基于随机矩阵理论的配电网时空大数据研究

Spatial-temporal Big Data Analytic Framework for Distribution Network Based on Random Matrix Theory

现代电力系统是迄今为止最复杂的工程系统,数据驱动方案将在未来电网特别是配电网分析中扮演重要角色。大数据模型和分析方法是当前电力大数据有效利用方案设计的核心和瓶颈。从数据的视角来重新审视当前配电网的发展及所面临问题,利用近年来高维统计的基础突破,提出基于随机矩阵的配电网时空大数据研究框架。旨在形成一套系统性的电力系统数据驱动方案—原始数据、预处理操作、随机矩阵建模、随机矩阵分析、指标提取和可视化、辅助决策,为系统的优化运营决策提供支撑。

The modem power gird is one of the most complex engineering systems. Data-driven solution, rather than model-based one, is more suitable for future girds, especially for distribution networks. Data modeling and data analysis are two core parts and key bottlenecks for efficient data utilization. From data aspect, the developments and challenges are reviewed. With the latest breakthrough in high-dimension statistics, a spatial-temporal data analytic framework is proposed for distribution network based on random matrix theory. This work aims to form a data-driven methodology-raw data, data modeling, data analysis, indicators-for operators＊ decision making.