ESPRIT associated with filter bank for power-line harmonics, sub-harmonics and inter-harmonics parameters estimation

Abstract

Detection of power-line frequency components and estimating their parameters are crucial tasks for maintenance of applied power quality, especially due to uncertainties originated from growing utilization of renewable power sources, hybrid energy storages and increasing demand for electrical vehicles. This manuscript introduces the association of filter bank to ESPRIT (FB-ESPRIT) for higher efficiency in detection and estimation of the harmonics, inter-harmonics, and sub-harmonics parameters in power quality measurement. FB-ESPRIT initially applies an analysis filter bank of parameter L to the signal, then individually implements ESPRIT to each of L resultant signal sub-band while it is spread over the full bandwidth of the signal. Due to individual ESPRIT implementation on L times spectrum spread of subbands, FB-ESPRIT is expected to have L times higher accuracy and robustness in parameter estimation compared to ESPRIT. Its efficiency analysis has been carried out over three case studies composed of multiple harmonics, inter-harmonics, sub-harmonic and fundamental components; a synthesized signal with under 500 Hz bandwidth, a synthesized signal with a bandwidth up to 64th harmonic in 50,000 Monte Carlo simulations (MCS) at different noise levels, and a real power signal measured at the output of a photovoltaic solar power plant in a partly cloudy windy summer morning. The comparative evaluations approve the multiple times accuracy and robustness dominance of efficiency of ESPRIT with versus without filter bank association. It has been theoretically illustrated that a FB-ESPRIT calibrated to give the same efficiency as ESPRIT has the advantage of L2 times less complexity order, as well approved via the mean result of extensive simulations runs.