To promote new energy consumption capacity and economy and reliability of power supply, an industrial park microgrid with wind, solar, diesel, storage, living load and limited controllable industrial load is studied. An optimal scheduling scheme based on the combination of day-ahead and intraday is proposed. To ensure the economy of day-ahead scheduling, an improved particle swarm optimization algorithm with segmented mutation optimization is proposed to analyze the scheduling model. Compared to traditional particle swarm optimization, the improved algorithm has faster convergence speed, while improving economy and reducing the operating cost of the park. When running on the intraday, to correct the deviation in the day ahead prediction, the source, network, load and storage are jointly involved in the adjustment to ensure that the economic results, as much as possible, are followed and to improve the reliability of the microgrid system. Then, it establishes detailed simulation models within intraday using real-time simulation technology. A real-time simulation platform is constructed using real-time simulation software (RT-LAB) and machine (OP4510), to solve the problem of long simulation time and difficulty and verify the feasibility of the industrial park microgrid scheduling strategy.