Fault Feature Extraction Method For Aerospace Electromechanical Equipment Based On Time-frequency Enhanced Synchrosqueezing Transform

Aerospace electromechanical equipment from launch to service,not only to experience strong impact,high temperature,and large overload,but also in the strong radiation,wide temperature changes,high vacuum external space environment for long-term operation.The operating conditions of high temperature,fatigue,corrosion,variable load,and variable speed aggravate the damage of key components such as gears,rotors and bearings.Once the failure of these critical components will likely cause the entire space equipment to be paralyzed,or even cause an irreparable space accident.Preventive mechanical fault diagnosis technology can accurately identify in advance the faults that sprout and evolve during the operation of the mechanism,which is of great significance to ensure the operational reliability and stability of aerospace equipment.In the fault diagnosis of aerospace machinery equipment,gears,rotors and bearings have always been the key research objects in this field.For gears,typical failure characteristics include periodic/time-varying transient characteristics caused by local damage(e.g.,pitting,tooth root cracking,wear,etc.)and modulation characteristics caused by time-varying mesh stiffness/load.Multiple/weak fault feature extraction and a large amount of noise interference bring challenges to gear fault diagnosis.For rotor fault diagnosis,the difficulty lies in how to extract the instantaneous frequency curve with strong time-varying characteristics from the nonlinear,nonstationary,and fast-changing speed vibration signal.For bearing fault diagnosis,the synchronous extraction and comprehensive output of compound faults at a fixed speed and the extraction of the instantaneous frequency of strong time-varying fault features at variable speeds are all difficult problems at present.To address the above problems,this paper is based on the synchrosqueezing transform,in-depth study of the underlying principles and the corresponding innovative technologies to improve the performance and applicability of synchrosqueezing transform,to provide an effective analysis tool for aerospace electromechanical equipment fault diagnosis.The specific research contents are as follows.1.Aiming at the problem of feature extraction and diagnosis of multiple/weak gear faults under strong noise interference,a dual-core denoised synchrosqueezing wavelet transform method is proposed.(1)In order to eliminate the noise interference in the fault signal and accurately extract the multi-gear fault features,a dual-core denoised method is proposed.The method firstly divides the original signal into two pre-whitening signals and pseudo-characterizing signals with their own characteristics by cepstral editing.According to the characteristics of the two signals,two corresponding EMD denoising methods are proposed to extract the periodic impact features and modulation features.Finally,a complete and pure gear fault signal is obtained by mixing the denoising results of the two sub-signals after denoising.(2)To reveal the multiple/weak fault characteristics in the pure gear fault signal,a synchrosqueezing wavelet transform is applied to this signal.(3)A multiple denoising strategy is adopted to further improve the denoising accuracy and the readability of the final time-frequency representation.Combining these three steps,a time-frequency representation showing the multiple/weak fault characteristics of the gear in the form of high energy aggregation and high time-frequency resolution is finally obtained.The multicomponent noise-containing simulation signal,the case of space pointing mechanism harmonic reducer and the case of finishing mill gearbox have fully verified the feasibility of this method,and the results show that this method can effectively extract the multiple/weak fault characteristics of gears under strong noise interference,which can provide a reliable basis for gear fault diagnosis of aerospace electromechanical equipment.2.For the difficult problem of rotor fault feature extraction under non-smooth strong time variation,multi-lifting synchrosqueezing short-time Fourier transform is proposed.This method could enhance the energy aggregation of time-frequency representation of rotor signal(fast-varying instantaneous frequency,strong noise),and has the features of signal reconstruction and low computational effort.The innovations include:(1)Constructing multiple second-order lifting operators to accurately estimate the fastvarying instantaneous frequency.In this method,phase second derivative estimation is first introduced to improve the basic estimation accuracy.Then,the estimation accuracy is gradually improved by multiple compression operators.Finally,the original banded instantaneous frequencies are compressed into energy-concentrated curves.(2)Correct the individual instantaneous frequency ambiguity points through the correction operator.That is,the problem of allocation error in individual instantaneous frequency points in the process of compression and rearrangement is corrected,so that the entire timefrequency curve is close to an ideal state.(3)A time–frequency fault measure cluster is proposed to accurately evaluate the energy concentration of the frequency of interest in mechanical fault diagnosis,and this indicator is used as the optimization target of the whole method.The nonstationary strong time-varying bat signal and the rotor fault signal of the heavy oil catalytic cracking unit fully verify that this method can provide favorable support for the research on aerospace rotor fault diagnosis.3.Aiming at the problem of synchronous extraction of bearing single/compound faults under different speed types,the bearing fault diagnosis is divided into two cases: variable speed and fixed(small fluctuation)speed.(1)For the variable speed case,the multi-lifting synchrosqueezing short-time Fourier transform in Chapter 4 is introduced to extract the time-varying fault characteristics of the bearings.(2)For the fixed(small fluctuation)speed case,a time-frequency energy aggregation spectrum diagnosis method based on multisynchrosqueezing wavelet transform is proposed.Firstly,through multisynchrosqueezing wavelet transform,the high energy aggregation time-frequency spectrum is obtained,which lays the foundation for accurately selecting the demodulation band of the bearing compound fault.Secondly,the optimal demodulation bands of each fault are accurately selected by the energy aggregation spectrum index,which provides an effective diagnostic tool for compound fault features extraction and identification.Finally,multiple optimal demodulation bands at constant speed are output simultaneously.The two methods are integrated to develop a bearing fault diagnosis system for aerospace electromechanical equipment,which not only realizes the synchronous extraction of single/compound faults of bearings under different rotational speeds,but also realizes synchronous extraction of bearing compound fault features under the same rotational speed.Finally,the validity of the proposed methods is verified by the variable/constant speed experimental bearing data,which provides a basis for the theoretical research on bearing fault diagnosis of aerospace electromechanical equipment.