Day-ahead residential load forecasting with artificial neural networks using smart meter data

Load forecasting is an important operational procedure for the electric industry particularly in a liberalized, deregulated environment. It enables the prediction of utilization of assets, provides input for load/supply balancing and supports optimal energy utilization. Current residential load forecasting is mainly based on the use of synthetic load profiles due to lack of or insufficient historical data. However, the advent of smart meters presents an opportunity for making accurate residential load forecasting possible. In this paper artificial neural networks are used with weather data and historical smart meter data for day-ahead load prediction. Extensive error analyses are performed on the model to investigate the suitability of the model for day-ahead prediction. The forecast model can be implemented by energy suppliers and distributed system operators for submission of day-ahead bids and for management of network assets respectively.