Abstract: With the rapid development of residential electricity equipment, the proportion of residential electricity in social electricity consumption is increasing. Mobilizing residential load to participate in demand side response is worthy of in-depth research. First, a two-layer scheduling model for intelligent community participation in demand side response is established. The upper layer maximizes the profits of distribution network operators, and the lower layer has the lowest total scheduling cost for intelligent community users. Second, the model considers the load models of various types of electrical equipment, user comfort, and incentive measures based on peak shaving contribution index constraints. Finally,considering the issue of power fluctuations caused by residential load participation in scheduling, and taking high-power air conditioning loads as an example, it proposes corresponding control strategies to address the phenomenon of power drops caused by air conditioning participation in scheduling. The results show that the double-layer scheduling model can enable users to schedule according to the scheduling expectations of distribution network operators, the peak shaving effect is good. It also takes into account the interests of both distribution network operators and users. The two proposed control strategies effectively reduce power drop and improve the stability of the distribution network.