In order to meet the demand of low carbon and cleanliness in the transformation of energy structure, the situation that microgrid participates in the energy market, real-time energy market and carbon trading market at the same time is considered. In view of the high forecasting accuracy and error distribution law of market electricity price, the stochastic programming method is used to deal with its uncertainty. In view of the randomness and intermittence of photovoltaic output, the dynamic robust optimization method is used to deal with it. A two-stage robust optimal scheduling model of microgrid considering the uncertainty of electricity price and photovoltaic output is constructed, and the bad scene identification algorithm is used to decompose the original problem into the main problem and sub-problem iteratively. The sub-problem is used to identify the worst photovoltaic scenario, and the single-layer optimization model under this scenario is solved by the main problem, which greatly reduces the number of scenarios needed and improves the computational efficiency of the model. A numerical example is given to verify the effectiveness of the algorithm, and the results show that the microgrid participating in multiple power markets at the same time can significantly increase profits, and the use of bad scene identification algorithm can reduce the computational burden and effectively improve the computational efficiency, and it is more adaptable in large-scale scenarios.