Wavelet-based data compression of power system disturbances using the minimum description length criterion

This paper introduces a compression technique for power disturbance data via discrete wavelet transform (DWT) and wavelet packet transform (WPT). The data compression leads to a potential application for remote power protection and power quality monitoring. The compression technique is performed through signal decomposition up to a certain level, thresholding of wavelet coefficients, and signal reconstruction. The choice of which wavelet to use for the compression is of critical importance, because the wavelet affects reconstructed signal quality and the design of the system as a whole. The minimum description length (MDL) criterion is proposed for the selection of an appropriate wavelet filter. This criterion permits selection not only of the suitable wavelet filter but also the best number of wavelet retained coefficients for signal reconstruction. The experimental study has been carried out for a single-phase to ground fault event, and the data compression results of using the suitable wavelet filter show that the compression ratios are less than 11 % and are reduced to more than a half of that percentage value by implementing an additional lossless coding.

I. Introduction

The transients due to ground faults, load switchings, and other disturbances may cover a broad frequency spectrum in the order of kilohertz to megahertz. A single captured event recorded for several seconds using monitoring instruments can produce megabytes of data. As a result, the volume of the generated and maintained data increase significantly, which lead to a high cost in storing and transmitting such data. Therefore, it is necessary to develop an effective compression technique which has capability to reduce the volume of data necessary for storing and to speed up the transmitted data for remote monitoring [1] [2] [3].