Combined Prediction Model of Concrete Arch Dam Displacement Based on Signal Residual Correction and PSO-RVM
Traditional prediction models for displacement of concrete arch dams only consider the influence of various factors on the prediction results,without fully exploiting the valuable information within the residual sequence.There-fore,this paper proposed particle swarm optimization-relevance vector machine(PSO-RVM)model for displacement pre-diction of arch dams,which combined the RVM with the PSO.Firstly,the singular spectrum analysis(SSA)was applied to decompose the residual sequence,and the components were reconstructed based on the cumulative contribution rate of singular values.Subsequently,the random forest(RF)algorithm was employed to predict the reconstructed sequence and obtain the correction values for the residual sequence.Finally,the predicted values from the PSO-RVM model and the correction values of the residual sequence were superimposed to establish the PSO-RVM+ model,a combined prediction model for the displacement of concrete arch dams based on signal residual correction and PSO-RVM.Engineering exam-ples demonstrate that the SSA and RF algorithms are effective in extracting valuable information from the residual se-quence and correcting the residuals.Compared to the PSO-RVM,RVM,and traditional regression models(SWR),the PSO-RVM+ composite prediction model exhibits superior predictive performance and stronger adaptability.This study provides a new perspective for dam safety monitoring and health diagnosis.
万方数据
维普
