AN ALGORITHM TO FIND THE TWO SPECTRAL LINES ON THE MAIN LOBE OF A DFT

Abstract

Using interpolation methods to find the true harmonic component of a signal on an inter-line position in the spectrum is a usual practice in frequency estimation. The disadvantage of this method is that the points on which the interpolation is made have to be on the main lobe. Most researchers consider the maximizer and its biggest neighbor to satisfy this requirement, but in many cases this presumption is false. We propose an algorithm that permits finding the two points on the main lobe of the spectrum a certain number of samples. The evolution of the maximizer indicates which neighbor belongs to the main lobe. The algorithm is tested successfully for signals generated with known frequency and amplitudes. By estimating accurately the natural frequencies, the cracks can be detected in early stage by vibration-based methods.