Hybrid Method of AFDD With ECOD for Oscillation Modal Extraction From Synchronized Ambient Data

Modal parameters extraction from synchronized ambient data for small-signal stability evaluation (SSSE) plays a crucial part in power system oscillation events prevention. While the randomness and uncertainty of large-scale renewable energy bring risks of error to extraction results. This paper proposes a hybrid method of automated frequency domain decomposition (AFDD) with empirical cumulative distribution-based outlier detection (ECOD) for modal parameter extraction from synchronized ambient data. First, the “automated peak picking” algorithm of AFDD is proposed to guarantee the accuracy of modal parameters extraction by avoiding the clumsy manually and inefficiently peak selection process. Then, ECOD is proposed with the potential to further enhance the reliability of bi-dimensional modal parameters including both frequency and damping ratio through outlier detection and elimination. The effectiveness of the proposed method is studied based on the simulation data of IEEE 16-generator 5-area system and synchronous ambient data of real power system, where the maximum steady-state error of the parameter extraction results was only 0.035, proving the accuracy and reliability of the proposed method.