In this paper, an island integrated energy system (IES) combining power, heat and hydrogen (CPHH) is built, and optimal power-heat-hydrogen scheduling is studied under actual plateau meteorological data. Considering the comprehensive demands of residents, an IES including renewable energy sources (RESs), fuel cell, electrolyzer, electric boiler, thermal energy storage device and new energy vehicles is established and operated. In the first stage of operation, according to the prediction of RESs and demands, day-ahead optimization is carried out which aims to minimize operation cost, while electric vehicles (EVs) are considered as special demand response (DR) loads. For the second stage of operation, ultra-short-term prediction is implemented to provide prediction data for model predictive control (MPC), realizing real-time operation. Functioning as a CPHH system, the fuel cell and electrolyzer collaborate to meet various needs of the plateau residential area with high efficiency. In addition, the method and IES structure proposed in this paper is compared with other options.

With the increasing presence of intermittent energy resources in microgrids, it is difficult to precisely predict the output of renewable resources and their load demand. In order to realize the economical operations of the system, an energy management method based on a model predictive control (MPC) and dynamic programming (DP) algorithm is proposed. This method can reasonably distribute the energy of the battery, fuel cell, electrolyzer and external grid, and maximize the output of the distributed power supply while ensuring the power balance and cost optimization of the system. Based on an ultra-short-term forecast, the output power of the photovoltaic array and the demand power of the system load are predicted. The off-line global optimization of traditional dynamic programming is replaced by the repeated rolling optimization in a limited period of time to obtain power values of each unit in the energy storage system. Compared with the traditional DP, MILP-MPC and the logic based real-time management method, the proposed energy management method is proved to be feasible and effective.