摘 要

随着我国城市不断的发展，很多城市都在修建地铁。但是由于地铁施工会受到地质条件、周边环境等多种因素的影响，地铁车站基坑施工中存在众多风险，地铁施工事故也经常发生。由此，将信息化监测技术用于地铁施工过程中，研究基坑工程的监测数据，探索基坑在施工过程中的变形规律，对于预报工程风险，控制工程事故的发生具有重要意义。但是在实际工程中，原始监测数据却并没有得到有效的利用与分析。首先，无法直接通过监测数据的超限，来判断当前施工状态是否存在安全风险。其次，目前仅采用常规的方法，针对某一项监测类型的数据进行统计。而不同监测类型的数据之间可能存在的联系，以及监测数据与风险事件之间的关联并未得到挖掘。事实上，施工风险的前兆一定会反应在监测数据的异常上，但是如何准确挖掘并获取他们之间的关系，如何有效的利用已有的监测数据，发挥原始数据的优势还急需研究。在此背景下，本文提出了基于地铁施工的原始监测数据，采用数据挖掘技术和复杂网络的理论知识，发掘监测数据特征和工程风险事件之间的联系。

本文基于明挖地铁车站的施工安全的原始监测数据，分别构建关联规则挖掘算法模型和复杂网络模型，研究不同风险事件之间的耦合关系。首先，利用基于Apriori的关联规则算法挖掘出不同监测类型异常情况之间的频繁项集，获得有助于地铁施工安全风险管理的关联规则信息，从而提出更准确的地铁车站基坑工程施工风险监测项目，能帮助决策者进行科学决策，有利于减少工程事故的发生。接着，利用挖掘出来的风险事件之间的关联规则构建施工风险事件之间的复杂网络模型，从而发现了类似于地表沉降、建筑物沉降及混凝土支撑轴力这些不同风险事件之间的耦合关系。并通过复杂网络的统计特征来获取关键节点的风险类型。对处于关键节点位置的风险进行有效控制能减缓风险事件在整个耦合网路中的传播效率。最后，通过将上面的模型运用到具体地铁施工案例中，验证其有效性。

关键词：地铁基坑工程 风险耦合 关联规则挖掘 复杂网络

Abstract

With the development of China's economy, more and more cities build metro. Due to the complicated construction geological conditions, metro foundation pit involves many inaccuracies, which suffers lots of risk during the construction period. In view of the real problem of metro construction accidents, the information of monitoring equipment is used to grasp the rules of construction data.

However, the original monitoring data has not been effectively utilized and analyzed in fact. First of all, the abnormal monitoring data can not directly determine whether the current construction situation has risks. Secondly, at present, only a certain type of monitoring data can be analyzed statistically by the conventional methods.Different types of monitoring data and the association between the risk events have not been excavated.In fact, the precursor construction risk can be reacted in the abnormal monitoring data. How to effectively use the monitoring data and get accurate relationship between them is urgent. Therefore, it is necessary to use the data mining technology to find out the link between the monitoring data and the risk events, which is based on the monitoring data collected by the system.

In this paper, the association rules mining and complex network analysis are applied to analyze the risk monitoring data. We use the ARM to dig out frequent itemsets and association rules between different monitoring abnormal situations. In addition, we introduce the complex network theory to analyze the network characteristics of the coupling relationship between other monitoring types that lead to the monitoring target anomaly. Then, the frequent itemsets and association rules between the risk monitoring projects are excavated. So as to find out the coupling relationship between surface settlement, building settlement and concrete support axial force between. The key monitoring risk events which conforms to the metro construction safety management are obtained. The effective control of the key risk can reduce the transmission efficiency of the event in the whole coupling network. Finally, the effectiveness of the above two models in the construction of subway station foundation pit is analyzed by the actual case.

Through the research, this paper reveals the coupling of risk in the construction of subway station excavation from the qualitative and quantitative point of view.Related conclusions can not only provide a scientific and rational support for the prevention of construction accidents, but also improve the metro construction safety early alert mechanism.

Keywords:Metro foundation pits; Risk coupling; Association rule mining; Complex network