基于贝叶斯方法的节理岩质边坡随机裂隙网络修正及可靠性研究

A large number of complicated discontinuities exist within rock slopes, which play a dominant role in the stability of rock slopes. To tackle the difficulties in reliability analysis of complex jointed rock slopes, this project aims to propose an adaptive Bayesian updating approach to infer the probability distributions of geometric parameters of joints by a strategy combining theoretical modeling, numerical simulation, experimental verification and engineering application. An approach for characterizing the measurement uncertainties of geometric parameters of structural planes will be developed. Thereafter, the influences of prior information and sampling size on the updating of probability distributions of the geometric parameters will be addressed. Then, a technique for correcting the discrete fracture network of jointed rock mass will be proposed using Bayesian approach. Laboratory experiments will be carried out to validate the proposed technique. It is anticipated that the structural planes network models of jointed rock slopes that are well consistent with engineering practice will be constructed by taking full advantage of limited field measurements and laboratory testing data. Finally, an efficient reliability analysis approach for jointed rock slopes based on surrogate models will be presented. The slope at the left bank of Jinping I Hydropower Station will be taken as a typical case to explore the effects of the uncertainties of geometric parameters as well as mechanical parameters of structural planes on the reliability of jointed rock slopes. The research outcomes of this project will lay the foundation for understanding the failure mechanism of jointed rock slopes and formulating the rational strategies for landslide risk mitigation.

岩质边坡广泛发育着大量复杂的不连续结构面，对边坡稳定性起着决定性作用。针对复杂节理岩质边坡可靠性分析难题，本项目拟采用理论分析、数值模拟、实验验证和工程应用相结合的研究方法，提出基于自适应贝叶斯更新的节理几何参数概率分布推断方法，发展结构面几何参数测量不确定性表征方法，阐明先验信息和样本量对几何参数概率分布推断的影响规律。在此基础上，开发一套基于贝叶斯方法的岩体随机裂隙网络修正技术并进行实验验证，实现基于有限的野外实测资料和室内试验数据建立切合工程实际的节理岩质边坡结构面网络模型。最后，提出基于代理模型的节理岩质边坡可靠性高效分析方法，并依托锦屏一级水电站左岸边坡开展应用研究，揭示同时考虑结构面几何参数及力学参数不确定性的节理岩质边坡可靠性演化规律。本项目研究成果将为深入认识节理岩质边坡变形破坏机理和指导制定合理的滑坡风险防治措施提供重要的理论与技术支持。

岩质边坡广泛发育着大量复杂的不连续结构面，对边坡稳定性起着决定性作用。针对复杂节理岩质边坡可靠性分析难题，本项目采用理论分析、数值模拟、实验验证和工程应用相结合的研究方法，提出了节理几何参数概率分布推断的修正贝叶斯更新方法，发展结构面互相关产状模拟的Copula函数方法，阐明了先验信息和样本量对几何参数概率分布推断的影响规律。在此基础上，开发了一套基于贝叶斯方法的岩体随机裂隙网络修正技术并进行实验验证，实现了基于有限的野外实测资料和室内试验数据建立切合工程实际的节理岩质边坡结构面网络模型。最后，提出了基于代理模型的节理岩质边坡可靠性高效分析方法，并依托锦屏一级水电站左岸边坡开展了工程应用研究，揭示了同时考虑结构面几何参数及力学参数不确定性的节理岩质边坡可靠性演化规律。围绕本项目研究，出版学术专著1部、英文编著1部，共发表相关学术论文36篇，其中SCI论文17篇、EI论文11篇、CSCD论文3篇、国际会议论文4篇、中文会议论文1篇，有7篇SCI论文被一区Top期刊收录，并取得计算机软件著作权登记9项，授权国家发明专利4项、实用新型专利1项。本项目研究成果将为深入认识节理岩质边坡变形破坏机理和指导制定合理的滑坡风险防治措施提供重要的理论与技术支持。