A low-quality PMU data identification method with dynamic criteria based on spatial–temporal correlations and random matrices

•Considering the spatial–temporal correlation of random matrices, a system operating state analysis method based on the single-ring theorem and correlation coefficient analysis is proposed to provide the basis for low-quality PMU data identification to determine the reasons for PMU data mutations.•Starting from singular value decomposition and reconstruction, a low-quality PMU data identification method based on the spatial–temporal correlation between original data and low-rank reconstructed data is proposed under different operating states.•The physical meanings of singular value decomposition are analyzed, and a dynamic threshold selection method is adopted for low-quality PMU data identification criteria based on the spatial–temporal correlation to improve the identification accuracy.