DOI: 10.66106/xinna7.20250103 ISSN: 3105-7527

基于人工智能的微电网能量管理策略(Energy management strategy of microgrid based on artificial intelligence)

张明昭 Mingzhao Zhang
Show PDF Cite
Abstract: The global energy structure is transforming towards a low-carbon direction. A large number of new energy power generation technologies such as photovoltaic and wind power are connected to the micro grid. This kind of power output has intermittent and random characteristics, and the traditional energy management strategy cannot simultaneously take into account the economy, reliability and real-time. Most of the existing methods rely on fixed rules or offline optimization schemes. In the face of complex and changeable operating environment, they lack adaptive adjustment ability. There are relatively large deficiencies in the coordinated operation of multiple power sources and the accuracy of load forecasting. In this paper, an AI driven micro grid energy management strategy is proposed. The core technical framework is deep learning and reinforcement learning. In this paper, the short-term prediction of new energy output and load demand is completed by using the long-term and short-term memory network, and the online decision-making model is built by combining the deep Q network to promote the collaborative optimization of energy storage system charging and discharging, controllable unit scheduling and interaction with the main network. On the basis of ensuring the stable operation of the microgrid, the strategy dynamically adjusts the output ratio of each unit, optimizes the operation cost and improves the new energy consumption rate. The test data shows that this method can deal with the uncertainty brought by the fluctuation of new energy. In many typical operation scenarios, the adaptability is stronger than the traditional PID control and model predictive control, reducing the wind and light rejection rate and extending the life of energy storage equipment.

More from our Archive