作者简介:王朝义(1988-),男,工程师,主要从事地质灾害调查评价与防治、地质灾害预警监测研究。E-mail: c595743526@163.com
DOI: 10.13512/j.hndz.2026.05.22
备注
为解决多时间尺度信息处理困难,导致滑坡体位移预测精度较低的问题,提出基于PSO-SVM算法和时间序列模型的滑坡体位移预测方法。以时间戳为指标建立各类影响因素的数据时间序列,并采用灰色关联分析确定各影响因素与滑坡体位移的关联度;根据数据时间序列,采用差分自回归移动平均模型构建时间序列预测模型,设置逼近预测精度的目标函数,添加滑坡移动过程的限制条件;创新性地在时间序列预测模型中引入支持向量机的回归函数,采用支持向量机(SupportVectorMachine,SVM)算法优化模型,改进位移预测的目标函数,引入粒子群优化算法(ParticleSwarmOptimization,PSO)迭代求解模型,获得最优模型解,实现滑坡体位移预测。实验结果表明:该方法对于滑坡体位移的预测均方根误差均值仅为0.98mm,预测精度较高。
To solve the problem of low accuracy in landslide displacement prediction caused by difficulties in pro‐cessing information at multiple time scales, a landslide displacement prediction method based on PSO-SVM algo‐rithm and time series model is proposed. The data time series of various influencing factors were established with timestamp as indicators, and the correlation degree between each influencing factor and landslide displacement was determined by grey correlation analysis. Based on the data time series, an autoregressive integrated moving average (ARIMA) model was used to construct the time series prediction model, the objective function was set for approach‐ing the prediction accuracy, and the constraints conditions were added to the landslide movement process. The re‐gression function of support vector machine was innovatively introduced into the time series prediction model. The support vector machine (SVM) algorithm was used to optimize the model and improve the objective function of dis‐placement prediction. The particle swarm optimization (PSO) algorithm was introduced to iteratively solve the model to obtain the optimal model solution and realize the displacement prediction of landslide. The experimental results show that the average root mean square error of this method for predicting landslide displacement is only 0.98 mm, indicating high prediction accuracy.





















