Machine learning-based optimization of resonant LLC converters for improved power quality in electric vehicle chargers

Abstract

Abstract When it comes to developing a strategy for EV charging networks, improving the utilization and planning of charging stations becomes critical with the fast growing market for electric cars. Improvement of power quality has the largest effect on the performance and dependability of EVs essential for their large-scale implementation. Issues like low power quality and very high THD present in conventional systems are addressed in this research through a two-stage EV charger consisting of an active power factor correction (PFC) front-end and an LLC resonant DC-DC converter, achieving an input power factor of 0.98, THD of 5% and efficiency of 95%. Applying Support Vector Regression (SVR) as the type of a supervised machine learning algorithm, we employ a predictive maintenance model which processes real-time information and decreases downtimes by a third. Based on the identified shortcomings of ordinary diode bridge rectifiers, this approach extends these advantages and enhances the functionality of EV chargers dramatically.