Evolution of Rail Contact Fatigue on Crossing Nose Rail Based on Long Short-Term Memory

The share of rail transport in world continues to rise. As the number trains increases, so does load on railway. rails are direct contact with loaded wheels. Therefore, it is more easily damaged. In recent years, domestic and foreign scholars have conducted in-depth research railway damage detection. weakest part track system, switches prone damage. Assessing predicting surface can improve safety operations allow for proper planning maintenance reduce capital expenditure increase operational efficiency. Under premise that functional paramount, service life rails, especially turnouts, significantly costs ensure transportation. This paper understands evolution fatigue crossing noses through long-term observation sampling turnouts. authors get images from new After image preprocessing, MPI (Magnetic Particle Imaging) divided into blocks containing local crack information. obtained texture information used regression prediction using machine-supervised learning LSTM network (Long Short-Term Memory) methods. Finally, a technique capable thoroughly evaluating wear process proposed.