Study On Seepage And Slope Safety Analysis Method Of Earth-Rock Dam Based On Integrated Learning

Earth-rock dams are widely used in dam engineering construction in China,and the safety analysis of earth-rock dams is of great significance for the safe operation of earth-rock dams and the development of the national economy.The seepage safety and slope safety of earth-rock dams are important components of safety analysis.Dam safety monitoring is an important means of safety analysis for earth-rock dams,and ensemble learning is an advanced data analysis method that efficiently integrates multiple machine learning algorithms.Therefore,by comprehensively applying the safety monitoring theory and integrated learning methods of earth-rock dams,research on the analysis methods of seepage and slope safety of earth-rock dams is carried out.The impact of the two core influencing factors of reservoir water level fluctuation and rainfall on the seepage and slope safety of earth-rock dams is analysed,which has practical application value for the operation management and safety evaluation of earth-rock dams.Monitoring data analysis and numerical simulation are commonly used methods for conducting the above research.However,statistical models or machine learning models are often used alone to build monitoring models based on monitoring data,which makes the generalization and prediction ability of a single model insufficient and cannot effectively integrate the interpretability of statistical models with the high fitting ability of machine learning models.At the same time,the back analysis of dam material parameters is a key factor in improving the accuracy of numerical simulation results.However,conventional back analysis methods need to be further improved in terms of speed and accuracy due to the limitations of analysis methods.In further dam slope safety analysis,there are also few control methods that provide feedback on the threshold of reservoir water level changes based on dam slope safety changes.In response to the above issues,the main research content of this article is as follows within the framework of integrated learning models:(1)Fully considering the lag effect of seepage in earth rock dams,constructing water level components and rainfall components based on lag effect functions,and applying swarm intelligence optimization algorithms and partial least squares regression to establish parameter solving methods to solve the lag effect parameters,thus establishing an improved statistical model for seepage in earth rock dams,quantitative analysis of reservoir water level and lag time of rainfall.At the same time,a comparative analysis will be conducted between the improved seepage statistical model and the seepage statistical model without considering lag effects,further demonstrating the necessity of improving the seepage statistical model.Finally,within the framework of the ensemble learning model,a new ensemble learning combination strategy based on the DREAMzs algorithm is proposed.The improved statistical model is integrated with multiple classic machine learning models,such as LSTM,to establish an ensemble learning model for seepage monitoring of earth-rock dams,effectively improving the predictive and generalization performance of the model.(2)The hybrid kernel function is used to replace the single kernel function to improve the kernel limit learning machine,and combined with the Latin hypercube sampling algorithm and the sparrow search algorithm,the inversion method of the earth rock dam permeability coefficient based on integrated learning is established.The effectiveness of the inversion method and the superiority of the improved algorithm are verified through examples,which can quickly and effectively estimate the permeability coefficient of dam materials.Then,based on the inverse estimation of the permeability coefficient of the dam material,the effects of the rate of reservoir water level change,rainfall intensity,and type on the seepage field of the dam body were analysed.Finally,based on the analysis of the seepage field,a safety analysis method for dam seepage is established using the criterion that the seepage slope of key parts of the dam is less than the allowable seepage slope.This method can quickly analyse the safety behavior of dam seepage during the process of reservoir water level changes.(3)Based on the integrated learning framework,the forward and backwards analysis method for the safety of earth-rock dams is used to analyse the impact of reservoir water level fluctuations and rainfall on dam slope safety using the limit equilibrium method.The impact of reservoir water level fluctuations on dam slope safety mainly analyses the impact of the initial reservoir water level,reservoir water level variation amplitude,and reservoir water level rise and fall rate on upstream dam slope safety.The impact of rainfall on dam slope safety mainly analyses the impact of rainfall type,rainfall intensity,and rainfall duration on downstream dam slope safety.Then,by combining the dam slope safety analysis model with the dam slope damage index when the reservoir water level changes,a dam slope safety analysis method was established,which can provide a reference for the rapid judgment of dam slope safety when the reservoir water level changes.Finally,by integrating the dam slope safety analysis method with the dam body seepage safety analysis method,a threshold determination method for reservoir water level change is established,which can provide guidance for reservoir water level change in the operation and management of earth-rock dams.