Artificial intelligence-based strategies for sustainable energy planning and electricity demand estimation: A systematic review

Access to electricity is a cornerstone for sustainable development and is pivotal to a country's progress. The absence of electricity impedes development and elevates poverty. The first step in sustainable energy planning is accurately estimating the people's electricity demand. However, accurately estimating or modelling electricity demand for localised communities has been a longstanding challenge since the inception of electricity, exacerbated by the continuous introduction of new electrical appliances, the need for more accurate and available data, and the unpredictable behaviour of individuals when using these appliances. This study seeks to develop a systematic review of existing research on predicting or forecasting electricity consumption in rural and urban areas. The study considered a bottom-up, top-down and hybrid approach with Machine Learning (ML), Deep Learning (DL), decomposition ensemble and AI-based optimization as techniques leveraged. The limitations of the models employed were also outlined, and lastly, open challenges and future directions were proposed. It was observed from the model categorization that decomposition ensemble and hybrid techniques may give a promising result; hence, they could help create an accurate and robust prediction or forecasting model for electricity demand. [Display omitted] •The absence of electricity impedes development and elevates poverty.•The proliferation of new appliances complicates accurate demand estimation.•Techniques include ML, DL, AI-based optimization, and decomposition ensemble.•Models' limitations are emphasized in the systematic review.•Enhancing load forecasting accuracy is a worthwhile direction for further study.