基于时空演化模型的地铁隧道施工地表沉降安全风险预警

Current control standard for ground settlement cannot meet the urgent requirement of early warning for safety risk of metro tunneling construction in urban area with various and complex environment. This study, based on methodology of systems engineering, investigates the spatio-temporal pattern of ground settlement from 3 million items of monitoring data and documents. A spatio-temporal model of ground settlement induced by twin-tunnel is proposed based on Logistic model and mixed Gaussian model. The safety risk factors are extracted from the model parameters which are deduced by curve-fitting process using Levenberg-Marquardt algorithm and Nelder–Mead algorithm. The patterns of safety risk factors in various geology condition and construction context are analyzed using multivariate statistical method. The effect of ground settlement on adjacent buildings are evaluated through the distribution of ground settlement and the patterns of building damages induced by differential settlement. The evaluation of settlement affordability of the adjacent buildings are proposed based on the utility and stiffness of the building. Finally, a model of safety assessment for ground settlement is proposed based on the aforementioned 2 evaluation methods, and it is used for the early-warning system. The result will be implemented in a number of metro construction projects, which are invested over billions, to improve the ability of safety early warning in construction process.

现行固定的地表沉降30mm控制标准不能满足多样性复杂城市环境下地铁隧道施工安全预警的迫切需求。本项目基于系统工程理论与方法，从300余万条监测数据及工程资料中挖掘地表沉降时空演化规律；构建安全预警模型。创新点包括：基于Logistic模型和混合高斯模型提出地铁双隧道施工地表沉降时空演化模型；综合运用Levenberg-Marquardt和Nelder–Mead算法拟合并提炼模型中代表沉降影响范围和塌陷大小的风险因子；运用多元统计方法挖掘风险因子在不同水文地质条件和施工情境下的规律；基于沉降分布规律和建筑物因差异沉降造成损坏的模式提出沉降对建筑物影响大小的评估方法；基于建筑物重要性及刚度抗变形能力提出建筑物对沉降可承受程度的评估方法；最后综合二者提出地表沉降安全风险评估模型，并根据风险分级结果进行施工安全预警。研究成果将在投资数百亿的全国多个地铁建设项目中推广应用，提高施工安全预警能力。

大规模的地铁建设缓解了城市交通压力，却激发了城市岩土工程的安全风险。岩土工程问题的复杂性促使工程领域迫切的需要一套系统化的方法进行风险综合评价。.本课题主要针对地铁隧道工程施工所引起的地表沉降开展研究，构建了面向隧道施工地表沉降分析的时空情境多维数据仓库，将零散的多源异构数据整合成集成情境的价值信息；运用离散小波变换技术滤除了日沉降中的监测误差，综合提升了数据的质量和信息的价值；从时间、空间、情境等多个视角挖掘了地表沉降规律，提炼了能够指示地表沉降安全风险的相关指标；基于大量的数据拟合结果，发现了沉降槽宽度的时不变性及其与地质水文分布的相关性；综合运用前文得到的地表沉降时、空、情境规律，建立了地表沉降安全风险预警机制以指导工程实践。.在研究过程中还发现：地铁施工安全风险除了受地表沉降所影响，还容易受其它因素所诱发。因此本课题还针对单个风险事件，以专家经验知识建立风险评价规则决策表，基于模糊粗糙集理论方法化简论域空间，从而得到单个风险事件的模糊评价规则。采用该方法得到的风险评价规则是对专家经验知识的高度抽象。与精确规则相比，模糊规则能最大限度地减少规则数量，同时提供给决策者更多有价值的信息以支持科学决策。该方法是为实现地铁施工安全风险识别与评价自动化而开发，能够完成安全风险评价知识库的构建，是地铁施工安全风险自动识别系统的理论基础，能够实现无专家参与的风险识别与评价，有利于提高地铁施工安全管理的效率与质量。