滑坡位移EEMD-SVR预测模型

滑坡位移预测是滑坡灾害实时监测预警的重要组成部分,良好的滑坡位移预测模型有助于预测地质灾害发生。滑坡变形受多种外界因素影响呈现出随机性和非线性的特点,在现有的滑坡位移预测方法中,机器学习方法在滑坡位移预测中得到了广泛的应用。针对滑坡位移预测是趋势项位移和周期项叠加的特点,本文研究采用基于集成经验模态分解(EEMD)的滑坡趋势项和周期项位移提取方法,结合支持向量回归(SVR)模型实现对滑坡的位移预测。首先,详细介绍了该模型的构建过程和预测性能,并以均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分比误差(MAPE)和决定系数(R²)作为评估模型的预测性能指标。然后,分别利用EEMD-SVR、SVR、Elman模型对贵州省岩溶山区的一处滑坡进行位移预测,结果表明,EEMD-SVR模型连续1 d预测的RMSE值、MAPE值和R²值分别为0.648 mm、0.518%和0.996 8,可以提供更高可靠的滑坡位移预测精度,对同类滑坡的位移预测具有一定的参考价值。

Displacement prediction model of landslide based on ensemble empirical mode decomposition and support vector regression

Landslide displacement prediction is an important part of real-time monitoring and early warning of landslide disasters. A good landslide displacement prediction model is helpful to predict the occurrence of geological disasters. The deformation of the landslide is affected by a variety of external factors and presents the characteristics of randomness and nonlinearity. Among the existing landslide displacement prediction methods, machine learning methods have been widely used in landslide displacement prediction. Prediction of landslide displacement is the superposition of trend displacement and periodic displacement. In this study, the displacement extraction method of landslide trend term and period term based on integrated empirical mode decomposition (EEMD) and the support vector regression (SVR) model were used to predict the landslide displacement. The construction process and prediction performance of the model are introduced in detail, and the root mean square error (RMSE), average absolute error (MAE), average absolute percentage error (MAPE) and coefficient of determination (R²) are used as the predictive performance indicators of the evaluation model. The EEMD-SVR, SVR, and Elman models are used to predict the displacement of a landslide in the karst mountainous area of Guizhou province. The results showed that the RMSE, MAPE and R² values of EEMD-SVR model were 0.648 mm, 0.518% and 0.996 8, respectively. This model can provide higher and more reliable landslide displacement prediction accuracy, and has certain reference value for the displacement prediction of similar landslide.