Design of ANN Based Controller for Battery Charging Using Synchronous Buck Converter

Abstract

This article presents the design of an artificial neural network (ANN) controlled one-and-a-half kilowatt capacity synchronous buck converter(SB) that can be used to charge lead-acid batteries. The most commonly used lead-acid battery charger requires a buck converter with a constant current voltage (CC-CV) algorithm to charge batteries. The conventional charges have various disadvantages; the traditional buck converter losses are negotiated by replacing the diode with MOSFET to improve efficiency. The conventional PI-based control algorithm design is complex and entirely depends on the system dynamics. The proposed data-driven ANN controller simplifies the control logic for battery charging. The proposed scheme has been tested on a test bed wherein the hardware components like a solar emulator, SB, controller and a 12V, 42Ah lead-acid battery has been used. The pulses from ANN-based control logic have been given from the micro-controller to the gate driver circuit that drives both MOSFETs’. The complete ANN algorithm has been performed in standalone mode by deploying the algorithm entirely onto a micro-controller. It is also observed that 95% of efficiency has been achieved with the proposed scheme.