Dynamic Energy Management: Predictive Maximum Demand Control with Short-Term Power Forecasting

Industries face rising energy costs and stricter regulations, making energy efficiency crucial. A key challenge for building operators is controlling maximum demand (MD), which significantly impacts electricity bills. Traditional MD control methods in Malaysia, used in Building Management Systems (BMS), rely on historical averages that may not provide accurate predictions. This paper introduces a novel predictive approach using Gradient Boosting for short-term power forecasting (STPF) to optimize MD control. The contributions are twofold: i) Unlike ultra-short-term forecasting (USTPF) models like Prophet- VMD, this method employs Gradient Boosting for improved accuracy. A three-stage MD control framework dynamically adjusts energy usage by turning off non-essential equipment to manage peak demand. Testing at a prestigious Malaysian hotel showed up to 21 % MD reduction, saving approximately 10,102 RM. ii) A user- friendly dashboard enables BMS operators to make data-driven decisions based on precise forecasts. This integration of advanced forecasting and dynamic management enhances cost efficiency, sustainability, and operational performance, offering a robust solution to modern energy challenges.