Hourly electric load forecasting using Nonlinear AutoRegressive with eXogenous (NARX) based neural network for the state of Goa, India

Accurate models for electric power load forecasting are essential for the operation and planning of power system from technical as well as financial perspective. Paper proposes an approach for short term electric load forecasting based on parameters which have been arrived from past load data using artificial neural network based Non linear autoregressive network exogenous technique. Novel approach for obtaining the seasonality factor, weekly trend and load increase pattern from past electricity consumption data are also proposed. Proposed methodology requires lesser real time inputs such as weather information. The real time active power load consumption data in MW for two and half years of Goa Electricity Board of Goa state from India is used for predicting future load demand. The results obtained from the model successfully predicts the future load data for week days with mean square error less than 1.67% and mean absolute deviation of 3.6%, which proves suitability of our proposed technique for forecasting.