随着分布式光伏渗透率不断提高，传统的光伏建模方法已无法满足建模需求。为建立大规模地区分布式的光伏系统的模型并研究其运行特性，提出一种分布式光伏发电系统动态分群多机等值建模方法。文中选择具有代表性的故障曲线的特征点作为聚类指标，然后根据改进的FCM（fuzzy-C means）聚类算法以选取的特征点为指标对区域内的分布式光伏系统进行聚类，然后对等值模型进行参数设计，对于难以获得的控制参数，提出一种基于非支配性排序遗传算法II（non-dominated sorting genetic algorithm II, NSGA-II）和主成分分析的多目标优化算法来对控制参数进行辨识，其他参数根据容量加权法获得。从而构建区域分布式光伏系统多机等值模型。通过Simulink仿真工具验证了所提等值建模方案能够明显地提高分布式光伏等值模型的准确性。
英文摘要:
With the continuous increase in the penetration rate of distributed photovoltaics, traditional photovoltaic modeling methods are no longer able to meet modeling needs. To establish a model for large-scale regional distributed photovoltaic systems and study their operational characteristics, a dynamic clustering multi-machine equivalent modeling method for distributed photovoltaic power generation systems is proposed. Firstly, representative feature points of the fault curve are selected as clustering indicators. Then, based on the improved FCM (fuzzy-C means) clustering algorithm, the distributed photovoltaic systems in the region are clustered using the selected feature points as indicators. Subsequently, the parameters design of the equivalent model is carried out. For control parameters that are difficult to obtain, a multi-objective optimization algorithm based on NSGA-II (non-dominated sorting genetic algorithm II) and principal component analysis is used to identify them. Other parameters are obtained using capacity weighting method. Thus a multi-machine equivalent model for regional distributed photovoltaic systems is constructed. Finally, the feasibility and accuracy of the proposed equivalent modeling scheme of distributed photovoltaic equivalent model is significantly improved using Simulink simulation tools.