Detection and classification of electrical faults in an electrical power system using the Wavelet transform and neural networks

• Autor(a) principal: Erazo-Velasco, Ismael Elías
• Data de Publicação: 2022
• Outros Autores: Bautista-Sánchez, José Vicencio, Rodríguez-Jijón, Roberto Iván, González-Quiñonez, Luis Adrián, Chere-Quiñónez, Byron Fernando
• Tipo de documento: Artigo
• Idioma: spa
• Título da fonte: Sapienza (Curitiba)
• Texto Completo: https://journals.sapienzaeditorial.com/index.php/SIJIS/article/view/537
• topic: Transformada Wavelet, técnicas de detecção e detecção de falhas, redes neurais inteligentes.; Wavelet transform, fault detection and detection techniques, intelligent neural networks; Transformada de Wavelet, Técnicas de detección y detección de falla, redes neuronales inteligentes

The main objective of this research was the detection and classification of electrical faults in an electrical power system using the Wavelet transform and Neural Networks. The methodology consists of two steps; In the first step, the calculation and design process of a radial power system is carried out. The second phase throws the modeling and simulation of fault detection and classification in Matlab/ Simulink. The results obtained indicate that the neural network predicts and classifies the types of faults in the electrical power system. It is concluded that although the neural model with Bayesian regularization and early completion offers relatively low errors, it has the drawback of being a rigid black box, that is, the neural network simply evaluates its inputs and produces its outputs, but it is not known What. For this reason, when a misclassification occurs, there is no way to make changes and the network must be completely retrained, which is not appropriate from a practical point of view.