Projeto otimizado de conversores cc-cc sintetizados pela teoria de grafos e algoritmo genético aplicando parâmetros avaliativos de performance

Abstract

In the search for more efficient dc-dc converters, designers seek to use mathematical methods for optimized designs in order to minimize losses and volume. Scanning and genetic algorithms are the methods used to determine the best operating points in the design of inductors, semiconductor and diode choices, for example. However, the choice of topology based on some criteria is as important as its optimized design. Thus, this dissertation has as main focus: to develop a converter synthesis procedure and apply an evaluation metric in the synthesized topologies. The work initially focuses on developing an algorithm based on methodologies available in the literature, which use graph theory for the synthesis of dc-dc converters. The input of this algorithm is only the number of switches, active and passive, in the topology. This synthesis is performed with the application of graph theory, which also provides a simple method to obtain the operating values of the converter elements in each operating interval. Afterwards, the MSEF and processed power metrics are applied to the synthesized topologies in order to obtain a value that allows classifying it to know which one will have the best performance. Finally, all these steps are summarized in an example of a search for a step-up converter with up to four keys for an input voltage of 100 V, an output voltage of 120 V and an output power of 100 W. To corroborate the results found by the algorithm developed, the topologies were simulated in the PSIM 2021 software by Powersim using the loss module present in the program.