A comprehensive survey on load forecasting hybrid models: Navigating the Futuristic demand response patterns through experts and intelligent systems
Kinza Fida, Usman Abbasi, Muhammad Adnan, Sajid Iqbal, Salah Eldeen Gasim Mohamed
Abstract
Load forecasting is a crucial task, which is carried out by utility companies for sake of power grids' successful planning, optimized operation and control, enhanced performance, and guaranteed reliability and adequacy. It mainly reduces the gap between demand and supply and thus assures an uninterrupted power supply and load shedding free operation. The electricity demand can be forecasted by several Artificial Intelligence (AI), Deep Learning (DL), and Machine Learning (ML) approaches of which hybrid methods are the most famous. This paper examines various present technologies and algorithms of load forecasting including digital twin (DT), data-mining (DM), federated learning (FL), and transfer learning (TL). In this work, a comprehensive assessment of forecasting with single and hybrid models is performed, their functions are reviewed and pros and cons are investigated. The Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Squared Error (RMSE) are used for performance assessment and comparison of the different models searching for best load forecasting model. The role of different experts and intelligent systems used for optimal load forecasting in power systems will be clearly elaborated along with their pros and cons, thus providing a roadmap research path for the readers, who wants to contribute to the scientific society related to this research field. • The key contributions of this paper are as follows: • LF techniques like digital twin (DT), data mining (DM), federated learning (FL), and transfer learning (TL) are examined. • An assessment of forecasting models, their pros, and cons, along with applications and limitations are reviewed. • Best LF models are concluded on the basis of MSE, RMSE, MAE, and MAPE.