Demand response (DR) technologies can cut peaks and fill valleys, improve energy efficiency and power system stability. Direct load control (DLC) makes the demand-side load available by directly controlling the power load. It is an effective method of implementing excitation-based DR by directly controlling the power load, which is relatively easy to implement. Air conditioning has good heat storage or cold storage capacity, and its electricity consumption is an important part of total household energy consumption, which can be transferred power demand to off -peak hours. Therefore, air conditioning can affect the peak-to-valley difference of the total load curve and is an important demand response resource. The thermodynamic model of air conditioning is established. The model of large-scale air conditioning load aggregation based on Monte Carlo method is proposed. The air conditioning load tracking control scheme of DLC is studied. The load reduction potential of air conditioning load participating in DR is discussed. The reference signal for controlling the air conditioner temperature or the open or close state is calculated according to the load peaking demand, and then the adaptive hill climbing (AHC) control method is used to track the implementation of the reference signal. The simulation results show that the DLC that tracks the reference signal to implement the air conditioning load can effectively reduce the load as needed.