摘要
为了减少库区滑坡失稳可能带来的损失,确保水电工程的安全稳定运行及保障周边民众的生命财产安全,探讨提升库区滑坡位移预测精度的方法。提出结合频率因子(frequency factor,FF)和变分模态分解-灰色关联度分析-误差反向传播神经网络(variational mode decomposition-grey relational analysis-back propagation neural network,VMD-GRA-BP)模型的库区滑坡位移预测方法,并应用于三峡库区滑坡。研究结果表明:FF对周期位移分量的预测效果有一定的提升,VMD-GRA-BP模型能够简化复杂非线性库区滑坡位移预测问题,BP模型预测性能优异,预测精度多高于支持向量机(support vector machine,SVM)和k最近邻(k-nearest neighbor,KNN)模型。研究结果可为库区滑坡位移预测和预警提供参考。
In order to reduce the possible losses caused by landslide instability in the reservoir area and ensure the safety of hydropower projects and people,the method of improving the prediction accuracy of landslide displacement in the reservoir area is discussed.A landslide displacement prediction method combining frequency factor(FF)and variational mode decomposition-grey relational analysis-back propagation neural network(VMD-GRA-BP)model is proposed and applied to landslides in the Three Gorges Reservoir area.The results show that FF has a certain improvement on the prediction effect of periodic displacement component.The VMD-GRA-BP model can simplify the prediction of landslide displacement in complex nonlinear reservoir area.The BP model has high prediction performance,and the prediction accuracy is mostly higher than those of support vector machine(SVM)and k-nearest neighbor(KNN)models.The research results can provide reference for the landslide displacement prediction and early-warning in the reservoir areas.
作者
贺明晓
刘阳
唐怡
杨辉宗
吴卓恩
张建经
罗宏森
HE Mingxiao;LIU Yang;TANG Yi;YANG Huizong;WU Zhuoen;ZHANG Jianjing;LUO Hongsen(Institute of Technology,Sichuan Normal University,Chengdu Sichuan 610101,China;Institute of Public Safety and Emergency Response,Sichuan Normal University,Chengdu Sichuan 610066,China;School of Civil Engineering,Southwest Jiaotong University,Chengdu Sichuan 610031,China)
出处
《中国安全生产科学技术》
CAS
CSCD
北大核心
2024年第9期96-104,共9页
Journal of Safety Science and Technology
基金
国家重点研发计划项目(2017YFC0504901)
四川省科技计划项目(2023NSFSC1038)
四川省大学生创新训练计划项目(S202410636063)
四川师范大学教改项目(20210471XKC)。
