Advances in Train and Rail Monitoring with DAS

Abstract

DAS resembles thousands of microphones along the sensing fiber. In railway applications, only a small fraction of all DAS data carries information about the train-rail interaction. Transforming the data to a rail view or train view simplifies analysis and reveals valuable rail and train properties.

Fiber Optic Train Monitoring with Distributed Acoustic Sensing: Conventional and Neural Network Data Analysis

“…While there are individual time samples where the correct number of bogies are visible in the measured intensity along the fiber, in general, the signal is too noisy to determine the correct number from a single sample. Therefore, we use the train position as obtained in the previous step to shift all DAS time samples by x center (t) to create a 'train-view' diagram [6] in which the train and bogie peaks are always localized at the same position (see Figure 2c,d), The train-view diagram corresponds to the DAS signal after transformation into a reference frame moving with the train, and one can directly see the train length of ca. 200 m over time in the diagram, The 'train-view' may potentially be useful for detecting train defects via train signals that change over time.…”

“…Thereby DAS may help to increase railway capacity by enabling efficient train driving and disposition, but more importantly by enabling moving block operation, if the safety-critical performance of DAS can be established. Further, not only can the train position and velocity be monitored, but, in principle, also faults of the train such as flat wheels, and wear in the train track can be detected due to their characteristic acoustic signatures [6].…”

“…While DAS is not yet routinely used in train monitoring, it offers several advantages over established techniques such as a track circuit or wheel counters, In contrast to wheel counters at a limited number of positions, fiber optic sensing is a truly distributed measurement and monitors the position of the train at all times. With its sensing range of tens or even up to 100 km [6] it eliminates the necessity of power and data cables because a single telecom fiber that often is already installed next to tracks can serve both, as sensor and data line, The DAS fiber is also immune to electromagnetic interference or lightning strike, In comparison to global navigation satellite systems, fiber sensing has the advantage of providing signals in tunnels and it does not need wireless communication between trains and base stations. Beyond train monitoring, events such as trespassing or cable theft can be detected with DAS.…”

“…Fiber optic sensing techniques such strain sensors based on Brillouin scattering [7,8], fiber Bragg gratings and fiber interferometers [9,10] have been demonstrated in railway applications. However, most activities in the field of DAS have centered on Raleigh backscatter based (C-OTDR) systems [6,7,11,12]. Two main DAS techniques can be distinguished.…”

“…Simpler DAS systems have been successfully used in a range of publications [4,5,7] and are also used in this work. Recent true-phase systems have shown significantly better signal-to-noise ratios as well as a long sensing range beyond 80 km [6]. A common problem of both DAS and true-phase DAS systems is the large amount of raw data that is acquired and must be processed precisely and quickly enough to extract the features of interest from the noisy data.…”