Multinodal load forecasting using an ART-ARTMAP-fuzzy neural network and PSO strategy

• Autor(a) principal: Antunes, Juliana Fonseca
• Data de Publicação: 2013
• Outros Autores: De Souza Araujo, Nelcileno Virgilio, Minussi, Carlos Roberto [UNESP]
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1109/PTC.2013.6652373; http://hdl.handle.net/11449/227394
• topic: Adaptive Resonance Theory; Artificial Neural Network; Multinodal Load Forecasting; Particle Swarm Optimization

This work presents a system based on Artificial Neural Networks and PSO (Particle Swarm Optimization) strategy, to multinodal load forecasting, i.e., forecasting in several points of the electrical network (substations, feeders, etc.). Short-term load forecasting is an important task to planning and operation of electric power systems. It is necessary precise and reliable techniques to execute the predictions. Therefore, the load forecasting uses the Adaptive Resonance Theory. To improve the precision, the PSO technique is used to choose the best parameters for the Artificial Neural Networks training. Results show that the use of this technique with a little set of training data improves the parameters of the neural network, calculated by the MAPE (mean absolute perceptual error) of the global and multinodal load forecasted. © 2013 IEEE.