A time domain approach for the analysis of non stationary signals in power systems

Voltage and current waveforms produced by several power systems are often characterized by a wide spectrum that contains numerous components, namely the fundamental, harmonics, and interharmonics. They can be properly analyzed through discrete Fourier transform (DFT)-based methods only if they remain stationary in time; in fact, in the presence of nonstationary components, DFT-based approaches do not grant reliable results because of the absence of a well-defined periodicity interval. The digital signal processing method proposed in this paper aims at analyzing a large category of nonstationary signals observed in power systems. It is capable of analyzing waveforms produced by power systems that seemingly operate in steady-state conditions, in which time-varying components represent anomalies, as well as waveforms generated by some power apparatuses, which supply nonstationary waveforms by design. The wide range of applicability of the proposed method is granted by a suitable processing scheme, including distributed filtering and phase unwrapping operations. Distributed filtering operations are performed to distinguish and separately analyze the major components contained in complex waveforms. Phase unwrapping operations are needed in the presence of nonstationary components characterized by heavy drifts and/or large swings in frequency.