A Hybrid AC/DC Microgrid Energy Management Strategy Based on Neural Network

An energy management strategy (EMS) for hybrid AC/DC microgrid is proposed in this paper. Since the accuracy of microgrid management is affected by the error of power prediction, this proposal is aiming to improve the economic performances of a microgrid while considering the prediction error. The management process is divided into 3 stages: day-ahead scheduling, intraday pre-scheduling and intraday scheduling. Economic performance is considered as the most important factor in the day-ahead scheduling stage. In this situation, the power forecast results are used to calculate the management command. Neural network is introduced in the management strategy of intraday pre-scheduling stage. In this process, the model of power generation, energy storage and other devices of microgrid is proposed, which is important for the dispatch of next stage. This process can reduce the error of management caused by power forecast. In the stage of intraday scheduling, the models of microgrid are taken used to calculate the dispatching command. The power forecasting results are used in this stage. Simulation results are shown in this paper and verified the effectiveness of the proposed strategy. The economic performances and operation reliability of the microgrid are both improved.