Intelligent phase balancing system for campus energy management: A comparative study of AI, automatic feeder shifting, and battery energy conversion

Abstract

In the context of thailand's continuously increasing electricity consumption, effective energy management in educational buildings has become a crucial strategy to reduce operational costs and concretely promote environmental sustainability. This study aims to analyze real-time electrical load data from the integrated industrial technician building at Rajamangala university of technology srivijaya, in order to develop a demand side management (DSM) strategy that aligns with actual usage conditions. Three types of automatic load control systems were designed and tested in this study: feeder-shifting, which redistributes load to balance the system, AI-agent, an intelligent system capable of learning and adapting to dynamic load conditions, and battery inverter, which alleviates peak demand during critical periods. Experimental results revealed that the AI-agent system demonstrated the highest effectiveness in reducing the peak load, with an average reduction of 21.7%. The battery inverter and feeder-shifting systems followed with reductions of 15.3% and 12.8%, respectively. Notably, the AI-agent system showed superior learning capability and responsiveness to fluctuating load conditions compared to other techniques. The findings indicate that integrating intelligent load control systems with real-time energy analytics holds high potential in driving educational buildings toward becoming net-zero energy buildings (NZEB). Furthermore, the approach can be effectively applied to other building types with similar load characteristics, such as office buildings and public facilities, to enhance national energy efficiency.