A new approach for seepage quantity prediction of earth and rock dams based on SSA-LSSVM

The enhancement of stability and safety in earth and rock dams is imperative for effective monitoring and prediction of seepage. Existing seepage prediction models frequently encounter challenges, including slow convergence and model parameters that are susceptible to local optimal solutions in practical applications. This paper puts forth a methodology for predicting seepage quantity in earth and rock dams using the Sparrow Search Algorithm (SSA) and the Least Squares Support Vector Machine (LSSVM). The methodology begins by identifying factors that influence the seepage quantity of an earth and rock dam. Subsequently, a LSSVM is trained to predict the seepage quantity of earth and rock dams, with the parameters being optimised by the SSA. Ultimately, a new method for predicting seepage quantity in earth and rock dams is proposed. Test samples demonstrate an MAE, MAPE, and RMSE of 0.304 L/s, 1.727%, and 0.357 L/s, respectively for the model, showing enhanced stability and prediction accuracy. For the prediction of small-sample high-dimensional data, the method demonstrates remarkable reliability and stability, offering a novel approach for the safety monitoring and operation of earth and rock dams.