A spectral estimation method for nonstationary signals analysis with application to power systems

• We propose a new method for estimating spectral content for the power grid signals.
• We address problems relevant to instantaneous phase estimation.
• We analyze electric non stationary signals.
• We evaluate the performance of the proposed algorithm.

In this paper we analyze a new method for estimating the spectral content of frequency and amplitude modulated waveforms in the context of power line signals where, as well known, the frequency of the fundamental component (50 Hz or 60 Hz) slightly changes over time. The method, proposed here in two distinct implementations, is based on a different choice of the harmonic functions that are usually used in the DFT as a basis to analyze signals. Indeed their frequency is allowed to change over time, according to the output of an instantaneous frequency detector. The orthogonality of the basis is preserved by performing a change of reference in the time. The theory is developed in continuous time and subsequently implemented in discrete time. Simulations and experimental results are provided in order to verify and characterize preliminarily the proposed method.