Han Hong, Cai Xiaopei, Gao Liang
School of Traffic and Transportation, Beijing Jiaotong University, Beijing, 100044, China.
School of Rail Transportation, ShanDong JiaoTong University, Jinan, 250300, China.
Sci Rep. 2025 Aug 8;15(1):28993. doi: 10.1038/s41598-025-14806-2.
With the rapid development of urban rail transit, subway track structures have an increasingly serious risk of damage under high-load operations. Traditional detection methods experience several problems, such as limited coverage and a lack of real-time performance, which make it difficult to ensure operation security. Therefore, a new method for diagnosing subway track structure states based on distributed fiber sensing is proposed. First, a method for constructing a correlation model for strain monitoring data based on the optimal space window is proposed to realize the division of measuring points to reduce the computational complexity, and then, the deep generative adversarial network model with residual learning is constructed. Through spatial correlation analysis of the strain of symmetric measuring points, the Mahalanobis distance of the predicted residual is used as the diagnostic factor to realize accurate identification of the orbital structure state. Finally, practical engineering verification shows that the proposed method can effectively eliminate periodic interferences such as temperature and accurately detect local strain anomalies with a positioning error that is less than the measuring point interval (20 cm), providing reliable technical support for the intelligent monitoring and safety of subway track structures.
随着城市轨道交通的快速发展，地铁轨道结构在高负荷运营下的损坏风险日益严重。传统检测方法存在覆盖范围有限、缺乏实时性等问题，难以保障运营安全。因此，提出了一种基于分布式光纤传感的地铁轨道结构状态诊断新方法。首先，提出一种基于最优空间窗口构建应变监测数据相关模型的方法，以实现测量点划分，降低计算复杂度；然后，构建具有残差学习的深度生成对抗网络模型。通过对对称测量点应变的空间相关性分析，将预测残差的马氏距离作为诊断因子，实现轨道结构状态的准确识别。最后，实际工程验证表明，该方法能有效消除温度等周期性干扰，准确检测局部应变异常，定位误差小于测量点间距（20厘米），为地铁轨道结构的智能监测与安全提供了可靠的技术支持。
