| Abstrakt: |
Considering that the landslide displacement is influenced by multiple factors, a time series decomposition-model construction-model training method of KMD-BP-TIGWO landslide displacement prediction is proposed by combining signal decomposition and intelligent algorithms. Firstly, the landslide monitoring data are decomposed into multiple IMF components and a residual using the EMD method, and the decomposed components are divided into period term and trend term displacements; secondly, the BPTIGWO model is constructed, the Tent mapping and adaptive weights are introduced to improve the convergence speed and global search capability of the whale algorithm, and the weights and thresholds of BP neural network are optimized using the TIGWO algorithm; subsequently, the optimized BP-TIGWO model is used to predict the landslide displacement using the optimized BP neural network and the TIGWO algorithm. Subsequently, the optimized BP model was used to output the predicted values, and finally, the predicted values of each component were superimposed to obtain the predicted values of landslide cumulative displacement, and the model prediction accuracy was evaluated. The experimental results show that the RMSE, MAE and R2 of the EMD-BP-TIGWO model are 0. 64,0. 51, and 0. 97 respectively under the feature of no rainfall input for 1 d. The prediction accuracy of the model is significantly higher than that of the EMD-GWO-BP, EMD-GWO-BP, BP-TIGWO, and BP models that do not consider the time lag, which can provide a reference for predicting the displacement of the landslide. [ABSTRACT FROM AUTHOR] |
