基于因果分析的滑坡位移预测模型  

Landslide displacement prediction model based on causal analysis

作　　者：孙华芬[1,2] 曹念 侯克鹏 SUN Huafen;CAO Nian;HOU Kepeng(Faculty of Land Resource Engineerng,Kunming University of Science and Technology,Kunming 650000,China;Key Laboratory of Sino-German Blue Mining and Utilization of Special Undeground Space,Kunming 65ooo0,China)

机构地区：[1]昆明理工大学国土资源工程学院,昆明650000 [2]云南省中-德蓝色矿山与特殊地下空间开发利用重点实验室,昆明650000

出　　处：《测绘科学》2024年第3期47-57,共11页Science of Surveying and Mapping

基　　金：云南省面上项目(202001AT070205);云南省大学生创新创业项目(S202310674105)。

摘　　要：针对滑坡位移预测模型对防灾减灾和损失评估有重要意义的问题,以八字门滑坡为例,基于位移监测数据建立了考虑时滞的滑坡位移CEEMD-SIWOA-BP组合预测模型。利用互补集合经验模态分解方法将位移监测数据分解成多个信号分量,重构为滑坡趋势项及周期项;构建BP-SIWOA模型,引入Singer混沌映射及自适应权重因子提高鲸鱼算法的全局搜索和收敛能力,利用改进的鲸鱼算法对BP神经网络模型的连接权重及阈值项赋值,对趋势项位移用三次多项式进行预测,而对周期位移项考虑时滞效应,利用收敛交叉映射法对降雨量及库水位与周期位移间的因果关系进行了分析和位移预测;将各分量结果叠加得到滑坡位移累计预测值,并评价了预测精度。结果表明,该方法性能优于其他模型,验证了考虑时滞的CEEMD-SIWOA-BP组合预测可行性,能为滑坡灾害预警预报提供技术参考。The landslide displacement prediction model is of great significance for disaster prevention and loss assessment.Taking the Bazimen landslide as an example,A CEEMD-SIWOA-BP combined prediction model for landslide displacement considering time delay was established based on displacement monitoring data in this paper.Firstly,using the complete ensemble empirical mode decomposition(CEEMD)method,the displacement monitoring data is decomposed into multiple signal components and reconstructed into landslide trend and periodic terms;Then,a BP-SIWOA model was constructed,and Singer(SI)chaotic mapping and adaptive weight factors were introduced to improve the global search and convergence ability of the whale optimization algorithm(WOA).The improved whale optimization algorithm was used to assign connection weights and threshold terms to the BP neural network model,predict the trend term displacement using a cubic polynomial,and consider the time delay effect for the periodic displacement term.The causal relationship between rainfall,reservoir water level,and periodic displacement was analyzed and predicted using the convergent cross mapping(CCM)method;Finally,the cumulative prediction value of landslide displacement was obtained by overlaying the results of each component,and the prediction accuracy was evaluated.The results show that the performance of this method is superior to other models,verifying the feasibility of CEEMD-SIWOA-BP combined prediction considering time delay,and providing technical reference for landslide disaster warning and prediction.

关 键 词：滑坡位移 因果分析 互补集合经验模态分解 BP神经网络 改进鲸鱼优化算法 

分 类 号：P237[天文地球—摄影测量与遥感]