Estimação do preço de liquidação das diferenças no mercado de energia elétrica de curto prazo utilizando redes neurais artificiais

Abstract

After the restructuring of the commercialization models and rules of electric energy in Brazil, a new way of negotiating electricity purchase and sale in the short term market based on the differences settlement price (PLD). Due to the characteristics of the Brazilian energy matrix, which is mostly composed of hydraulic generation, the PLD tends to be a volatile value due to variations in hydrological conditions, demand requirements, cost of deficit and price of fuels used in thermoelectric plants. In this work the PLD will be estimated through an artificial neural network using historical series of several variables that make up or influence the formation of electric energy price in short term market. This work will use twoartificial neural networks in the estimation process, the Time Delay Neural Network (TDNN) and the Multlayer Perceptron (MLP), also the Pearson coefficient will be used as a proposal for the selection of input variables in the MPL network.