Vibration-based fault diagnosis of dynamic rotating systems for real-time maintenance monitoring

This paper presents a diagnostic study on the monitoring, detection, and localization of faults in mechanical rotor dynamic systems. It is well known that the vital function of machine condition monitoring relies on the ability to reliably measure the "vital signs" of a machine, of which vibration is among the most important. The current diagnostic methods and other measurement and detection tools now allow for a much broader assessment of a machine's condition from its monitored vibration than simply checking the vibration level against the healthiest machines, i.e., the machine's vibration levels will exceed or have already exceeded the "normal" or the "allowable" levels based on experience for the machine. With this in perspective, this paper presents a study on the monitoring of bearing imbalance and failure through vibration analysis. The detection of imbalance and early deterioration of a bearing is highlighted by the use of a temporal quantity and the kurtosis. The proposed methodology can easily detect the most frequent imbalance and bearing faults and reflect their severity. It allows making better recommendations and maintenance decisions. It allows the development of a new form of maintenance commonly called conditional maintenance as well as the evaluation of predictive maintenance.