Bearing Fault Feature Analysis

Time: 2018-06-27
Summary: BEARINGS are among the most important machine components in the vast majority of rotating machines and exigent demands are made upon their carrying capacity and reliability. Generally, a rolling bearing cannot rotate for ever. It often works well in non-ideal conditions, but sometimes minor problems cause bearings to fail quickly and mysteriously without any notable warning.

BEARINGS are among the most important machinecomponents in the vast majority of rotating machines andexigent demands are made upon their carrying capacity andreliability. Generally, a rolling bearing cannot rotate for ever.It often works well in non-ideal conditions, but sometimesminor problems cause bearings to fail quickly andmysteriously without any notable warning. The bearingfailures are mainly resulted from excessive wear or damage inrolling ball elements as well as in the inner/outer races of thebearing. Presently real-time condition monitoring systems forbearing systems often fail to provide sufficient time betweenwarnings and on the other hand, inaccurate interpretation ofoperational conditions may result in false alarms andassociated unnecessary costs and downtime .Traditionally, the detection of faults has become possible bycomparing the sensitive features of signals from sensors in themachinery while running in normal and faulty conditions. 

This method of the detection of faults has showedconsiderable success and several techniques have beendeveloped. The use of vibration signals is quite common inthe field of condition monitoring of rotating machinery.Analyzing the vibration signals directly in the time domain isone among the simplest and cheapest diagnosis approaches . However, as the damage increase, the vibration signalbecomes more random and the temporary statistical valuesreduce to more like that of normal bearing levels. This is the most important shortcoming of this approach . In thefrequency domain approach the major frequency componentsof vibration signals and their amplitudes are used for trendingpurposes. One of the drawbacks of frequency-domainapproaches is that they require the bearing defect frequenciesto be known or pre-estimated. The time-frequency domainapproach use both time and frequency information allowingfor the transient features, such as impacts. However, thisapproach fails to analyze the continuously smooth signal.In this paper, the powerful method of RQA is used to studyand characterize the experimental sensor signals generatedduring the normal and faulty states of the bearing under study. 

Previous:Bearings – Most Common Bearing Materials

Next:Replacement and replacement cycles of lubricants