Frequency Estimation of Multifrequency Signals Based on the 3-Point and 4-Point Spectrum Interpolation for Short Measurement Time in PV Systems

Multifrequency signals occur e.g. in photovoltaic systems where the estimation of frequency is very important to drive the inverter. This paper presents an comparison of the estimation method based on the FFT procedure, MSD time windows and the 3-point/4-point spectrum interpolation. Results show that more accurate (up to several times) is the method with three points of the spectrum taken into calculations. Simulations were performed for the tested signal without disturbances, for harmonic components and for AWGN noise for various measurement parameters. The method is very accurate and very fast (even below one cycle of the tested signal). The accuracy for the 3-point interpolation is approximately $2\times 10^{\wedge}-8 Hz/Hz$ for 128 samples in the measurement window and 1.2 cycle of the signal. Obtained results can be very useful because they show the accuracy in various measurement conditions and also show that the method with three points of the signal spectrum is optimal in the tested range up to two measured signal cycles.