Grid supply-demand balance faces severe challenges, and air conditioning loads, as typical controllable loads, have significant demand response potential. Although the individual capacity is small, the large user base makes it a sizable demand-side resource after aggregation. In this paper, a grid supply and demand optimization scheduling method based on air-conditioning load is proposed for the grid supply and demand imbalance problem. Firstly, an indoor temperature prediction model is constructed based on the extreme learning mechanism to realize the indoor temperature prediction after 5 minutes using the historical temperature data as input, and the adjustable capacity of air conditioning load is determined accordingly. Second, an air conditioning load regulation strategy considering human comfort is designed, and the comfort temperature interval is set to 22-28°C, with the goal of minimizing the comfort cost for optimal scheduling. Finally, a supply-demand cooperative optimization model including time-of-use tariff and incentive-based demand response is constructed to optimize scheduling with the objective of minimizing the operating cost of the user’s optical storage microgrid. The simulation results of the algorithm show that when the TSV index is used to evaluate the central air-conditioning load clusters, the comfort users can participate in the scheduling for 7.56 minutes, and the dispatchable capacity reaches 10.3 MW, while the economy users can participate in the scheduling for up to 21.8 minutes, and the dispatchable capacity reaches 13.8 MW. In the real-time scheduling strategy, the time granularity of 5 minutes is used during the time period 17:00- 19:00 In the real-time scheduling strategy, when scheduling with 5-minute time granularity from 17:00 to 19:00, the power difference of the contact line is 0.40kW, and the number of iterations is 170, which is a significant improvement compared to the scheduling effect of 15-minute time granularity. The method in this paper provides a feasible technical path for grid supply and demand regulation.