Detecção de correntes de inrush em transformadores de potência: análise comparativa e uma nova proposta baseada em Transformada Wavelet Sincronizada e Bag of Features

Also known as inrush current, magnetizing current arises during transformer energization. Due to its transient characteristics and high amplitude, this current can cause unwanted operations of protection relays, resulting in incorrect tripping. The main objective of this work is to compare different...

Full description

Bibliographic Details
Author: Durlo, Luccas dos Santos
Format: master thesis
Status:Published version
Publication Date:2025
Country:Brasil
Institution:Universidade Federal do Pampa (UNIPAMPA)
Repository:Repositório Institucional da UNIPAMPA
Language:Portuguese
OAI Identifier:oai:repositorio.unipampa.edu.br:123456789/10514
Online Access:https://repositorio.unipampa.edu.br/handle/123456789/10514
Access Level:Open access
Keyword:ENGENHARIAS
Engenharia elétrica
Transformadores elétricos
Sistemas de energia elétrica
Correntes de Inrush
Electtrical engineering
Electrical transformers
Electrical power systems
Inrush currents
Description
Summary:Also known as inrush current, magnetizing current arises during transformer energization. Due to its transient characteristics and high amplitude, this current can cause unwanted operations of protection relays, resulting in incorrect tripping. The main objective of this work is to compare different detection methods found in the literature, which are based on the analysis of differential current to distinguish transformer energization events from fault-related events. For this purpose, simulated datasets generated in PSCAD were used, and the methods were implemented and tested in the MATLAB environment. This study also proposes an approach based on the Synchrosqueezed Wavelet Transform (SWT) combined with the Bag of Features (BoF) model, evaluating its effectiveness under various operating conditions, such as diferente switching angles, fault inception times, and knee-point voltage levels of the magnetic saturation curve. The results highlight the importance of developing and applying accurate detection methods capable of precisely identifying inrush currents, minimizing false trips, and ensuring the reliability of protection systems. In this context, the proposed method — DIPI — achieved superior performance, standing out for its image-based classification strategy supported by SWT and BoF.