人民长江2024，Vol.55Issue(7)：108-114,7.DOI:10.16232/j.cnki.1001-4179.2024.07.014
基于FEEMD-GRU-FC模型的滑坡位移预测
Landslide displacement prediction based on FEEMD-GRU-FC model
摘要
Abstract
Landslides in reservoir areas are affected by rainfall and periodic changes in reservoir water level,which will interfere with the prediction performance of landslide displacement prediction models.A FEEMD-GRU-FC(fast ensemble empirical mode decomposition-gated recurrent unit-fully connected layer)landslide displacement prediction model based on the periodic environmental changes was proposed.The model fully considered the influence of noise among different frequencies of displace-ment data,divided the displacement into sub-sequences of different frequencies to predict them respectively,and used the PCC(Pearson correlation coefficient)method to select external environmental factors with large influence.The ability of the GRU mod-el to extract periodic change features from environmental data was improved by adding a full connection layer.The model was ap-plied by combining with the actual data of the Bazimen Landslide in the Three Gorges Reservoir area.Different prediction models and evaluation indicators were used for multi-perspective comparison and verification.The experimental results showed that the model proposed was better than the other three models in predicting Bazimen Landslide displacement.Its fitting degree was 98.6%,and three kinds of error indexes were the lowest,which proved that the model can accurately predict the landslide dis-placement in the reservoir area.The research results can guide disaster prevention and reduction.关键词
滑坡位移预测/皮尔逊相关系数/门控循环单元/全连接层/非线性/八字门滑坡/三峡库区Key words
landslide displacement prediction/Pearson correlation coefficient/gated recurrent unit/fully connected layer/nonlinearity/Bazimen Landslide/Three Gorges Reservoir area分类
建筑与水利引用本文复制引用
何清,李丽琳,林子安..基于FEEMD-GRU-FC模型的滑坡位移预测[J].人民长江,2024,55(7):108-114,7.基金项目
国家自然科学基金项目(U23A20280) （U23A20280）
广西科技重大专项(桂科AA23062038) （桂科AA23062038）
广西重点研发项目(桂科AB21196041) （桂科AB21196041）
