Abstract The application of a large-scale slurry-balance shield machine in super-large tunnel excavation (diameter >14 m) is much more difficult and risky than the common shield tunneling. The control of shield machine becomes a complex and difficult problem under frequently encountered unforeseen geological conditions. The operators of shield machine need to recognize the influential parameters for excavation stability, and then primarily adjust these influential parameters which can be controlled in shield machine tunneling. Aiming to find out the influential parameters for excavation stability, this study proposes a framework of global sensitivity analysis (GSA) by (i) employing three different meta-models namely polynomial chaos expansion (PCE), radial basis function (RBF), and support vector machine (SVM) methods to mimic the mapping relation between input and output parameters, and (ii) adopting three kinds of GSA methods namely Morris elementary effects, Sobol method, Expand Fourier amplitude sensitivity test (EFAST) algorithm to classify the parameters as important or unimportant (insensitive). The results show that the synchronous grouting pressure, advancing speed, and penetration are influential parameters to bubble chamber pressure, whereas the cutter rotation speed, cutter torque, penetration, and actual volume excavation are influential parameters to deviation angle. Here, we pay close attention to the most influential parameters that must be prioritized for parameter control, which can help the administrators optimize their management scheme of influential parameters during tunnel excavation. The follow-up research can focus on how to develop and explore additional particular details about the GSA method or some other data mining technologies for optimizing the management scheme of shield tunneling.

中文翻译：

---

地铁隧道开挖稳定性影响参数的全局敏感性分析

摘要 大型泥浆平衡盾构机在超大型隧道开挖（直径>14 m）中的应用比普通盾构掘进难度大、风险大。在经常遇到的不可预见的地质条件下，盾构机的控制成为一个复杂而困难的问题。盾构机操作人员需要识别影响开挖稳定性的参数，然后对这些影响参数进行初步调整，这些参数在盾构机掘进中是可以控制的。为了找出影响开挖稳定性的参数，本研究通过（i）采用三种不同的元模型，即多项式混沌展开（PCE）、径向基函数（RBF）、和支持向量机（SVM）方法来模拟输入和输出参数之间的映射关系，以及（ii）采用三种GSA方法即Morris基本效应，Sobol方法，扩展傅立叶幅度灵敏度测试（EFAST）算法对参数进行分类重要或不重要（不敏感）。结果表明，同步注浆压力、推进速度和穿透是影响泡腔压力的参数，而刀具转速、刀具扭矩、穿透和实际开挖体积是影响偏角的参数。在这里，我们密切关注必须优先进行参数控制的最有影响的参数，这可以帮助管理员在隧道开挖过程中优化其影响参数的管理方案。