An Approach to Recognize the Transient Disturbances With Spectral Kurtosis

Abstract

Spectral kurtosis (SK) is a statistical tool, which can indicate both non-Gaussian components in a signal and their locations in the frequency domain. The computation methods of SK decide the characteristics of transient disturbances. We propose a new computation method based on Choi-Williams distribution to recognize the transient disturbances. Through the comparisons and analysis of different computation methods for five transient disturbances, the proposed method can better reflect their characteristics difference. In addition, the relationships between SK and transient disturbance parameters (amplitude, phase, duration, noise, and so on) are analyzed and discussed in detail, and a recognition plan is proposed. We, respectively, adopted the simulation signals with random disturbance parameters and different noise, power systems computer aided design, and real-life signals to prove the recognition performance. The experimental results show that the method is feasible and simple, and the recognition rate of five transient disturbances is high and satisfied. Especially, the transient impulsive and oscillation with noise can be completely recognized using the proposed method in this paper.