基于复杂网络的地铁深基坑施工地表沉降风险分布及演化研究

蒋双南

摘  要

近年来，我国地铁投资建设得到迅速发展，地下车站的深基坑工程体量巨大。作为一项综合性很强的工程，地铁车站深基坑工程在施工中对基坑及周边地表、建筑物安全性进行风险评价的传统方法存在诸多不足和缺陷。在这种情况下，如何利用地铁车站基坑地表系统大量的监测数据，对深基坑从时间与空间两方面进行动态的安全风险评估，成为地铁车站基坑施工信息化及安全管理急需解决的重要问题。

本文以复杂网络理论为基础，以地铁车站深基坑地表系统为对象，分析与提出了一套基于深基坑地表系统监测数据，建立深基坑地表系统时空复杂网络模型，并定量评价深基坑系统风险及其时空演化的方法。最后基于武汉市某地铁车站深基坑工程及险情案例进行有效性与进步性的实证。

首先本文分析指出了对深基坑复杂系统风险进行有效量化评价的难点以及复杂网络理论的可行性。作为一个高度抽象的复杂系统，复杂网络模型提供了一种能描述真实系统内微观基本单元到宏观复杂结构及时空演化的方法。深基坑地表系统内的监测点及其相关关系可构成复杂网络模型，以抽象模拟深基坑施工与基坑、周边地表、建筑物多因素的关联关系，并量化深基坑复杂系统内多个元素对系统整体安全的影响，使定量、动态研究深基坑复杂系统的风险及其演化具有可行性。

其次本文基于深基坑现场监测点的实测数据建立了反映深基坑地表沉降系统的复杂网络模型。为了解决时序数据存在的时间上的不连续与噪音问题。本文运用插值法与小波分析方法对监测时序数列进行预处理，通过对数据进行补全与去噪实现了从数据到信息的转换与对施工监测数据的深入挖掘。本文基于相关系数法将时序数据转化为相关矩阵，初步建立复杂网络模型。出于对模型优化和关键信息提取的目的，本研究利用阈值法将原始相关矩阵转化为邻接矩阵，做出优化后的深基坑系统静态模型。

为了实现深基坑复杂网络模型随施工进程的动态演化，本研究利用时间观察窗口对施工全过程的时序数据进行划分，分别构建不同时间窗口下的网络模型，然后基于复杂网络模型的拓扑参数建立风险评价指标，其中基于介数中心性从空间视角量化地表沉降风险及其分布情况；基于网络结构熵的量化从时间视角量化深基坑系统整体风险，并对风险在时空上的演化进行评价，实现深基坑地表系统风险分布及演化的研究。

最后本研究基于武汉市某地铁车站深基坑案例对提出的方法进行实证检验。结果表明，该方法较传统方法能更有效地利用客观监测数据，量化地铁车站深基坑施工系统风险及其时空演化，对于加强深基坑地表系统安全风险监控，从事前预警和措施防范方面增强深基坑施工的安全管理能力具有重要意义。

关键词：地铁工程、深基坑系统风险评价、时空演化、时序数据分析、复杂网络模型

Abstract

In recent years, China's metro construction has developed rapidly. The amount of the construction of deep foundation pits of underground stations is huge. As a comprehensive project, there are many defects in the old methods of risk assessment of foundation pit,surrounding surface and building safety. In this case, it is the most important problem to be solved that how to use the large number of monitoring data collected by metro station foundation pit surface monitoring system to do the dynamic risk assessment from the aspects of time and space.

This research based on the complex network theory and took the metro station deep excavation surface system as research object. The research put forward a method to build the deep foundation pit space-time complex network model based on the surface monitoring data of deep foundation pit and had a quantitative risk evaluation of deep foundation pit and its space-time evolution. Finally, the validity and advantages of the mothed were verified by the case of a metro station construction in wuhan city.

Firstly, the research pointed out the difficulty of quantifying the risk of deep foundation pit complex system and the feasibility of complex network theory. As a highly abstract and complex system, the complex network model provides a way to describe the micro basic unit and also the macroscopic complex structure and statistical law. Monitoring points of deep foundation pit complex system may constitute a complex network model and nodes in the model can reflect the influence between deep foundation pit, the surrounding terrain, and buildings. It can also simulate the multiple impact on the risk of overall system made by different factors. It makes the quantitative and dynamic risk study of deep foundation pit become feasible.

Secondly, the complex network model of surface subsidence system of deep foundation pit is established based on the monitoring data collected by the deep foundation pit monitoring points. In order to optimize the temporal discontinuity and noise of data. The research used the interpolation method and the wavelet analysis method to have pre-treatment for monitoring data. Based on correlation coefficient method, the temporal sequence data were transformed into correlation matrix and a complex network model is established. For the purpose of model optimization and extraction of key information, this research used threshold method to transform the original correlation matrix into adjacency matrix and made the optimized complex network model of deep foundation pit system.

In order to describe the dynamic evolution of the complex network model of deep foundation pit, this research made a moving time observation window on the whole construction process of time-series data, thus built various network models under different time observation window respectively. Then based on topological parameters of complex network model the research established risk evaluation index. From the space view, the research established risk evaluation index based on the betweenness centrality to describe surface subsidence and its distribution; From the perspective of time, the research established risk evaluation index based on the network structure entropy to quantify whole risk of deep foundation pit system and evaluate the evolution of the risk.

Finally, this research based on the case of a metro station construction in wuhan city to verify the method proposed. The results showed that compared with traditional methods, the method proposed can more effectively describe the quantitative risk of metro station deep foundation pit system and its time-space distribution and evolution. The method is of great significance to strengthen the safety risk monitoring and the safety management of deep foundation pit construction.

Key words： Metro construction Deep foundation pit system Risk evaluation  Time-space distribution and evolution Analysis on time series data Complex network model