| With the complex structure,the failure rate of aero-engine is high.The fault recognition method based on vibration signal analysis is one of the most effective means to identify structural faults of aeroengine rotor system(AERS).However,many weak fault information is coupled together and submerged in strong noise,which brings great challenges to the fault diagnosis of AERS in the early stage of faults or composite faults.In this paper,the fault characteristic frequency is extracted by spectrum analysis method to realize fault recognition by studying the multiwavelets,noise reduction theory and fault signal separation method.Aiming at the influence of threshold selection on multiwavelet denoising effect,the threshold setting strategy has been improved based on the existing multiwavelets denoising methods on the neighboring coefficient theory and translation invariant theory.Then,modified TI-MWT-NCD method has been proposed.Combininng the standard deviation of high-frequency coefficients,energy ratio of multi-dimensional coefficients and neighboring coefficients,and making full use of the advantages of hard threshold and soft threshold,the threshold can be set adaptively according to the measured signal.Through analyzing the simulated signal of impact fault with noise,this method has proved to have better noise reduction performance and can achieve high correlation between the processed data and the pure impact signal.In order to solve the problem that the preprocessing of traditional multiwavelets transform makes the input signal redundant or destroys its orthogonality,a hybrid method based on maximum correlation kurtosis deconvolution(MCKD),multiwavelets and Hilbert transform demodulation analysis(HTDA)has been proposed.The original vibration signal with fault is divided in two by MCKD transform twice.Then multiwavelets transform is used to further enhance the noise reduction effect.As the postprocessing of MCKD,multiwavelets transform can reduce effectively the effect on fault pattern recognition by the parameters of MCKD.There are two problems.One is that the fixed wavelet basis function is often unable to match the fault characteristics.The other is that the parameters of MCKD can’t be preset in practical application.AMCKD-MWT-HT method has been proposed based on HTDA,adaptive MCKD and multiwavelets in view of the two problems.According to the relationship between the fault characteristics of the tested system and the parameters of MCKD,the parameter ranges are preset.Considering the periodic shock characteristics in time domain and cyclostationarity in frequency domain,a compound optimization criterion based on maximum correlation kurtosis and square envelope spectral entropy is proposed.Then,the parameters of MCKD and multiwavelets basis functions have been set adaptively by the criterion.ABMWT-AMCKD-HT method based on adaptive balanced multiwavelets,adaptive MCKD and HTDA has been proposed to solve the fault recognition problem of compound fault.Considering the energy fluctuation and periodicity of vibration signal with fault in time and frequency domain,a compound optimization criterion based on square envelope spectral entropy,the square root and entropy of the square envelope signal,is prosed to construct flexibly balanced multiwavelets.Based on the composite index composed by square root and entropy of square envelope signal,square root and entropy of square envelope spectrum,the spectrum value of characteristic frequency and octave frequency,the optimal setting of T and M in MCKD are realized.The vibration displacement signal on the vibration test bench has been analyzed by modified TIMWT-NCD method,MCKD-MWT-HT method and AMCKD-MWT-HT method.The vibration acceleration signal on aero-engine rotor fault tester has been analyzed by modified TI-MWT-NCD,MCKD-MWT-HT and ABMWT-AMCKD-HTmethod.Compared with other methods,the effectiveness and superiority of the proposed method have been verified.In AERS fault pattern recognition,the research in this paper has a certain theoretical significance... |
PDF Full Text Request