Blind extraction and characterization of power system oscillatory modes

• Identifying and extracting electromechanical modes from simultaneous measured data is an important aspect in power systems.
• A BSS method to identify modal components and complex mode shapes from measured data is proposed.
• The BSS method is based on second order statistics.
• We propose an analytic interpretation of the time-lagged covariance matrix.

Wide-area monitoring substantially improves modal identification and characterization under noisy conditions. This paper discusses the use of blind source separation (BSS) techniques to extract and identify modal responses and mode shapes from simulated data. The method has advantages over other global analysis methods in that it allows for the analysis of both, transient and ambient data and is thus well suited for global system monitoring of power system oscillatory behavior using wide-area measurement systems (WAMS) data.Methods for analysis of complex datasets using BSS techniques are developed, and a physical explanation is offered. The developed procedures are tested on simulated data, and the impact of various parameters such as noise and time lags on the quality of signal separation are examined.