Signal-tuned spectral gabor transform as a tool for power quality monitoring
Silvia M. C. Victer, Júlio César Ferreira, José R. A. Torreão · 2018
Here we report on a work concerning signal-tuned spectral Gabor functions, and their application to power-signal disturbance analysis. The signal-tuned spectral Gabor functions allow the representation of spectral signals by means of Gabor functions whose parameters are defined by the signal's inverse Fourier transform. Basing a time-frequency transform on such functions amounts to analyzing the spectral signal by means of its own coding functions, what leads to similar properties as those of the Wigner transform. The resulting Signal-Tuned Spectral Gabor Transform (STSGT) tunes itself to the input signal, allowing the accurate detection of time and frequency events. In the experimental study reported here, the time- frequency spectrograms yielded by three different approaches (STSGT, Wigner transform and S transform) have been taken as input to a multilayer-perceptron classifier. Our results show that the STSGT, besides yielding higher-quality, more informative spectrograms, is able to outperform the Wigner and S transforms in the classification task.