The primary frequency control (PFC) is becoming more important for maintaining the system balance between the generation and consumption. The progressed intelligent home system has made it easier for household appliances to provide PFC services to power systems. Among the appliances, inverter air conditioners (IACs) can adjust the operating power rapidly by changing the compressor frequency and have little influence on the customer comfort. In this paper, an IAC model is developed considering the thermal model of the room, whose compressor operating frequency is determined by the system frequency deviation and the temperature deviation between the room and the set value. Two control methods for the aggregated IACs are proposed, the centralized detection control (CDC) method and the distributed detection control (DDC) method. The system frequency is detected by the control center in the CDC method, while the system frequency is monitored locally by each IAC controller in the DDC method. Generally, the CDC method can get more accurately frequency detection than the DDC method. However, the control center in the CDC method has to send control signals to each IAC in real time, which causes communication delays for the IACs to adjust the operating power. As for the large-scale aggregated IACs, the simulation results show that the signal delays in the CDC method have more negative effects than the detection errors in the DDC method.