Smart Grid Seminar

Phasor measurement units (PMUs) can provide synchronized phasor measurements of remote points in the power system at a sampling rate of 30  samples per second or more.  Since the DOE smart grid investment program started in 2009, the number of PMUs in the North American power system has increased tremendously. The collection of high-dimensional PMU measurements will be futile if they are not supported by efficient data management and information extraction methods.  

Compressed sensing theory and low-rank matrix theory show that the acquisition, storage, and processing of high-dimensional data can be much simplified if the data exhibit certain low-dimensional structures such as sparsity and low-rankness. Due to the wide existence of low-dimensional models, compressed sensing and low rank methods have been applied to medical imaging, computer vision, collaborative filtering, etc. 

The focus of this talk is to draw a connection between PMU data management and low-rank methods. After observing the low-rankness of spatial-temporal blocks of PMU measurements in Central New York Power System, we propose a common framework of leveraging low-rankness for multiple PMU data management tasks. Specifically, I will talk about the recovery of missing PMU data and the detection of cyber data attacks and show the theoretical and numerical results of the proposed low-rank methods.