关键词: 滑坡变形预测, 集合经验模态分解, 遗传算法优化, 支持向量机

Abstract: In the ordinary landslide deformation prediction method, under the influence of external factors, the displacement sequence is easy to show random and nonlinear characteristics, which makes the landslide deformation difficult for prediction, so it is necessary to put forward more effective prediction methods. Based on the ensemble empirical mode decomposition (EEMD), the original landslide displacement time series was decomposed, so several groups of new displacement deformation subseries with obvious complexity difference were obtained. Then, according to the characteristics of each deformation subsequence, the deformation subsequence prediction model based on GA-SVM can be established respectively. And then each subsequence prediction model can be superimposed, finally a support vector machine landslide deformation prediction model based on sensemble empirical mode decomposition and genetic algorithm optimization (EEMD-GA-SVM) was constructed. Taking Xiangluba Village landslide in Enshi City as an example, by comparing the prediction accuracy of various slope deformation prediction models of EEMD-GA-SVM, BPNN, SVM and GA-SVM, it was found that the support vector machine prediction model based on ensemble empirical mode decomposition and genetic algorithm optimization was more accurate and reliable, which can provide valuable reference for landslide safety monitoring.

Key words: landslide deformation prediction, ensemble empirical mode decomposition, genetic algorithm optimization, support vector machine