M. Glickman and P. O'Shea (Australia)
Windows, Prony, and least-squares average.
This paper describes a new Fourier based method for estimating the damping of decaying exponential modes that occur in a distributed power system as a result of an electric disturbance. It is a “sliding spectral window” technique which builds on the work in [1], [2] and [3]. The key innovation in the new method is the use of multiple orthogonal windows rather than just a simple rectangular window (as used in [1]) or a simple smooth window (as used in [2] and [3]). The use of these multiple orthogonal windows allows least-squares averaging strategies to be performed, enabling lower variance estimates to be obtained as a result. Achieving lower variance estimates for damping is important because the more accurate the estimation is, the more quickly damping anomalies (which correspond to major electrical disturbances) can be detected. Simulations are presented which show that the use of multiple orthogonal sliding windows coupled with least squares estimation gives better results than are obtainable with a single window. The new method has a particular advantage at moderately low signal to noise ratios (SNRs).
Important Links:
Go Back
