Previsão de demanda de energia elétrica com a utilização de redes neurais artificiais

Abstract

The importance of electric power in people’s lives is undeniable, as it contributes to the development and well-being of a society. Therefore, it is necessary to seek methods and tools to ensure safe planning for the supply of electric power to the population. Thus, this work proposes the use of Artificial Neural Networks (ANN) for this purpose. Multiple Layer Perceptron (MLP), Radial Basis Function (RBF), Extreme Learning Machine (ELM), Echo State Network (ESN), and Long-Short Term Memory (LSTM) neural networks were developed to forecast time series of electric power demand for the states of Paraná, Santa Catarina, and Rio Grande do Sul. The results obtained show the feasibility of the proposal, as all the networks achieved good results, presenting acceptable errors and generating prediction curves consistent with reality, with the MLP being the network with the best predictability.