Autocorrelation approaches to time frequency analysis of machinery vibration signals

This paper discusses the technical aspects of three well-known time frequency analyses of machinery vibration signals. It begins by reviewing the short time Fourier transforms. It then considers the Wiener Kinchine theorem, which relates the autocorrelation function to the square of the Fourier transform magnitudes. The symmetrical autocorrelation function leads to the Wigner distribution, and the interference terms that compromise its usefulness are discussed. Finally the reduced interference distributions (most notably the Choi and Williams distribution) that greatly reduce the interference terms are presented. Three example analyses of machinery data, along with transient simulations illustrate results that can be expected.