Deep Learning-Based Acoustic Emission Scheme for Rail Crack Monitoring

This research proposes a single-sensor acoustic emission (AE) scheme for detection and localization of crack in steel rail (rail head, rail web, and rail foot) under load. In the operation, AE signals were captured by the AE sensor and converted into digital signal data by AE data acquisition module. The digital data were used total variation denoising (TVD) algorithm to remove ambient and wheel/rail contact noises, and the denoised data were processed and classified to localize cracks in the steel rail using a deep learning algorithmic model. The AE signals of pencil lead break at the head, web, and foot of steel rail were used to train (80 % of the input data) and test (20%) the algorithmic model. In training and testing the algorithm, the AE signals were divided into two groupings (150 and 300 AE signals) and the classification accuracy compared. The deep learning-based AE scheme was also implemented on-site to detect cracks in the steel rail. The total accuracy under the first and second groupings were 86.6 % and 96.6 %. The novelty of this research lies in the use of single AE sensor and AE signal-driven deep learning algorithm to detect and localize cracks in the steel rail, unlike conventional AE crack-localization technology which relies on two or more sensors and human interpretation.