Estimation of the switching losses in DC-DC boost converters by various machine learning methods

Abstract

DC-DC converter circuits are topologies commonly used in power electronics applications such as renewable energy sources, electric vehicles, uninterruptible power supplies and DC transmission systems. The most important factors affecting efficiency and thus performance is the choice of the power semiconductor switching element as well as the circuit design and types of these topologies. In this context, power semiconductors are determined according to the switching frequency and current-voltage parameters. However, due to other operating modes of the circuit and load variation during the power conversion, the losses of the switching elements do not remain constant. In this study, a parametric simulation is performed in a conventional DC-DC boost converter circuit using the parameters related to the Insulated-Gate Bipolar Transistor (IGBT) power-switching element selected at a certain current-voltage capacity. These parameters are switching frequency, duty ratio and load change of the converter. Finally, using the data obtained, the loss of switching losses are estimated by the Multilayer Perceptron (MLP), Support Vector Machine (SVM), K- Nearest Neighbors (KNN) and Random Forest (RF) Machine Learning (ML) techniques.