盾构法施工风险的多态贝叶斯网络模型分析

Risk Analysis of Shield Construction by Polymorphic Bayesian Networks Model

摘要:

地铁隧道具有投资规模大、风险因素多、社会影响大等特点，这决定了在施工过程中进行风险管理的必要性。鉴于风险的模糊性，很难用一个具体的概率值或概率区间来描述其发生概率，由此提出了用多个概率区间来分析盾构法施工过程中的风险。以划分地铁及地下工程风险概率等级的五个概率区间为度量标准，设计出了风险因素概率等级调查表，通过专家调查法可统计出盾构法施工风险因素概率等级的概率分布；基于调查表，采用贝叶斯网络建立了分析风险事故概率等级概率分布的多态系统。利用该系统对一个地铁隧道工程进行风险评估，得到了该工程的风险概率等级，并分析出相应风险概率等级下进出洞洞口密封效果不好、中轴线偏离过大造成管片破裂以及管片密封材料损伤、破坏等因素为重点控制因素。

Abstract:

Subway tunnel is characteristic of large-scale investment, many risk factors, large social impact, etc., which determines the need of risk management during the construction. As the fuzziness of risk, it is difficult to use an accurate probability or a probability interval to describe risks. So this paper suggests that shield excavation risks be analyzed by multiple probability intervals. Based on the five probability intervals which divide risk probabilities into five ranks in the subway and underground engineering, risk factor questionnaire has been designed, and the probability distributions of probability ranks about shield construction risk factors could be figured out through expert survey method. According to the questionnaire, Bayesian Networks is applied to establish a multi-state system to analyze the probability distributions. The multi-state system is used in the risk assessment of a subway tunnel, the probability rank of the project is figured out and at the corresponding probability rank，such risk factors as poor seal in the portals, axis deviation which causes segment failure and the damage, destruction of sealing material are considered as the key control objects.