In recent years, high-speed railway (HSR) is developing rapidly and has become a critical component of intercity transportation. As the next-generation CR450 high-speed train undergoes extensive testing, higher operation speed underscores the growing demand for HSR health inspection. Currently, track recording vehicle (TRV) with various sensing instruments is the standard and most reliable method, but the low inspection frequency limits its immediate response to sudden track faults. Therefore, advancing real-time HSR health monitoring can effectively complement existing methods and further enhance the safety of HSR system.
On May 3, a study titled “Laser Interferometry for High-Speed Railway Health Inspection using Telecom Fiber along the Line” was published in Nature Communications by Prof. Bo Wang’s team from the Department of Precision Instrument at Tsinghua University. The team developed a large-scale optical fiber interferometry platform (Fig. 1) using existing telecommunication fiber along a 12-km section of the Beijing–Guangzhou HSR. A neural network was employed to map the relation between train-induced vibration patterns and specific railway sections. By continuous vibration localization and section-by-section analysis, the 12 km railway was divided into several monitoring units, each with an A-PSD error margin that effectively indicates HSR health status.
Train-induced vibrations carry information of the train-track-bridge (TTB) system. Tailored to the long track length and strong vibration of HSR, the study employs fiber-based laser interferometry to capture the vibrations with high fidelity. A Residual Neural Network (ResNet) enhanced long short-term memory neural network is developed to localize the vibrations, enabling distributed monitoring along the railway line.
Figure 1. The HSR platform under measurement
The study proposed average power spectral density (A-PSD) of train-induced vibration as an indicator for HSR health monitoring (Fig. 2). A-PSD error margins were established for each railway section and validated through over 14 months of continuous monitoring. During that time, there were earthquakes, rainstorms, and seasonal changes. However, the A-PSD of railway sections remained stable within the error margins (Fig. 2c–f), consistent with the healthy railway condition.
Figure 2. The average power spectrum density (A-PSD) analysis of 4 typical HSR infrastructures
During the monitoring period, the HSR Engineering Section of the Beijing Railway Bureau carried out maintenance on a track section with minor creep deformation. Analysis showed that A-PSD of this section exceeded the error margin before maintenance, and returned within the margin afterward (Fig. 3). Similar cases further demonstrated the sensitivity of the A-PSD indicator to track faults.
Figure 3. Sensing results of a section with creep deformation before and after maintenance.
The proposed real-time health monitoring method enables long-distance distributed sensing without the need for special sensors and significantly narrows the HSR inspection period. Besides, during the nighttime maintenance window when high-speed trains stop and the environment is quiet, signals such as seismic waves and vibrations from coal trains were detected using the same experimental platform. This study demonstrates that a distributed sensing system based on existing fiber networks can act as a real-time supplementary tool for HSR health inspection and has the potential to establish a large sensing network.
Prof. Bo Wang is the corresponding author of this work. Guan Wang (2024 Ph.D. graduate, Tsinghua University) and Dongqi Song (2024 Ph.D. student, Tsinghua University) are co-first authors. Special appreciation is conveyed to the High-Speed Railway Engineering Section and Communication Section of the Beijing Railway Bureau, and Gongjian Hengye Communication Technology Corporation Limited for arranging the experimental platform and providing track recording vehicle inspection data. This work was supported by National Natural Science Foundation of China (62171249), National Key R&D Program of China (2021YFA1402102), and Tsinghua Initiative Scientific Research Program.
Link to paper:
https://www.nature.com/articles/s41467-025-59507-6
Editor: Li Han
