A Dual Monitoring Technique to Detect Power Quality Transients Based on the Fourth-Order Spectrogram

Abstract

This paper presents a higher-order statistics-based approach of detecting transients that uses the fourth-order discrete spectrogram to monitor the power supply in a node of the domestic smart grid. Taking advantage of the mixed time–frequency domain information, the method allows for the transient detection and the subsequent identification of the potential area in which the fault takes place. The proposed method is evaluated through real power-line signals from the Spanish electrical grid. Thanks to the peakedness enhancement capability of the higher-order spectra, the results show that the procedure is able to detect low-level transients, which are likely ignored by the traditional detection procedures, where the concern pertains to power reliability (not oriented to micro grids), and this, by analyzing the duration and frequency content of the electrical perturbation, may indicate prospective faulty states of elements in a grid. Easy to implement in a hand-held instrument, the computational strategy has a 5 Hz resolution in the range 0–500 Hz and a 50 Hz resolution in the range of 0–5 kHz, and could be consequently used by technicians in order to allocate new types of transients originated by the distributed energy resources. Four real-life case-studies illustrate the performance.

Keywords

• higher-order statistics
• micro grids
• power quality (PQ)
• spectral kurtosis (SK)
• spectrogram
• time–frequency analysis
• transient detection