Development of a novel real-time AE source localisation technique using ANN for health monitoring of rail section: an experimental study

Abstract

Almost all countries are using railways to transport passengers and freight. Due to heavy load and environmental exposure, these rail sections are prone to damage. Thus, structural health monitoring of the rail section is an essential topic of research to ensure safety and predict the remaining useful life of the track. Consequently, the detection and localisation of damages/cracks in rail section are crucial. It also helps in taking corrective action in time. In this context, the acoustic emission (AE) technique is found to be a powerful non-destructive testing technique which has great potential to localise cracks in rail sections. Actually, the initiation or growth of a damage/crack in a structure emits AE waves, which may be captured using AE sensors mounted on different locations of the structure. The damage itself acts as an AE source. Subsequently, in this study, a novel approach is developed to localise the AE source in a rail section using the artificial neural network (ANN). The ANN model is based on the experimental crack signals captured through a single AE sensor by placing it at different regions of the rail section. For each position of the sensor, cracks are simulated using pencil lead break at different locations along the length and depth of the rail section. It is observed that the developed ANN model predicts the location of the AE source with better accuracy in comparison to the existing localisation methods. The minimum error is found to be less than 1%. Therefore, it can be concluded that the developed ANN model is a suitable option for predicting potential crack locations during real-time monitoring of rail sections.