基于Copula理论的库岸边坡变形预测模型及概率失稳判据

Monitoring and forecasting is one of important methods to prevent and minimize the geological hazards of reservoir bank slopes. All the existing prediction methods and failure criterions don’t properly address the uncertainty characteristics of soils and rocks. For that reason, typical landslides from the Three Gorges Reservoir area are going to be considered as case study to meet the objectives of this research. Effective identification of optimal joint distribution of variables will be achieved from the random variables (i.e., key responsible factors for the landslide) by improving the entropy weight of Copula function. Copula based landslide prediction model will be established to perform a probabilistic evaluation. Based upon the prediction model, a probabilistic failure criterion will be formulated. Then the proposed prediction model and failure criterion will be compared and contrasted with a physical model test of landslide under a combined action of rainfall and reservoir water. The relationships between key landslide responsible factors and their impacts will be investigated by using a measured data from the Three Gorges Reservoir area. The probabilistic failure criterion method can be used to predict the probabilistic failure criterion for colluvial landslides. The outcomes of this research program will have an immense contribution to enhance landslide prediction methods and theories. In the meantime, it can be used as a guideline for practical engineering problems.

监测预报是防止和减轻库岸边坡地质灾害的重要手段之一，现有的滑坡预测预报方法、失稳判据研究都试图给出某一确定的预测结果或判据信息，较少考虑岩土体的不确定性特征。基于此，以三峡库区典型堆积型滑坡为研究对象，通过库岸边坡关键致灾因子识别选定随机变量；改进熵权Copula函数，实现变量最优联合分布的有效识别；通过构建基于Copula函数的滑坡变形预测模型，实现对滑坡变形的概率预估；基于变形预测模型研究，提出滑坡概率失稳判据方法，实现从不确定性角度推求滑坡失稳判据；通过降雨及库水联合作用下的滑坡物理模型试验，验证所提出的变形预测模型及失稳判据方法；基于三峡库区典型堆积型滑坡降雨量、库水位及测点变形监测数据，开展关键致灾因子及作用效应相关关系研究，并借助概率失稳判据方法，实现典型堆积型滑坡一定概率的失稳判据推求。预期研究成果有助于丰富滑坡预测预报方法和理论，为库岸边坡灾害防治提供重要参考。

三峡库区岸坡变形预测及灾害预报是当前亟待解决的问题之一。现有的预测预报方法、失稳判据都试图给出某一确定的预测结果或判据信息，忽略了岩土体的不确定性特征，研究项目从不确定性角度出发，以三峡库区典型堆积型滑坡为对象展开工作。在理论方法方面，建立了坡体变形、降雨量及库水位多元监测数据变量的熵权Copula联合分布模型；提出了提出了堆积型滑坡多元可靠性分析方法，包含堆积型滑坡g-line可靠性分析方法和基于无味变换的边坡可靠性分析方法；构建了Copula理论下滑坡变形预测模型，并提出了变形表征的滑坡概率失稳判据方法，实现了一定保证率下的滑坡变形预测和判据推求。在试验研究方面，开展了堆积型滑坡变形破坏三维物理模型试验，重点分析了不同作用下堆积型滑坡的变形破坏时空演化过程，分别建立降雨和库水升降联合作用、冰雪融水与库水联合作用下的堆积型滑坡变形破坏模式，研究了堆积体力学、渗流参数的时间劣化效应和空间变异规律。在工程应用方面，推求了由暴雨、地震因素表征的白水河、树坪等滑坡的失稳判据，并开展了考虑堆积体力学参数沿高程变异的树坪滑坡稳定性分析，部分研究成果被三峡库区地质灾害预警指挥应急监测专业技术团队所采纳，为库区岸坡地质灾害防治提供了重要技术支撑。研究项目已发表论文13篇，其中SCI/EI论文9篇，申请发明专利2项，培养博士、硕士研究生7人，项目负责人晋升副教授，获评三峡大学三峡学者，研究成果支撑获批省部级科技进步一等奖1项、二等奖1项。