The rise of the concept of smart city has spawned a large number of pilot projects in smart community. Its complete IoT, regulation and measurement system can support the smooth implementation of automatic demand response (ADR). It is necessary to design ADR model of smart community to improve the reliability, robustness and cost-effectiveness of smart community operators. Therefore, an ADR model based on robust optimization is proposed. The model aims to maximize the economic benefits of smart residential areas. In order to ensure the high implemensibility of ADR scheduling plan, two types of block ADR strategies are designed to solve the contradiction between the continuous demand response mode and the electricity consumption law of residential users, and the strategy of using the gas turbine under its jurisdiction to stabilize the wind power forecast deviation. The nested segmentation algorithm is used to solve the model, so that the solution converges to the global feasible region. The example is carried out in the modified Garver-6 node system to verify the effectiveness of the proposed model.