Displacement prediction method of rainfall-induced landslide considering multiple influencing factors

Predicting rainfall-induced landslide displacement is one of the important means of disaster prevention and mitigation. Considering the Tanjiawan landslide in the Three Gorges Reservoir area as the research object, the daily rainfall and soil moisture content as influencing factors, complementary ensemble empirical mode decomposition (CEEMD) was used to decompose the time series of displacement and influencing factors, followed by K-means clustering to determine the periodic displacement, random displacement, trend displacement, and their corresponding influencing factor components after decomposition. The Grey System theory was used to test the correlation between the influencing factor and decomposition displacement, and the least squares support vector machine based on particle swarm optimization (PSO-LSSVM) and the least square method were used to predict the decomposition displacement. The results showed that after decomposition and clustering, the grey relational degree between the influencing factor and the decomposition displacement is up to 0.91, which showed that the selection of the displacement decomposition and the influencing factor is reliable. A coefficient of determination of 1.00 indicated that the quadratic least squares function model can predict the trend displacement well, and the root mean squared error value of the PSO-LSSVM model predicting displacement did not exceed 21.62 mm. At the same time, compared with the prediction results without considering water content as the influencing factor, the results show that the prediction effect considering water content as the influencing factor is very reliable, and the model in this study can achieve the displacement prediction of rainfall-type landslides satisfactorily.

Automated summary

The study successfully predicted rainfall-induced landslide displacement using methods such as complementary ensemble empirical mode decomposition and K-means clustering, combined with Grey System theory, PSO-LSSVM, and least square methods. The results showed that the model is reliable and effective in predicting landslide displacement, with a high grey relational degree and low root mean squared error value.