含高渗透并网变流器电力系统宽频强迫振荡机理及监测方法研究

Wind power, solar photovoltaics and other renewables are mostly integrated to power system through grid-connected converters with high flexibility, low inertia and strong nonlinearity. Hence, the disturbance source of forced oscillations varies, the complexity of forced oscillation mechanism raises and the band of frequency widens with time-varying characteristic, posing a serious threat to the safe and stable operation of power grid. Aiming at the wideband forced oscillations in power system with high penetration of grid-connected converters, this project will carry out an in-depth study from the following aspects:1) to fill in the gap of ignoring the interactions among disturbance sources, grid-connected converters and the grid, the multi-time scale coupling power system model suitable for the analysis of forced oscillations will be constructed; 2) propose the analysis method for forced oscillations based on magnitude/phase motion dynamics and harmonic balance method, breaking through the limitation of mechanism analysis based on linearized model in electromechanical timescale, revealing the mechanism of forced oscillations from the aspects of linearity and nonlinearity comprehensively; 3) propose the monitoring method of wideband forced oscillations based on topology data analysis, solving the problem of detecting and identifying wideband forced oscillations with the emphatic characteristics of the nonstationarity, nonlinearity and time-variation. The results of the research hopefully will reveal the complex mechanisms of wideband forced oscillations in power system with high penetration of grid-connected converters, also providing important theoretical basis and effective methodological support for monitor and analysis of wideband forced oscillations.

风电、光伏等新能源多采用具有高灵活性、低惯性、强非线性特点的并网变流器接入电网，导致强迫振荡扰动源多样化、激发机理复杂性提高、频率范围拓宽且时变性显著，给电网安全稳定运行造成了严重威胁。本项目针对含高渗透并网变流器电力系统宽频强迫振荡问题，拟从以下几方面开展深入研究：1)弥补传统忽略扰动源-并网变流器-电网交互影响的不足，建立适用于宽频强迫振荡分析的电力系统多时间尺度耦合模型；2）提出基于幅相动力学和谐波平衡法的宽频强迫振荡机理分析方法，突破现有基于机电尺度线性化模型机理分析的局限性，从线性和非线性角度全面揭示振荡机理；3）提出基于拓扑数据分析的宽频强迫振荡监测方法，从高维振荡数据空间拓扑分析的角度解决非平稳、非线性和时变性特征突出的宽频率范围强迫振荡检测和辨识难题。研究成果有望揭示含高渗透并网变流器电力系统中宽频强迫振荡的复杂发生机理，并为其监测与分析提供重要的理论基础和有效的方法支撑。

风电、光伏等新能源多采用具有高灵活性、低惯性、强非线性特点的并网变流器接入电网，导致强迫振荡扰动源多样化、激发机理复杂性提高、频率范围拓宽且时变性显著，给电网安全稳定运行造成了严重威胁。本项目针对含高渗透并网变流器电力系统宽频强迫振荡问题开展了以下研究：.1）研究了调制算法中直流母线电压动态对直驱风机功率振荡的影响，明确了是否可以忽略机侧部分的前提是调制算法中是否考虑了直流母线的动态过程。.2）提出了一种基于阻抗分段降阶模型零点辨识的风电并网系统稳定性分析方法，所提方法显著降低了高阶系统的稳定性分析难度，在实际含大规模风电并网的电力系统中具有较好的适用性。.3）提出了一种基于对角化LDPC压缩感知和k-近邻算法的广域系统宽频振荡监测方法和一种基于自编码器与LSTM的宽频振荡信号压缩及其扰动源定位方法。能够实现在现有带宽下的宽频振荡信号数据传输，提高振荡检测的快速性和准确性，实现广域系统宽频振荡源的定位分析。.4）提出了一种基于多维特征及ReliefF-mRMR的低频振荡类型判别方法。通过ReliefF-mRMR提取多维特征中的最优特征子集，并采用一种基于遗传算法(GA)优化关键参数的改进支持向量机(SVM)，提高了低频振荡类型判别的准确性。.5）提出了一种基于两阶段深度迁移学习的强迫振荡扰动源分层式定位方法。将定位问题转化为图像识别问题，并在此基础上提出了一种两阶段深度迁移学习(DTL)算法来定位强迫振荡扰动源。该方法提高了振荡源定位的快速性和准确性，能够在有效减轻数据通信压力的同时，满足电力数据保密性的要求。.本项目研究内容和所提方法为电力系统宽频振荡的分析和监测提供了有意义的参考，其中部分方法正在进行实用化推广。