滑坡预测对于减轻地质灾害的危害十分重要，但对科学研究却很有挑战性。基于变形特征和位移监测数据，建立了三峡库区白水河滑坡的时间序列加法模型。在模型中，累计位移分为3个部分：趋势、周期和随机项，解释了由内部因素（地质环境，重力等）、外部因素（降雨，水库水位等）、随机因素（不确定性）共同作用的影响。在对位移数据进行统计分析后，提出了一个3次多项式模型对趋势项进行学习，并利用多算法寻优的支持向量回归机（SVR）模型对周期项进行训练与预测。结果表明，在预测精度上，基于时间序列与遗传算法－支持向量回归机（GA-SVR）耦合的位移预测模型要明显优于网格寻优（GS）以及粒子群算法（PSO）优化的支持向量回归机模型。因此，GA-SVR 模型在滑坡位移预测方面可以得到较好的应用。在“阶跃型”滑坡位移预测中，GA-SVR 将具有广阔的应用前景。%Prediction of landslides is very important for mitigating geo-hazards but is scientifically very challenging. Based on the deformation characteristics and the data measured by monitoring of Baishuihe landslide,Three Gorges Reservoir,a time series addition model was established for the landslide displacement prediction.In the model,the accumulative displacement is divided into three parts:trend,periodic and random terms,which can be explained by the internal factors(geological environment,gravity,et al.),external factors(rainfall,reservoir water level,et al.),random factors(uncertainties).After statistically analyzing the displacement data,a cubic polynomial model was proposed for the trend term displacement prediction.And using multi algorithm to find the optimal support vector regression (SVR)model to train and predict the periodic term.The results show that based on Genetic algorithm(GA-SVR)and the time series model,the displacements prediction of Baishuihe landslide are better than the landslide displacements based on the particle swarm optimization(PSO-SVR)and the grid search(GS-SVR) model.Therefore,the coupling model based on the time series and GA-SVR can be used to predict the landslide displacement.GA-SVR will have broad application prospects in the prediction of the “step type ” landslide displacement.
展开▼
