A Survey Of Artificial Intelligence Methods For Renewable Energy Forecasting: Methodologies And Insights

Blessing Olatunde Abisoye,Yanxia Sun,Wang Zenghui

DOI: https://doi.org/10.1016/j.ref.2023.100529

2023-12-22

Renewable Energy Focus

Abstract:The efforts to revolutionize electric power generation and produce clean and sustainable electricity have led to the exploration of renewable energy systems (RES). This form of energy is replenished and cost-effective in terms of production and maintenance. However, RES, such as solar and wind energies, is intermittent; this is one of the drawbacks of its usage. In order to overcome this limitation, studies have been undertaken to forecast its availability and power output. The current trending method of forecasting availability and the power generated by the RES is the artificial intelligence (AI) method. However, with all its potential, traditional AI, such as Artificial Neural Network (ANN), Support Vector Machine (SVM) and many more, does not have it all. Because of this, metaheuristic algorithms are being explored as optimization techniques to increase the performance accuracy of these AI forecasting methods and overcome some challenges of traditional AI models. This study presents an insightful survey of AI (traditional AI and metaheuristic) methods explored to forecast the availability of solar and wind renewable energy systems. A survey of the existing surveyed literature was presented. The taxonomy of the explored AI in RES was formulated, and the theoretical backgrounds of some of the traditional AI algorithms were presented. Also, the various forms of metaheuristic algorithms and the improved versions applied to optimize classical AI methods to forecast RES systems' availability and power output were surveyed. A conceptual framework of the hybrid AI application in RES was formulated. Finally, the survey discussion, insight, challenges of the existing models and future directions were presented.