Health State Assessment of Metro Train Traction Converter under Variable Operating Conditions Based on Feature Extraction

The traction inverter is a key component of subway trains, affecting the safe operation of the vehicle. Frequent starting and stopping and high-intensity operation subject it to periodic electrical and thermal stress, resulting in the attenuation of system performance. The nonlinear variable operating condition degradation mode also makes the performance degradation analysis more complicated. This paper proposes a variable operating condition traction inverter health assessment method based on feature extraction. First, a mathematical model of the traction inverter system is established to analyze the key components that have the greatest impact on the system; then, a degradation model of the key components is established, and degradation analysis under variable operating conditions is performed, which is verified by simulation; finally, key feature parameters are extracted from multiple dimensions, and machine learning is combined to achieve accurate assessment of the health status.