A Method for the Monthly Electricity Demand Forecasting in Colombia based on Wavelet Analysis and a Nonlinear Autoregressive Model

This paper proposes a monthly electricity forecast method for the National Interconnected System (SIN) of Colombia. The method preprocesses the time series using a Multiresolution Analysis (MRA) with Discrete Wavelet Transform (DWT); a study for the selection of the mother wavelet and her order, as well as the level decomposition was carried out. Given that original series follows a non-linear behaviour, a neural nonlinear autoregressive (NAR) model was used. The prediction was obtained by adding the forecast trend with the estimated obtained by the residual series combined with further components extracted from preprocessing. A bibliographic review of studies conducted internationally and in Colombia is included, in addition to references to investigations made with wavelet transform applied to electric energy prediction and studies reporting the use of NAR in prediction