Engineering and Data-Driven Approaches for Load Forecasting: A Review

Abstract

Precise load forecasting is critical for ensuring the equilibrium between electricity supply and demand, reducing operational costs, and optimizing resource scheduling in power systems. In this study, a comprehensive literature review on the engineering and data-driven approaches for load forecasting is conducted. A systematic review was performed using the Scopus database, narrowing down 635 initial results through filtering. VOSviewer software was utilized to analyze keyword relationships, revealing major themes and potential areas for further research. The literature reviewed highlights two primary categories of load forecasting methods: engineering (or physics-based) models and data-driven (or AI-based) models. Engineering methods utilize thermodynamic principles and detailed contextual data, such as building structure and HVAC systems, to estimate energy consumption, as demonstrated in software like EnergyPlus and eQUEST.