Research On Fault Feature Extraction Methods For Gears Of Aero-engine Accessory

The accessory gearbox plays an important role in power transmission in the aero-engine,and the health of its internal gears have a great impact on the safe operation of the gearbox.When the gear in the accessory gearbox fails,the accurate rotation speed is difficult to obtain and the vibration signal can only be collected from a limited number of measuring points on the housing of the gearbox.The fault feature of the gear is easily covered by the vibration of other parts and noise,which greatly reduces the success rate of fault diagnosis.Therefore,the research on fault feature extraction methods is carried out for gears of aero-engine accessory in this paper.The main contents are as follows:(1)To address the problem of harmonic vibration extraction of gearbox gears without rotation speed measurement,time synchronous averaging(TSA)based on the cross-power spectrum is proposed for feature extraction.The original vibration is intercepted by equal length based on the estimated fundamental frequency.The phase calculation signals of each segment and the corresponding reference cosine sequence are defined.The segments’ phases are detected from the cross-power spectrum and converted into the number of sampling points in the discrete domain for cyclic translation to compensate for the phase differences caused by rounding and frequency estimation.High synchronization between signal segments for average can be obtained.The research results shows that the proposed method can basically eliminate the accumulation of phase error caused by incomplete period interception,and realize the effective extraction of gear harmonic vibration.Compared with the typical method,the accuracy of the proposed method is higher.(2)Aiming at the issue that the fault feature of gearbox gears under multi-component vibration interference and complex path transmission is weak,a feature extraction method based on signal reconstruction and optimized pulse enhancement is proposed.Through the screening of empirical mode decomposition(EMD)results by an index that can effectively measure the strength of pulses,the reconstructed signal is obtained,and the component with less fault information is removed.The reweighted kurtosis and whale optimization algorithm are combined to optimize the fault period in the multipoint optimal minimum entropy deconvolution(MOMEDA),and then the optimal pulse enhancement signal can be obtained.The research results show that the proposed method can effectively eliminate the extraneous interference in the broadband vibration,avoid the feature weakening caused by the inaccurate prediction of the fault period,and complete the directional enhancement and extraction of the weak fault pulse of the gearbox gear.The performance of the proposed method is better than the typical methods.(3)The weak gear fault feature in the complex vibration signal of gearbox housing may not be significant after pulse enhancement.To cope with this challenge,a feature extraction method based on optimized pulse enhancement and envelope synchronous averaging is proposed.Firstly,the optimal pulse enhancement signal is located by signal reconstruction and fault period optimization.Then the envelope segments of equal length are cut from the envelope signal,and the phase detection,conversion and compensation are completed based on the cross-power spectrum of the phase calculation signal and the corresponding reference cosine sequence.Finally,the harmonic components related to the fault are extracted by the average process of synchronous envelope segments.The research results show that envelope synchronous averaging can highlight the characteristic frequency in the optimal filtered signal,and the gear fault diagnosis is realized after comparing it with the theoretical value.