Online Condition Monitoring of Three‐Phase Filter Capacitors for Vehicular Auxiliary Converter

ABSTRACT

Three‐phase filter capacitors are key components in urban rail vehicle auxiliary converters. They filter the output ripple voltage and ensure a stable power supply for onboard equipment. Their performance degradation poses significant safety risks. This paper proposes an online capacitance estimation method for condition monitoring. The method uses existing voltage and current sensors of the converter without adding hardware costs. To address the challenges of the difficult separate estimation of three‐phase capacitance values and high harmonic content in the data, an improved estimation model is developed. This model suppresses switching harmonics while reducing the impact of sensor DC bias. An improved Whale Optimisation Algorithm with a deterministic fitness‐level‐based update mechanism is adopted to achieve accurate capacitance estimation through search. Furthermore, comprehensive robustness analysis reveals that a minimum sensor accuracy of Class 0.5 is required to provide sufficient diagnostic margin for a typical 10% early warning threshold. Experimental validation on a subway auxiliary converter shows that the estimation error of this method is below 3.5% compared to measured values. The proposed method demonstrates better stability than other algorithms and effectively reduces misdiagnosis risks caused by measurement noise and harmonic distortion.