Smart multioutput fast charger for electric vehicles using deep learning controllers

Abstract

The increasing adoption of electric vehicles has intensified the demand for efficient, high-performance charging solutions to address long charging times and range limitations. This paper presents the development of a smart multioutput Direct-current fast charger leveraging deep learning techniques to enhance electric vehicle battery charging capabilities. The proposed Direct-current fast charger integrates a front-end converter to transform grid AC voltage and current into DC, managed by three deep learning controllers. The charger incorporates a high-frequency inverter, a high-frequency isolation transformer, and a diode bridge rectifier for AC-DC conversion. The inverter gate drive system is optimized by using a deep learning controller to regulate output performance. Two types of deep learning controllers such as custom neural network and neural net fitting are implemented to minimize settling time, overshoot, and harmonics, ensuring ripple-free, smooth DC bus voltage, current, and battery charging current. Comparative analysis reveals that the custom neural network-based Direct-current fast charger is superior to the neural net fitting in terms of settling time, overshoot, complexity, accuracy, and efficiency. MATLAB & Simulink simulations validate the effectiveness of the proposed system, demonstrating its potential for improving electric vehicle charging performance.