Resumo:
The identification of high-impedance faults is a recurring topic in the literature, given the importance of electric power in modern life and the disruptions caused by its absence. High-impedance faults are difficult to identify, presenting peculiarities and characteristics that can easily be mistaken for simple load variations in electrical systems. Conventional methods, such as those used in protection relays, may fail to correctly identify high impedance faults, delaying system restoration. The problems associated with delayed identification can have both economic impacts and pose safety risks to living beings. Therefore, considering the advancements in computational methods and their contribu tions to solving problems across various fields, the objective of this work is to apply these methods to the identification of high-impedance faults. To achieve this, machine learn ing algorithms were employed in combination with the Continuous Wavelet Transform (CWT) to analyze simulated electrical signals in IEEE 13-bus and 34-bus systems. The results showed improvements in the accuracy and precision of fault detection compared to existing methods in the literature, validating the effectiveness of the proposed approach.
