Research On Fault Diagnosis Technology Of Intershaft Bearings In An Aeroengine

Aviation engine is the heart of the aircraft,in order to satisfy the requirements of the aerodynamic stability,weight reduction and increase in esteem,double rotor structure or three rotor structure are used by modern aircraft,while the rotor bearing system uses one or more intermediate bearing.The high failure rate of intermediate bearing is a dreadful hazard to the safety operation because of the large dynamic load,high temperature and the high speed inner and outer ring when it is working.Obviously,it has a great significance,for the reliable operation of the aircraft,to carry out study of the Aeroengine intermediate bearing fault diagnosis and condition monitoring technology.This paper is based on an aeroengine intermediate bearing as the object,vibration signals of normal bearing,outer ring defect and rolling element defect of an aeroengine intermediary bearing fault simulation rig were measured ands using the aircraft engine bearing fault simulation test bench to get the fault vibration signal acquisition.Cause the engine intermediate bearing signal is easily disturbed and the fault signal is weak,a diagnosing method based on minimum entropy deconvolution is proposed,which improves the signal to noise ratio and recovers the original impulse signal.The intrinsic time scale decomposition(ITD)feature extraction method of signal combined with approximate entropy is brought up.The nonlinear and nonstationary signals of intermediary bearing were decomposed to a set of inherent rotating components(PR)using ITD method,and the approximate entropy were calculated.The random forest classifier is introduced into the intermediate bearing fault diagnosis,and values of different scale approximate entropy as feature vectors were input to the random forest(RF)classifier for fault classification and diagnosis.The diagnosis effect of different classifiers on the intermediate bearing data is analyzed and compared.Finally,the test data of a certain engine are analyzed.Research shows that the minimum entropy deconvolution method can effectively remove the noise components of the engine intermediate bearing fault.Vibration signals of the ITDapproximate entropy method can accurately extract intermediary signal of fault bearing characteristic.Random forest on the intermediate bearing fault signal classification improves the accuracy of fault diagnosis.Based on the test datas of an aeroengine,it's shown that the proposed method in this paper can effectively extract weak fault signal characteristics of aeroengine mediation bearings from casing surface vibration signals.The method has a high accuracy for aeroengine intermediate bearing fault diagnosisis with a certain engineering practicability.