Detection of vibration disturbances during the analysis of long realisations of vibration signals

Abstract

Creation of systems of proactive technical maintenance of the equipment requires conduction of a detailed study of changes in vibration condition of technical objects in the course of their use. Stationary systems of vibration control and monitoring in their work are guided by the standard criteria and signal about the appearance of emergency dangerous situations or carry out protection functions. However, the identification of the reasons for changes in technical condition of the controlled object is not their main function. Modern computer or plug-in systems allow reception of long (tens of seconds, minutes, hours, days) realizations of vibration signals which reflect vibration condition of the tested mechanism in various regimes without interruption. The visual analysis of a number of such vibration signals showed the existence of rare vibration eruptions and disturbances which are almost unidentified through the vibration signal parameters and spectral analysis. The method of processing of vibration signals is presented, based on their decomposition into periodic and noise-type components and wavelet analysis which focuses on detection of rarely created vibration eruptions and disturbances. These vibration anomalies indicate the origin of defects and their study is directed on the perfection of the means of technical diagnostics of mechanisms according to the vibration parameters.