Transient signal detection using higher order moments

Abstract

The asymptotic performance of transient detection based on higher order moments is theoretically derived, invoking the asymptotic normality of the decision statistics, and a detector based on the third-order absolute moment is proposed. The analysis shows that for transient duration less than half the observation window, the proposed detector outperforms the detectors based on the second- and fourth-order moments for a wide range of SNR values. Computer simulations assess the applicability of the theoretical analysis. © 1999 IEEE

Detection of Disturbances in Voltage Signals for Power Quality Analysis Using HOS

“…The disturbance detection process involves several variables that depend on the kind of application. For instance, the starting point d could be known or not; the duration L could be available or not; the wave shape of the disturbance (u + Δf ss ) L can be known, partially known, or completely unknown [19,22,23]. In this context, H 0 is a simple hypothesis and H 1 is a composite hypothesis.…”

“…Additionally, the higher-order-based cumulants provide more relevant information from the random process. The use of such relevant information for detection purpose and other applications such as parameters estimation and classification have been successfully investigated in several applications [16,17,[23][24][25] which are not related to power systems. Based on this discussion and assuming that v, f, and u carry out relevant information from the disturbance occurrence, then the hypotheses stated in (12) are reformulated as follows:…”

“…Some contributions have demonstrated that HOS-based techniques are more appropriate to deal with non-Gaussian processes and nonlinear systems than second-order-based ones. Remarkable results regarding detection, classification and system identification with cumulant-based technique have been reported in [16,17,22,23,25]. Assuming that components f and u of voltage signals are modeled as a non-Gaussian process, the use of cumulant-based technique appears to be a very promising approach for detection of abnormal behaviors in voltage signals.…”

Detection and parameter estimation of a transient signal using order statistics

“…If the signal part is known (up to few unknown parameters), it is often referred to as "parametric signal" [10]. If the signal s is random, it is referred as a "nonparametric signal" [1].…”

Weak transient signal detection based on continuous wavelet and cumulant analysis

“…This paper presents a simpler yet effective weak Signal dcrection method by thresholding the 41h order cumulants of the continuous wavelet transform (CWT) hased on the Morlet wafelet, with the threshold derived from the Neynian-Pearson criterion. Also presented in this paper is the investication carried out to compare the performance with other detection methods based on the CWT only, and based on other joint time-frequency transforms, namely, the short-time Fourier transform (STFT) and the Wigner-Ville distribution (WVD) [3], with and without cumulant analysis.…”