摘要
为解决滑坡危险性评价不准确的问题,以某道路滑坡为工程背景,首先,通过现场调查、监测数据统计来分析滑坡变形特征;其次,以滑坡变形数据为基础,利用集合经验模态分解开展数据分解处理,并通过门控循环单元神经网络、天牛群算法和广义回归神经网络构建滑坡变形组合预测模型,再根据高精度变形预测结果计算滑坡复活后的危险性分级指标,以准确实现滑坡复活后的危险性评价。分析结果表明:在滑坡变形特征分析方面,滑坡变形初始时间与公路建设基本一致,其后历年变形也时有发生,即变形位置与公路建设密切相关,变形时机与降雨密切相关。同时,通过滑坡变形组合预测,得出预测思路具有高精度预测能力,其预测结果的可信度高,计算得到滑坡危险性评价指标范围为0.728~1.062,说明滑坡不同监测点位置的复活危险性程度存在一定差异,对应复活危险性等级介于Ⅱ~Ⅳ级,其局部失稳风险较大。根据滑坡危险性评价结果,若开展滑坡整体防治,建议按照不利原则,滑坡整体防治建议按Ⅳ级执行;若开展滑坡局部防治,建议结合具体位置对应的危险性等级进行针对性防治。
To address the issue of inaccurate risk assessment after the revival of landslides,a landslide on a certain road was used as the engineering background to analyze the deformation characteristics of the landslide through on-site investigation and monitoring data statistics.Based on landslide deformation data,data decomposition processing is carried out using ensemble empirical mode decomposition,and a landslide deformation combination prediction model is constructed through gate controlled recurrent unit neural network,beetle swarm algorithm and generalized regression neural network.Based on the high-precision deformation prediction results,the risk classification index of the landslide after resurrection is calculated to accurately achieve the risk evaluation of the landslide after resurrection.The analysis results indicate that in terms of landslide deformation characteristics analysis,the initial time of landslide deformation is basically consistent with highway construction,and deformation also occurs from time to time in subsequent years,that is,the deformation position is closely related to highway construction,and the deformation timing is closely related to rainfall.At the same time,through the combination prediction of landslide deformation,it was found that the prediction idea in this article has high-precision prediction ability,and the reliability of its prediction results is high.The range of landslide hazard evaluation indicators was calculated to be 0.728~1.062,indicating that there are certain differences in the degree of revival hazard at different monitoring points of the landslide,corresponding to revival hazard levels ranging fromⅡtoⅣ,and its local instability risk is relatively high.According to the landslide hazard assessment results,if the overall prevention and control of landslides is carried out,it is recommended to follow the principle of disadvantage,and the overall prevention and control of landslides is recommended to be implemented at levelⅣ,if local landslide prevention and control is carried out,it is recommended to carry out targeted prevention and control based on the corresponding danger level of the specific location.
作者
李振基
薛晓辉
LI Zhenji;XUE Xiaohui(Suning Branch Safety Supervision Office,Guoneng Shuohuang Railway Development Co.,Ltd.,Cangzhou 062350,China;School of Engineering Management and Logistics,Shaanxi Railway Institute,Weinan 714099,China;Hebei Key Laboratory of Geotechnical Engineering Safety and Deformation Control,Hebei University of Water Resources and Electric Engineering,Cangzhou 061001,China)
出处
《甘肃科学学报》
2025年第5期21-27,49,共8页
Journal of Gansu Sciences
基金
河北省岩土工程安全与变形控制重点实验室(河北水利电力学院)开放基金(HWEKF202103)。
关键词
道路滑坡
变形组合预测
数据分解
广义回归神经网络
复活危险性
Road landslides
Deformation combination prediction
Data decomposition
Generalized regression neural network
Revival risk
