Projeto otimizado para minimização de perdas em um conversor bidirecional aplicado a sistemas autônomos de iluminação

Abstract

This work presents the development of an standalone street lighting system based on solar energy and light emitting diodes (LEDs). The system must be able to charge a battery bank during the day from a photovoltaic source and supply an LED load during the night. A bidirectional converter based on gallium nitride (GaN) transistors is used in order to achieve a high-performance system and, thereby, to optimize the size of the battery bank and photovoltaic array to maximize the system autonomy or reduce costs, depending on the application goals. A complete and optimized methodology was developed for the design of the elements of this standalone lighting system. This methodology is based on a discrete search algorithm that evaluates a set of commercially available elements to determine the best combination of components and the best operating point of the circuit, in order to minimize energy losses over the system operation period. This document initially presents a bibliographic review of theoretical aspects related to autonomous lighting systems, such as the photovoltaic source, batteries, LED load, converter topologies and commercially available solutions. A review of GaN switches is also presented, covering their history, operating principles, features and state of the art in terms of commercially available devices. To experimentally validate the proposed topology, a 100 W prototype with arbitrary switching frequencies was developed and evaluated. In battery charger mode, the circuit operates at 250 kHz and has efficiencies greater than 92% over the entire power range with a maximum value of 97.3%. In LED driver mode, the chosen operating frequency was 345 kHz and a maximum efficiency of 95.8% was achieved. To validate the proposed design methodology, four converters with power ratings of 30 W in driver mode and 150 W in charger mode were implemented and tested at different operating points. Thus, it was proved that, considering a set of components, the methodology is able to determine the best operating point to maximize the efficiency, and, considering different sets of components, determining the best in terms of energy loss reduction to maximize system autonomy