Considering the weak noise resistance and low identification efficiency of traditional bridge damage identification methods, a data-driven approach based on synchronous statistical moment theory and vehicle–bridge interaction vibration theory is proposed. This method involves two main steps. First, a two-axle test vehicle is used to collect acceleration response signals synchronously from adjacent designated measurement points while stationary. This operation is repeated to calculate the second-order statistical moment curvature (SOSMC) difference of entire bridge points corresponding signals in different states. By comparing with the reference value, the preliminary damage location of the bridge can be obtained. Second, the first-order modal shape curve is constructed using the second-order statistical moment (SOSM). The refined identification of bridge damage is then based on an improved direct stiffness back calculation of the bridge's stiffness. This article proposes the synchronization theory for the first time and combines it with the statistical moment clustering method, forming an innovative approach to obtaining structural vibration modes. The effectiveness of this method has been well validated through numerical simulations with different parameters and on-site bridge tests. The research results indicate that SOSMC indicators have better noise resistance and higher recognition efficiency in identifying damage locations, compared to modal curvature and flexibility curvature indicators. Additionally, compared to transfer rate and random subspace methods, the SOSM method results in smaller error and higher identification efficiency.

中文翻译：

---

基于车桥相互作用同步统计矩理论的桥梁损伤识别


针对传统桥梁损伤识别方法抗噪声能力弱、识别效率低的问题，提出一种基于同步统计矩理论和车桥相互作用振动理论的数据驱动方法。该方法涉及两个主要步骤。首先，使用两轴测试车在静止状态下从相邻的指定测量点同步采集加速度响应信号。重复该操作，计算不同状态下整个桥点对应信号的二阶统计弯矩曲率（SOSMC）差异。通过与参考值进行比较，即可得出桥梁的初步损伤位置。其次，使用二阶统计矩（SOSM）构建一阶模态曲线。然后，桥梁损伤的精确识别基于桥梁刚度的改进直接刚度反演计算。本文首次提出同步理论，并将其与统计矩聚类方法相结合，形成了一种获取结构振动模态的创新方法。通过不同参数的数值模拟和现场桥梁试验，很好地验证了该方法的有效性。研究结果表明，与模态曲率和柔度曲率指标相比，SOSMC指标在识别损伤位置方面具有更好的抗噪声能力和更高的识别效率。此外，与传输速率和随机子空间方法相比，SOSM方法的误差更小，识别效率更高。