A Performance Evaluation for Parameter Estimation Method Suitable for Machine Noise Analysis

Abstract

The parameter estimation of a signal containing multiple random periodic signals is applicable to a variety of fields such like the identification of the number of machines through the analysis of vibrations or sounds. However, it is not easy to achieve it with low computational load. A computationally-efficient approach termed as the accumulation for real-time serial-to-parallel converter (ARS) has been proposed for vital sensing, just as a computationally efficient parameter estimation method for a composite signal of multiple periodic signals. Although its computational efficiency has been analyzed, in this paper, the performance of ARS in terms of the identification of multiple random periodic signals is investigated compared with the fast Fourier transform (FFT).