A novel experimental/numerical hybrid methodology for the assessment of railway-induced ground-borne vibration in buildings based on experimental measurements in the soil surface is proposed in this paper. This methodology has been specifically designed for the prediction of railway-induced vibration in buildings to be constructed close to an operative railway infrastructure, although it can be applied for other types of vibration sources. The model of the incident wave field induced by the railway infrastructure consists of a set of virtual forces applied in the soil, which would be obtained from vibration experimental measurements in the surface of the ground where the building will be constructed. These virtual forces can be subsequently applied to a model of the building-soil system to obtain a prediction of the vibration levels that will be induced by the existing railway infrastructure to the studied building. In the present work, this methodology is theoretically defined and it is numerically validated for two-dimensional and two-and-a-half-dimensional cases. To numerically test the methodology, the measured ground surface responses are replaced by simulated ones obtained in a set of points called collocation points. In this context, a parametric study has been developed with the aim of finding out a robust criterion for the application of the present methodology with respect to the amount and location of the collocation points (representing vibration sensors) and virtual forces. It is found that the distance between virtual sources should be smaller than the S-wave wavelength of the upper soil layer corresponding to the highest frequency of the frequency range of interest to ensure the reliability of the methodology. Moreover, the proposed method is found to be insignificantly affected by the building-tunnel dynamic coupling for building-tunnel distances above 20 m. The proposed hybrid model would simplify the usual numerical prediction approach commonly adopted for dealing in detail with these problems, since a model of the railway infrastructure is no longer required. Moreover, it would reduce the uncertainty of the prediction due to the use of experimental measurements of the particular site to be studied. In addition, it would provide a higher accuracy and flexibility than empirical models based on experimental transmissibility functions between the soil surface and the building.

本文提出了一种新的实验/数值混合方法，用于评估基于土壤表面实验测量的建筑物中铁路引起的地传振动。该方法专为预测在运行中的铁路基础设施附近建造的建筑物中的铁路引起的振动而设计，尽管它可以应用于其他类型的振动源。由铁路基础设施引起的入射波场模型由一组施加在土壤中的虚拟力组成，这些力可以从将要建造建筑物的地面的振动实验测量中获得。这些虚拟力随后可以应用于建筑-土壤系统的模型，以获得由现有铁路基础设施对研究建筑引起的振动水平的预测。在目前的工作中，这种方法在理论上得到了定义，并在二维和二维半的情况下进行了数值验证。为了对该方法进行数值测试，将测量的地表响应替换为在一组称为搭配点的点中获得的模拟响应。在这种情况下，已经开发了一项参数研究，目的是找出一个稳健的标准，用于在搭配点（代表振动传感器）和虚拟力的数量和位置方面应用本方法。发现虚拟源之间的距离应小于感兴趣频率范围的最高频率对应的上层土层的S波波长，以确保方法的可靠性。此外，发现所提出的方法在建筑-隧道距离超过 20 m 时不受建筑-隧道动态耦合的影响。提议的混合模型将简化通常用于详细处理这些问题的常用数值预测方法，因为不再需要铁路基础设施模型。此外，由于使用要研究的特定地点的实验测量，它会降低预测的不确定性。此外，