Use of the Stationary Wavelet Transform to Characterize Transient Events in DC Power Distribution Systems

摘要

In real-time load monitoring systems, load transient behavior is often used to detect and classify load shifts in the plant. For this application, the wavelet transform allows for both time and frequency localization information to address the shortcomings of the Fourier transform in transient analysis. This paper uses a shift invariant form of the discrete wavelet transform, the stationary wavelet transform, to detect transient events on a simulated DC power system. The wavelet modulus maxima of the transformed signal are then used to distinguish between a previously characterized load transient and a faulty injected signal. The method is verified using simulation of a DC power system with a pulsed load and with the starting transient of a brushless DC motor to a commanded speed.