Slopes are frequently stabilized by reinforcement measures in geological and geotechnical engineering. Effectiveness of reinforcement measures is vital to slope failure risk mitigation, but it is affected by geological and geotechnical uncertainties arising from loads, geological formations and geotechnical properties. In face with these uncertainties, slope reinforcement measures are often monitored at site, and monitoring variables that are sensitive to the safety and reliability of the reinforced slope stability shall be carefully selected during monitoring design, which highly depends on engineering experience and judgments (i.e., prior knowledge). How to exercise the prior knowledge in a quantitative and transparent way in geotechnical monitoring design remains unexplored. This can be accomplished using reliability-based monitoring sensitivity analysis, which requires significant computational costs due to repeated inverse analyses and reliability analyses given different possible values of monitoring variables. This paper develops an efficient reliability-based monitoring sensitivity analysis framework based on BUS (i.e., Bayesian updating with structural reliability methods) with Subset Simulation methods. The proposed approach is illustrated using a real reinforced slope example. Results show that it quantifies reliability sensitivity of different monitoring variables (e.g., displacements at different locations on slope surface) in a cost-effective manner based on prior knowledge. Such reliability sensitivity information facilitates decision-making in selecting sensitive monitoring variables during the monitoring design of reinforced slopes.

在地质和岩土工程中，经常通过加固措施来稳定斜坡。加固措施的有效性对于缓解边坡破坏风险至关重要，但它会受到载荷、地质构造和岩土特性引起的地质和岩土工程不确定性的影响。面对这些不确定性，边坡加固措施往往在现场进行监测，监测设计时应谨慎选择对加固边坡稳定性安全性和可靠性敏感的监测变量，这高度依赖于工程经验和判断（即，先验知识）。如何在岩土工程监测设计中以定量和透明的方式运用先验知识仍有待探索。这可以使用基于可靠性的监测灵敏度分析来完成，由于给定监测变量的不同可能值，由于重复的逆向分析和可靠性分析，这需要大量的计算成本。本文利用子集模拟方法开发了一种基于BUS（即使用结构可靠性方法的贝叶斯更新）的高效的基于可靠性的监测灵敏度分析框架。所提出的方法使用真实的加固斜坡示例进行说明。结果表明，它基于先验知识以具有成本效益的方式量化了不同监测变量（例如，斜坡表面不同位置的位移）的可靠性敏感性。