Singular point detection of time-varying power events using filter banks and adaptive filters

The detection of disturbances in power signals is important to prevent damage to electrical machines. The disturbance waveforms contain singular points where events have occurred. This paper presents a method to detect singular points in a power quality disturbance waveform using filter banks and adaptive filters. The proposed method decomposes the input disturbance signal using a filter bank system of non-tree structure. The non-tree filter bank is designed to separate high frequencies to reduce spectral leakage. An adaptive filter is then used to predict the decomposed signal. The error signal from the adaptive filter becomes the residual signal. The energy of the error signal is used to detect singular points.