Supplementary information for 'Intelligent Decision Method for Stability Assessment of Shield Tunnel Based on Multi-objective Data Mining' from Intelligent decision method for stability assessment of shield tunnel based on multi-objective data mining

Due to the improper operation of the shield construction process and unknown geological survey, shield construction faces many risks in passing through the complex strata, among which the excavation face instability is the most serious disaster accident. To solve these issues, this research focused on the limit support pressure and the excavation face stability in the soil when crossing the Yangtze River. First, the analytical formula of limit support pressure of the excavation face is established through the wedge model. The support safety coefficient is given to assess the excavation face stability quantitatively. Then the rough set algorithm was used to analyze the sensitivity of each index to establish the reduced evaluation index system for the excavation face stability. The BP neural network was used to train the learning data, and a neural network evaluation model with a prediction error of 5.7675 × 10⁻⁴ was established. The prediction performance of BP was verified by comparing the TOPSIS prediction model and the cloud model. The evaluation method proposed in this paper provides an essential reference for evaluating the underwater shield tunnel excavation face stability.This article is part of the theme issue ‘Artificial intelligence in failure analysis of transportation infrastructure and materials’.

由于盾构施工工艺操作不当以及地质勘察不详，盾构隧道穿越复杂地层时面临诸多风险，其中开挖面失稳是最为严重的灾害事故。针对上述问题，本研究以穿越长江的盾构隧道为研究对象，聚焦其土体开挖面的极限支护压力与开挖面稳定性问题。首先，通过楔形体模型推导建立开挖面极限支护压力的解析公式，并引入支护安全系数以定量评价开挖面稳定性。随后采用粗糙集（rough set）算法对各指标进行敏感性分析，构建简化的开挖面稳定性评价指标体系。利用BP神经网络对训练样本开展学习训练，构建了预测误差为5.7675×10⁻⁴的神经网络评价模型。通过与逼近理想解排序法（TOPSIS）预测模型及云模型对比，验证了所提BP神经网络模型的预测性能。本研究提出的评价方法可为水下盾构隧道开挖面稳定性评价提供重要参考。本文属于“交通基础设施与材料失效分析中的人工智能”主题专栏的一部分。