Analysis of Changes in the Characteristics of the Multifractal Spectrum During Diagnosing the Technical Condition of Rolling Bearings

This paper presents a study conducted with a view to analysing the sensitivity of measures determined based on the characteristics of the multifractal spectrum when detecting damage in rolling bearings. The research was conducted using vibration signals of a real rotating machine shaft supported by bearings, each of which was in a different technical condition. The effects of damage-induced changes in vibration signals on the calculated waveforms of multifractal spectra and the measures determined from them were analysed. In addition, qualitative and quantitative analyses of the effect of additional signal interference on changes in the characteristics of the spectra were carried out. The study shows that the application of vibration signal processing using a multifractal spectrum enables fast and unambiguous demonstration of changes occurring in the signals that can be linked to damage of bearings. Irrespective of the type of bearing damage symptoms present in the vibration signals—whether high-energy quasi-correlated impulse or medium and low-energy stochastic—an increase in measures determined from the multifractal spectrum was identified. The study shows that, regardless of the level of vibration signal interference, the calculated quantitative measures of the spectra were indicative of damage to rolling bearings.