Cheng Yanming, Zhang Jinqi, Al Shurafa Mahmoud, Liu Dejun, Zhao Yulian, Ding Chao, Niu Jing
Electrical and Information Engineering, Beihua University, Jilin, 132000, China.
School of Electrical Engineering, Zhejiang University, Hangzhou, 310027, China.
Sci Rep. 2025 May 23;15(1):17988. doi: 10.1038/s41598-025-98665-x.
Microgrid (MG) is basically composed of different distribution generators (DGs) connected in parallel for supplying a specific set of loads managed by an Energy management system (EMS). EMS is a control system integrated within MGs for managing the operations of these DGs effectively to fulfill a power balance between power production and load demand in the most optimal way, especially in island MGs. In this paper, an EMS based on Multiple Adaptive Neuro-Fuzzy Inference System optimized by Genetic Algorithm (MANFIS-GA) is proposed for PV/Wind/Diesel Generator/Battery (PWDB) island MG system, to optimize the output power of diesel generator, manage charging-discharging operation of MG Battery Storage keeping its State of Charge (SOC) in acceptable limits, and improve the MG system reliability and stability by mitigating the effects of sudden changes in the electrical loading and Renewable energy sources (RES) Power. The prediction system is implemented by using 8760 samples based on an hourly meteorological data of a whole year. GA is used as an optimization technique for training MANFIS to accomplish the desired objects of EMS. For evaluation purpose, a real case study of a day-ahead data is tested and discussed in details. Experiments show that the proposed smart system provides accurate results for the expected outputs and achieves a good performance.
微电网（MG）基本上由不同的分布式发电机（DG）并联组成，用于为一组特定的负载供电，这些负载由能量管理系统（EMS）管理。EMS是集成在微电网中的控制系统，用于有效管理这些分布式发电机的运行，以最优化的方式实现发电与负载需求之间的功率平衡，特别是在孤岛微电网中。本文针对光伏/风能/柴油发电机/电池（PWDB）孤岛微电网系统，提出了一种基于遗传算法优化的多重自适应神经模糊推理系统（MANFIS-GA）的能量管理系统，以优化柴油发电机的输出功率，管理微电网电池储能的充放电操作，使其荷电状态（SOC）保持在可接受的范围内，并通过减轻电气负载和可再生能源（RES）功率突然变化的影响，提高微电网系统的可靠性和稳定性。预测系统基于一整年的每小时气象数据，使用8760个样本来实现。遗传算法用作训练多重自适应神经模糊推理系统的优化技术，以实现能量管理系统的预期目标。为了进行评估，对一个日前数据的实际案例进行了测试并详细讨论。实验表明，所提出的智能系统为预期输出提供了准确的结果，并取得了良好的性能。
