Artificial intelligence-based forecasting models for integrated energy system management planning: An exploration of the prospects for South Africa

The regional energy demand for Southern Africa has been predicted to increase by ten to fourteen times between the years 2010 and 2070. Thus, to address the proliferation of energy demand, South Africa’s integrated resource plan, which includes using renewable energy sources to increase the electricity supply and reduce the country’s carbon footprint, has been formulated. However, integrating renewable power into the power grid brings different dynamics for the system operators, as renewable power sources are variable and uncertain. Thus, accurate demand and generation forecasting become critical to the safe operation and ensuring continuity of supply, as consumers require. Due to the complexity of the earth’s atmosphere, weather forecasting uncertainty, and region-specific criteria, traditional forecasting models are limited. Thus, Machine Learning, Deep Learning, and other artificial intelligence techniques are attractive possibilities for improving classical forecasting models. This study comprehensively reviewed relevant works on AI-based models for generation potential and load demand forecasting toward intelligent energy resource management and planning. The approach involved searching research databases and other sources for studies, reports, and publications on location-specific energy resource management using criteria such as demography, policy, and sociotechnical information. Consequently, the review study has highlighted how AI predictive analytics can enhance long-term energy resource potential and load forecasting toward improving electricity sector performance and promoting integrated energy system management implementation in South Africa. • Most forecasting models’ capabilities are limited by lack of ‘localized’ information. • AI-based forecasting tools are essential for improved energy management systems. • Explored AI-based methods and the traditional approach for energy resources management. • Identified critical policy contexts and sociotechnical factors affecting energy sector activities. • Presented recent AI-based energy resource forecasting with data decomposition approaches. • Provided information on AI predictive model deployment for South Africa DSM adoption.