数据驱动的电力系统运行方式分析

随着电网中高比例可再生能源及电力电子设备渗透率的提高,电力系统的运行形态将发生深刻变化,基于经验选择的电力系统方式分析,越来越难以应对电力系统运行、规划、保护和稳定分析的要求。为此,该文基于电力系统时序运行模拟数据,提出了数据驱动的电力系统运行方式及其形态变化分析方法。该方法首先对高维运行模拟数据进行预处理,通过聚类和紧密度指标饱和点分别辨识典型运行模式及其数量,利用t-SNE降维算法提取主要特征并可视化运行方式的时空变化规律;接着,从定量的角度,提出了描述电力系统运行方式分散性、季节一致性、时序多变性的评估指标。最后,以甘肃电网实际算例系统分析了高比例可再生能源对电力系统运行方式的影响,验证了文中数据驱动方法的有效性。实验结果表明,电力系统运行方式的分散性及典型运行方式的数量会随着可再生能源渗透率的上升而迅速增加。此外,电力系统运行方式与季节的耦合性逐渐减弱。

Data-driven Power System Operation Mode Analysis

The power system operation mode changes substantially due to increasing renewable energy penetration and power electronic devices integration in power system. New operation modes are emerging in power system operation, planning, protection, and stability analysis, which can hardly be handled by traditional model and experience-driven method. To this end, this paper proposed using data-driven method to analyze power system operation mode and its variation based on high dimensional simulated chronological power system operation data. This method first preprocesses the data, identifies representative operation mode patterns and its number using clustering algorithm, and then extracts key features using t-SNE for operation modes visualization. In addition, several indices were proposed to evaluate power system operation modes space dispersion, seasonal consistency, and time variation. A case study on Gansu provincial power system in China showed the impacts of high renewable energy penetration on power system and validated the proposed data-driven method. The results show that the space dispersion and pattern number of operation modes are significantly increasing as renewable energy penetration increases. In addition, the operation mode are less correlated with season under high renewable energy penetration.