Deep learning-based fault diagnostic network of high-speed train secondary suspension systems for immunity to track irregularities and wheel wear

Abstract Fault detection and isolation of high-speed train suspension systems is critical importance to guarantee running safety. Firstly, the existing methods concerning fault or are briefly reviewed divided into two categories, i.e., model-based data-driven approaches. The advantages disadvantages these categories approaches summarized. Secondly, a 1D convolution network-based diagnostic method for designed. To improve robustness method, Gaussian white noise strategy (GWN-strategy) immunity track irregularities an edge sample training (EST-strategy) wheel wear proposed. whole network called GWN-EST-1DCNN method. Thirdly, show performance this multibody dynamics simulation model built generate lateral acceleration bogie frame corresponding different irregularities, profiles, secondary faults. simulated signals then inputted network, results correctness superiority Finally, 1DCNN further validated using tracking data CRH3 on railway line.