基于矩阵填充的电力系统功角曲线采样信息轻型化方法

互联电力系统发生扰动，振荡模式间非线性相关作用强时，功角曲线振荡模式具有时变性，但各时间阶段所含主要振荡模式分量少，且低频振荡模式是影响系统动态特性和稳定性的重要内在因素。文中提出一种基于矩阵填充(MC)实现电力系统功角曲线采样信息轻型化的方法。由最优随机采样矩阵对应的位串形成确定性采样矩阵，采样少量数据，采用奇异值阈值算法恢复功角曲线，该方法不受系统阶数影响，且具备较强的数据处理能力，可用于低频振荡在线分析。算例分析结果表明，MC采样方法相对传统奈奎斯特采样方法和压缩感知采样方法，具有采样数据量少、占用存储空间少等优点，验证了MC采样方法实现功角曲线采样信息轻型化的可行性和有效性。

Matrix Completion Based Method for Realizing Lightening of Sampling Information of Power Angle Curve in Power System

When disturbance occurs in interconnected power systems, and the nonlinear correlation between oscillating modes is strong, oscillation modes of power angle curve are time-varying. However, the main oscillation modes in each time stage are small and the low frequency oscillation modes are important intrinsic factors affecting the dynamic characteristics and stability of the system. So a method for realizing lightening of sampling information of power angle curve in power system based on matrix completion is put forward. A deterministic sampling matrix is formed by the bit string that corresponds to the optimal random sampling matrix to sample a small amount of data, then singular value threshold(SVT)algorithm is used to restore the power angle curve. This method is not affected by system order and has strong data processing ability. It can be used for on-line analysis of low frequency oscillation. The results of numerical example show that matrix completion method has less data than traditional Nyquist sampling method and compressive sensing sampling method, and occupies less storage and so on. The feasibility and validity of matrix completion method for lightening of sampling information of power angle curve are verified.