Performance Trade-offs of Machine Learning Hyperparameters in On-board Charger’s Power Factor Correction Fault Classification

Abstract

As EV adoption grows, reliable On-Board Chargers (OBCs) are essential for safe and efficient charging. Diagnosing OBC faults is challenging due to varied fault types. Traditional rule-based methods struggle with modern systems, prompting the use of machine learning. Our study shows that applying performance trade-off about various machine learning model, significantly shows fault classification F1 score while ensuring real-time performance in PFC fault diagnostics in OBCs.