| Texto completo | |
| Autor(es): |
Moraes, Anderson L.
;
Fernandes, Ricardo A. S.
;
IEEE
Número total de Autores: 3
|
| Tipo de documento: | Artigo Científico |
| Fonte: | 2020 IEEE POWER & ENERGY SOCIETY GENERAL MEETING (PESGM); v. N/A, p. 5-pg., 2020-01-01. |
| Resumo | |
Currently, research focused on the pattern classification in the area of Power Quality has been using Deep Learning techniques. However, some techniques, such as Convolutional Neural Networks, despite its good performance, require images as input data. In this sense, the present paper proposes an analysis of Recurrence Plots textures as a way to identify Power Quality disturbances. Among the disturbances were considered the short and long duration voltage variations, the oscillatory and impulsive transients, the voltage fluctuations and waveform distortions. These analyzes allowed to identify distinct patterns between disturbances and, consequently, demonstrate that Recurrence Plot can be properly employed as a novel feature engineering technique in the scope of Power Quality disturbances classification. (AU) | |
| Processo FAPESP: | 19/15192-9 - Monitoramento não invasivo de cargas baseado em gráficos de recorrência e deep learning no contexto de smart homes |
| Beneficiário: | Ricardo Augusto Souza Fernandes |
| Modalidade de apoio: | Auxílio à Pesquisa - Regular |