Research on power model calibration and control algorithm response characteristics of flexible load controller for multi-split air conditioners

In view of the current lack of specific multi-split air conditioner flexible load response control algorithms and controller hardware devices, this paper achieves a complete functional construction through equipment measurement point and functional networking design, experimental testing, BP neural network power model construction, ETP model simulation experiments, and flexible load response control algorithm design. In response to the significant power rebound phenomenon during the flexible load response process of air conditioners, control strategies of continuously adjusting the set temperature in small steps and partially turning off indoor units are designed and experimentally verified. The results show that for a 17.5 kW air conditioner, in the cooling mode, for every 1 ℃ increase in the set temperature, the actual power can decrease by approximately 2.4-3.4 kW, and the flexibly adjustable load within the comfort range accounts for about 40%-60% of the rated power. Continuously adjusting the set temperature in small steps can establish a new balance within about 1 hour, while partially turning off indoor units can achieve a load response of 35% to 55%.