Intelligent deep learning techniques for energy consumption forecasting in smart buildings: a review

Urbanization increases electricity demand due to population growth and economic activity. To meet consumer’s demands at all times, it is necessary to predict the future building energy consumption. Power Engineers could exploit the enormous amount of energy-related data from smart meters to plan power sector expansion. Researchers have made many experiments to address the supply and demand imbalance by accurately predicting the energy consumption. This paper presents a comprehensive literature review of forecasting methodologies used by researchers for energy consumption in smart buildings to meet future energy requirements. Different forecasting methods are being explored in both residential and non-residential buildings. The literature is further analyzed based on the dataset, types of load, prediction accuracy, and the evaluation metrics used. This work also focuses on the main challenges in energy forecasting due to load fluctuation, variability in weather, occupant behavior, and grid planning. The identified research gaps and the suitable methodology for prediction addressing the current issues are presented with reference to the available literature. The multivariate analysis in the suggested hybrid model ensures the learning of repeating patterns and features in the data to enhance the prediction accuracy.