Signal Selection for Oscillation Monitoring With Guarantees on Data Recovery Under Corruption
Signal Selection for Oscillation Monitoring With Guarantees on Data Recovery Under Corruption
复制标题
DOI：
10.1109/tpwrs.2020.2993196
复制
发表时间：
2020-11
影响因子：
6.6
通讯作者：
K. Chatterjee;N. Ray Chaudhuri;G. Stefopoulos
中科院分区：
文献类型：
--
作者：
K. Chatterjee;N. Ray Chaudhuri;G. Stefopoulos
见解发展到分组PMU信号，以保证稀疏损坏下的数据恢复。导出了用测量窗口组成信号的模态可观测性来表示子空间密度的解析关系。结果表明，通过最小化各弱阻尼模态的相位角和观测值幅值的变化对信号进行分组，可以最小化测量窗口的数值秩，增强子空间的密度，并有助于实现基于鲁棒主成分分析的信号重建方法保证精确恢复的充分条件。这些见解被构建为信号选择的引理和命题，并通过来自IEEE测试系统的综合数据以及来自美国公用事业公司的现场PMU数据进行验证。
Insights are developed into grouping PMU signals for guaranteeing data recovery under sparse corruption. Analytical relations are derived to express the denseness of the subspace spanned by a measurement window in terms of the modal observabilities of its constituent signals. It is shown that grouping signals by minimizing variation in phase angles and amplitudes of observabilities for each poorly-damped mode minimizes the numerical-rank of the measurement window, enhances denseness of the subspace, and helps in attaining the sufficiency condition guaranteeing exact recovery using Robust Principal Component Analysis-based signal reconstruction methods. These insights are structured into lemmas and propositions for signal selection and are validated on synthetic data from IEEE test systems, as well as field PMU data from a US utility.