Reliability Analysis Of Aviation Engine Spindle Bearing Based On Vibration Signal

As a key supporting component,aviation engine spindle bearing has always been in a complex and harsh working environment for,so it has the characteristics of high failure rate and easily damaged.So the traditional reliability analysis methods and maintenance strategies could not meet the equipment requirements.Therefore,fault diagnosis and life prediction analysis methods based on on-line monitoring emerge was proposed under the requirement,which could provide basis and decision-making for maintenance before the failure occurred,and it would greatly reduce the occurrence of catastrophic accidents.Firstly,the FMECA of an aero-engine spindle subsystem was carried out,and the qualitative analysis of the aero-bearing was carried out by FTA.Several kinds of failure modes with the most harmful impact and several failure reasons with the greatest structural importance were pointed out,which could provide the reliability optimization design for the aero-engine bearings and spindle subsystems.Then,this paper elaborated the testability of spindle bearing faults in depth,which established the relationship between the vibration mechanism and fault features,and transformed the key problem of spindle bearing fault diagnosis into the problem of fault-feature extraction based on vibration signals.After introducing the conventional methods of vibration signal analysis and multi-domain feature extraction,a new fault diagnosis method based on EMD-SVD and fuzzy neural network was proposed in this paper to extract and identify the early fault features of spindle bearings accurately.Firstly,vibration signals in known states was decomposed by empirical mode decomposition(EMD)to obtain the intrinsic mode functions(IMF)containing the main fault feature information.Then,the key IMF was operated by singular value decomposition(SVD)as the initial vector matrices,which was used as the input samples of the fuzzy neural network.The fault feature matrix was learned and trained by FNN,and then the diagnosis and recognition of spindle bearings in different states were realized in numerical value.On this basis,the output value of FNN was normalized to health index(HI),so we could describe the performance degradation state of the bearing by the distance between the test sample and the normal sample,and then the performance degradation curve of the spindle bearing in the whole life cycle was established.Finally,the rolling bearing fault diagnosis and life prediction system was developed by using the interface design function of MATLAB GUI.The experimental analysis proved that the method proposed in this paper could extract and identify the early fault features of spindle bearings accurately and steadily,and improved the accuracy of fault diagnosis,then it provided a basis for the condition-based maintenance(CBM)of the same type of aviation bearings.