一种基于数据的电力电子高渗透电力系统中多种功率振荡监测和分析方法研究

A rapidly growing number of power electronic converter (PEC)-connected systems such as various renewable energy sources, energy storage systems, plug-in electric vehicles, FACTS and HVDC have penetrated into the conventional power grid, which enhances the grid flexibility and controllability, but also considerably complicates the dynamic behavior of both transmission and distribution networks. The PEC system, on one hand, participates in the existing conventional power oscillations and makes the problems more complex; On the other hand, it also brings some new types of oscillations among various converters or between the converters and weak power network with different voltage levels (e.g., under/over frequency oscillations are observed in wind farm-connected systems, the dynamics of which are associated with PECs rather than the system or wind turbine. How exactly such oscillations happen and how to damp them remain an open problem for grid connections of wind power). Both aspects above have posed a very critical threat to power grid operation and thus deserve a careful examination. . However, large quantities of PEC systems with complex structures, time-varying parameters and ‘black-box’ controllers not only make the system modeling nearly impossible, but also drastically increase the dynamic model dimensions and thus the computational burden especially for a resource-constrained real-time operational environment. Hence, the above ‘new features’ of the modern power grid have raised an enormous challenge to system operators in analyzing the oscillatory stability by using the traditional model-based analysis. The fast development of the wide area measurement system (WAMS) and big data analytics techniques has enabled the data-based analysis and provided an effective solution to the above-mentioned challenge. To date, the data-based analysis has been applied in many aspects of system dynamic security assessment including estimating the oscillation modes and tracking the oscillation energy dissipation. However, most relevant research efforts have been devoted to investigating the conventional power oscillations in the transmission network only without considering the impact of high penetrating converters. In addition, most of current data-based methods and techniques perform like a black box, which lack a deep physical explanation on the problems. . Hence, this project aims to firstly develop two novel data-based key techniques, namely the WAMS-based local energy flow dissipation monitoring technique and machine learning-based system damping flow distribution as well as data pre-processing and knowledge extraction technique. Then by combining these two techniques together, a model-independent monitoring and analyzing method with the capability of bearing incomplete and false measurement is established to simultaneously visualize and investigate the system oscillations with a clear physical understanding for the first time and locate the oscillation sources. Since whatever oscillation problem are essentially related to energy flow fluctuation, the proposed method can deal with various types of oscillation problems in transmission and distribution networks including new emerging or potentially undiscovered oscillation modes brought by PECs. On this basis, efficient preventive and emergency control strategies are proposed to suppress the various oscillations appearing in the modern smart grid. The project will undoubtedly bring significant benefits to secure planning and operation of the future smart grid characterized by large-scale integration of converter-controlled components in China.

大量涌入的电力电子换流器连接的系统（后称为PEC系统）在提高电网灵活性和可控性的同时，也使电网动态日益复杂。PEC系统一方面参与现有功率振荡，另一方面又带来了存在换流器间及换流器和各级弱电网间的新振荡模式，给电网安全稳定运行带来了严重的威胁。PEC系统结构复杂、参数时变和控制黑箱化对传统基于模型的振荡稳定性分析提出巨大挑战。广域测量系统（WAMS）和大数据分析的迅速发展使基于数据的分析成为可能，给解决上述挑战提供了契机。本项目研发两种基于数据的关键技术（即基于WAMS的局域能量流衰减监测和基于机器学习的系统阻尼流分配及数据预处理），并结合这两种技术形成一套全新的振荡监测分析方法，首次将振荡分析可视化，定位振荡源并对振荡机理提供清晰的解释，从而提出有效的控制策略。该方法可以应对多种振荡问题，处理缺失和错误数据。项目无疑将对大规模PEC系统接入为标志的我国未来智能电网的安全运行有着重大的意义。

以新能源为主体的电力系统是我国实现双碳目标的必然选择。传统电网逐步向“高比例新能源”和“高比例电力电子”的“双高”新型电网转变，多种新的振荡问题涌现，给电网稳定运行带来了极大挑战。因此，本项目系统地研究了“高比例电力电子”新型电力系统中振荡监测、机理分析和抑制技术，并提出了新的理论分析方法研究系统中出现的复杂振荡交互问题。..首先，针对传统基于模型方法局限，研究并推广了基于测量的能量流分析法理论，将其应用于含高比例电力电子的新型电力系统的各种振荡监测和分析中。针对以同步机为主体的传统多机系统从理论上严格证明了所提出的EFA与阻尼转矩分析法（DTA）以及模态分析法的一致性，结论对任意同步机模型都适用。在此基础上，将上述研究推广到含大量电力电子变流器的新型电力系统中，对全功率变流器的新型振荡模式以及振荡模式之间的交互进行了定量分析，通过历史实测振荡数据分析各点聚合的阻尼转矩系数、特征根实部以及特定参数之间的关系，进而提出了一种在线阻尼调控方法。..其次，为了更深入地理解电力电子化电力系统中新型振荡问题，帮助推广能量流分析法，提出了模式叠加理论，建立了双开环子系统动态模型，并提出了一种基于双向DTA的模式位移评估方法来量化研究变流器与外部电网之间的模式交互，以及基于PMU测量的在线谐振监测控制。..最后，为应对测量数据质量和安全问题，提出了一些基于大数据技术的预处理方法，增强数据鲁棒性，为系统运行员提供更准确的振荡监测分析打下基础。..本项目所提出的基于能量流的振荡监测分析方法可以避免复杂的系统建模，有效解决模型不可得或维数灾的问题，在高维的电力电子化电力系统中有广阔应用前景；所提出的新型振荡分析方法理论给出了深入的振荡机理解释，帮助将所提出的振荡监测分析从同步机为主导的输电网推广至电力电子化的输配电网中，应对多种新型振荡问题；提出的针对测量数据质量和安全的预处理技术，有效提高了振荡监测分析的准确性和鲁棒性。