Adaptive filtering, wavelet and lapped transforms for power quality problem detection and identification

This paper focuses on digital signal processing techniques applied to power quality problem detection and identification. The detection and identification are based on a previous estimation of the fundamental component of the event under analysis by using the adaptive notch filter and Kalman filter for frequency, amplitude and phase estimations, respectively. After that, the non-stationary type signal, obtained by the subtraction of the estimated fundamental component from the acquired signal, is submitted to the discrete wavelet transform (DWT) and modulated lapped transform (MLT) for further parameters extraction. A set of information or event signature obtained from the transient event is comprised of several parameters extracted from the DWT and MLT domains, the estimated parameters of the fundamental component as well as THD of the transient event. The results obtained from real database report that the combination of the adaptive signal processing techniques, DWT, MLT and parameters extraction techniques provide a reliable signature of the power quality problems.