Analysis on Electrical Load Characteristics of Air Conditioning in Commercial Buildings

Commercial buildings’ AC systems consume significant energy and exhibit time-varying, nonlinear, and strongly coupled electrical load characteristics. This limits their operational efficiency and challenges grid stability. To enhance the accuracy and physical interpretability of existing analysis methods, a hybrid approach combining data-driven and physics-based analysis is proposed. It integrates multi-source data, identifies key load-affecting features, and analyzes their dynamic coupling. Then, a hybrid prediction model fusing machine learning and AC physical models is built. Finally, based on the prediction results, load characteristics such as peaks and valleys, fluctuations, and response potential are thoroughly analyzed to formulate refined energy-efficiency optimization strategies. This approach improves prediction accuracy and adaptability to unknown working conditions. Case studies show that it can effectively uncover the dynamic characteristics of AC loads, offering support for commercial building energy-saving and grid interaction.