Deep PCA-Based Incipient Fault Diagnosis and Diagnosability Analysis of High-Speed Railway Traction System via FNR Enhancement

In recent years, the data-driven based FDD (Fault Detection and Diagnosis) of high-speed train electric traction systems has made rapid progress, as safe operation system is closely related to reliability stability trains. The internal complexity external environment mean that fault diagnosis faces great challenges. this paper, a wavelet transform-based FNR Noise Ratio) enhancement realised highlight incipient information Deep PCA (Principal Component Analysis)-based diagnosability analysis framework proposed. First, scheme for enhancement-based data preprocessing with selection intelligent decomposition levels optimal noise threshold Second, technology on continuous transform proposed from perspective energy. Further, Deep-PCA detectability isolatability are provided via geometric descriptions. Finally, experiments TDCS-FIB (Traction Drive Control System–Fault Injection Benchmark) platform fully demonstrate effectiveness method in paper.