decomposition (EMD) can remain original data property. Periodicterms and trend terms are obtained by decomposing seepage monitoring data ofembankment dam using empirical mode decomposition method. Seepage timeseries hybrid model is established based on EMD combining time series analysismethod；(3) Ensemble Empirical Mode Decomposition（EEMD) can the eliminatemixing and smooth data. Seepage time series hybrid model is established base onEEMD by decomposing seepage monitoring data of embankment dam usingEnsemble Empirical Mode Decomposition（EEMD)；(4) The four models above areapplied to predict left shore seepage of Fenhe reservoir. Comparing fitting resultsand prediction accuracy of the four models,technical support is provided forconstruction of seepage model and safe operation of embankment dam.According to the study,this paper (1) Puts forward the time series model(ARI),the seepage hybrid model based on empirical mode decomposition and time series analysis and the seepage hybrid model based on ensemble empiricalmode decomposition and time series analysis,and three models are carried out tofit and forecast the seepage data of the left dam shore of Fenhe reservoir；(2)Relative error (RE),sum of squared residuals (SSE),goodness of fit and meanabsolute percentage error (MAPE) indexes are tested on the three models,theresults show that fitting effect and prediction accuracy of EEMD-ARI model isoptimal,EMD-ARI model followed,and ARI model is the worst of all；(3)Combined with the model residual graph to further verified the results thatfitting effect of EEMD-ARI model is better than that of EMD-ARI model,andEMD-ARI model is better than that of ARI model；(4) The EEMD-ARI modelcan be used to forecast the seepage flow of the left dam shore of Fenhereservoir,which can provide technical support for the seepage hybrid modelconstruction of the dam and the development of decision support system for thedam safety operation.This paper the research results of seepage early warning model combinedwith practical engineering have a certain reference value for the safetyoperation condition monitoring of earth-rock dam and model construction,alsohave a certain application value.