Modeless reconstruction of missing synchrophasor measurements

This paper presents a new framework of reconstructing missing synchrophasor measurements (erasures) without the modeling of power system dynamics. Leveraging the approximate low-rank property of phasor measurement unit (PMU) data, we connect the problem of reconstructing PMU data erasures with the recent advance in low-rank matrix completion methods. Erasures can be reconstructed through existing computationally-efficient algorithms such as singular value thresholding (SVT) and information cascading matrix completion (ICMC). Numerical experiments on actual PMU data are conducted to verify the effectiveness of the proposed method. Since existing analysis for matrix completion methods assumes an independent-erasure model that does not capture the correlation among PMU data erasures, we propose two models to characterize temporal correlation and channel correlation in PMU data erasures. We provide theoretical guarantees of the ICMC algorithm in reconstructing correlated erasures in both models.