Short-Term Multinodal Load Forecasting Using a Fuzzy-ARTMAP Neural Network

• Autor(a) principal: Abreu, T.
• Data de Publicação: 2019
• Outros Autores: Moreira, J. R. [UNESP], Minussi, C. R. [UNESP], Lotufo, A. D.P. [UNESP], Lopes, M. L.M. [UNESP]
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1109/ISGT-LA.2019.8895486; http://hdl.handle.net/11449/198214
• topic: Aggregate Load; Electrical System Distribution; Fuzzy ARTMAP Neural Network; Load Forecasting

The prediction of electric charges is essential in the electric power system, because it establishes when and how much of generation, transmission and distribution capacity must be arranged to meet the expected load without interruptions in supply. Therefore, the more accurate, reliable and fast the results, the better quality the forecast will be. This paper proposes an approach that performs the forecast considering several points of the electricity network (multinodal forecast), where different types of consumers are considered (industrial, commercial and residential). In this problem is used an ARTMAP Fuzzy artificial neural network , that is based in the theory of resonance adaptative (ART). The main characteristic of neural networks of the ART family is the stability and plasticity that provide results quickly and accurately. In order to test the proposed forecast system, results of 24 hours (48 points) ahead are presented for nine substations of a New Zealand Electrical Company.