Air conditioning loads (ACLs) are important demand-side resources. However, existing researches on the adjustable capability of ACLs based on temperature control mainly focus on system performance during the adjustment of temperature setpoints, without addressing the recovery of setpoints. These approaches typically assume one-time participation of ACLs in demand response, which limits the full utilization of the flexibility. To systematically describe the aggregated power characteristics of air conditioner clusters throughout the entire process of temperature setpoint adjustment and recovery, the widely used safe protocol (SP) in temperature control is expanded to the recovery phase. Based on this, an aggregated power model that fully describes the process of homogeneous ACLs participation in demand response is established. Furthermore, considering the consistency of air conditioning users’ comfort, a user selection model based on a comfort factor ranking index is proposed to determine the regulation strategy of air conditioning users participating in demand response in batches and multiple rotations. Simulation results show that the proposed control strategy enables multiple rounds of user participation, reduces differences in comfort levels among users, and achieves higher power control accuracy with smaller power rebound fluctuations.