A hybrid unscented filtering and particle swarm optimization technique for harmonic analysis of nonstationary signals

Abstract

This paper presents a modified unscented Kalman filter for accurate estimation of frequency and harmonic components of a time-varying signal embedded in noise with low signal-to-noise ratio. Further, the model and measurement error covariances along with the unscented Kalman filter parameters are selected using a modified particle swarm optimization algorithm. To circumvent the problem of premature convergence and local minima, a dynamically varying inertia weight based on the variance of the population fitness is used. This results in a better local and global searching ability of the particles, which improves the convergence of the velocity and better accuracy of the unscented Kalman filter parameters. Various simulation results for nonstationary sinusoidal signals with time varying amplitude, phase and harmonic content corrupted with noise, reveal significant improvement in noise rejection and speed of convergence and accuracy in comparison to the well known extended Kalman filter.