There is uncertainty over the amount of wind power, photovoltaic output and load size in a microgrid. To tackle this, starting from the two stage of day-ahead and real-time, a two-stage optimal dispatching model of a microgrid is established. In the day-ahead dispatching stage, an economic dispatching model based on the optimal total operation cost of the microgrid is established. Based on the results of day ahead scheduling optimization, the real-time scheduling stage considers the output adjustment sequence of each power supply in real-time optimization scheduling, and proposes a real-time rolling optimization strategy with dynamic incentive demand response participation. The incentive price and user participation capacity of the strategy presents dynamic changes with a time scale moving forward, aiming to use price to guide users to increase their participation in demand response. A calculation example shows that dynamic incentive model demand response, compared with static incentive model demand response, can not only improve the tracking effect of the day-ahead tie line plan, but also better eliminate the day-ahead prediction error of the microgrid. It can significantly improve user income and effectively reduce the system operation cost. This provides a reference for the optimal operation of the market-oriented microgrid. This work is supported by the National Natural Science Foundation of China (No. 52077146).