Method of vibroacoustic signal spectrum optimization in diagnostics of devices

Abstract

The vibroacoustic method of device diagnostics is most often based on spectrum analysis of digitally recorded vibration signal of their mechanical elements. In the steady-state, depending on the test object, the signal may contain only fundamental frequency and harmonics. Then, in case of poor mechanical condition, diagnostic signal contains also some foreign frequencies. Using signal of finite length for computing its spectrum causes some deviations in spectrum structure. It can be the reason for misinterpretation because of array frequencies occurring in DFT, which in fact do not exist in the signal and existing frequencies, which have incorrect amplitude. To reduce this effect, the window function can be used. Taking into consideration the special spectral structure of diagnostic signals in the steady-state, it is possible to adjust parameters of signals in such a way to eliminate any distortion for regular frequencies. For this purpose, if further analysis requires a fixed number of signal samples, by signal resampling, spectral resolution can be adjusted to the harmonics existing in the signal, especially, when processed are signals already recorded with unwanted sampling frequencies.