Proportion-Extracting Chirplet Transform for Nonstationary Signal Analysis of Rotating Machinery

Abstract

Multicomponent nonstationary signals with close instantaneous frequencies (IFs) are commonly encountered in rotating machinery condition monitoring and fault diagnosis. It is very challenging to accurately reveal the physical natures of such signals. To address the issue, this article presents the proportion-extracting chirplet transform (PECT). In the PECT, the proportional kernel functions can match with the time-varying patterns of all constituent components. This enables the PECT to eliminate the artifacts caused by spectral overlaps in short-time windows and to achieve fine frequency resolution. Meanwhile, the corresponding proportional frequency shifting operators ensure satisfactory time resolution. Therefore, the PECT provides high time-frequency resolution and suffices to characterize the time–frequency structures of nonstationary signals with close IFs. The experimental and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> measured data analysis results of typical rotating machinery validate the effectiveness of the PECT.