Research On Rolling Element Bearing Time-varying Nonstationary Fault Characteristics Extraction Under Time-varying Rotational Speeds

Rolling element bearings are one of the most prevalent components in rotating machines.Due to their complicated and severe working conditions such as time-varying speed,heavy load and high temperature,which all result in the inevitable performance regressions.As a result,time-varying nonstationary fault characteristics extraction of rolling element bearing is the key step of health monitoring and fault diagnosis of mechanical equipment,also important to ensure operation safety and reliability,reduce economic losses and avoid catastrophic accidents.It is difficult for traditional methods to extract bearing time-varying nonstationary characteristics,because of their theoretical defects,unsatisfactory results and depend on rotating speed measurement.In this dissertation,the characteristics of bearing single fault signal,multi-fault signal and faulty gear interferential signal is explored from the perspective of bearing vibration performance.On this basis,the improved computed order tracking and non-resampling algorithms have been developed.Focused on the heavy calculating pressure of resampling time marks of the even-angle sampling calculation in the process of angle domain resampling,the improved envelope resampling algorithm is proposed based on the equal division impulse.First,the faulty bearing signal and rotating speed signal are measured.Then,the time marks of every rising edge of the rotating speed pulse and the corresponding amplitudes of faulty bearing vibration signal are determined.Furthermore,every adjacent rotating pulse is divided equally,and the time marks in every adjacent rotating speed pulses and the corresponding amplitudes of vibration signal are obtained by the interpolation algorithm.Finally,all the time marks and the corresponding amplitudes of vibration signal are arranged and time marks are transformed into the angle domain to obtain the resampling signal.speed-up and speed-down faulty bearing signals are employed to verify the validity of the proposed method,and experimental results show that the proposed method is effective for diagnosing faulty bearing.Furthermore,the traditional order tracking techniques are applied to the experimental bearing signals for comparison,and the results show that the proposed method produces higher accurate outcomes in less computation time.Focused on the problems of stationary hypothesis of the instantaneous rotating speed and interpolation errors,the generalized demodulation transform based time-varying nonstationary fault characteristics extraction of rolling element bearing is developed.Both the traditional computed order tracking and the improved resampling algorithm cannot avoid stationary hypothesis of the instantaneous rotating speed and interpolation errors.Hence,the generalized demodulation transform is introduced.In the proposed method,the phase function set is firstly calculated using rotation speed signal and fault characteristic coefficients of the target bearing.Then,the interested time-varying fault characteristic component is transformed into a stationary one by applying the generalized demodulated transform to the envelope of the filtered signal.Finally,the reset fault characteristic component is quantitative expressed using the fast Fourier transform.The studies of simulated and experimental faulty bearing signals indicate that the proposed method is effective to diagnose faulty bearing under time-varying rotational speed without angle domain resampling.Focused on the problems of irrelevant interferences and rotating speed measurement devices,the time-varying nonstationary fault characteristics extraction method of bearing based on the rotating frequency and fault characteristics reset is developed.First,the higher amplitude time-frequency ridge is extracted from envelope time-frequency representation of filtered signal using the harmonic summation-based peak searching algorithm.The extracted time-frequency ridges can be used to calculate all possible rotating frequency functions,and its phase functions and frequency points.Then,make assumptions about the fault types of the extracted time-frequency ridges.Furthermore,the envelope signal is demodulated using the iterative generalized demodulation transform,and one of the assumptions is verified based on the demodulated spectra.Finally,the envelope signal is further demodulated and the other faulty types are determined.Both simulated and experimental results validated that the multi-fault feature of rolling element bearing under time-varying rotational speed can be effectively identified.In the proposed method,the time-varying mult-fault characteristics and rotating frequency are reset and quantitative expressed using the iterative generalized demodulation transform and fast Fourier transform.On the one hand,the rotating speed is a reference to determine the fault type,but not important tool to resampling signal;On the other hand,the rotating frequency is quantitative expressed,but not extracted,as a result,reduce the demand of high resolution ratio and high noise immunity of time-frequency algorithm.Focused on the problems of time-varying nonstationary multi-fault characteristics separation and quantitative expression of bearing and gear,a Vold-Kalman generalized demodulation based multi-fault detection method is developed.Specifically,the fault characteristic components are firstly extracted using the Vold-Kalman filtering.Then,the extracted time-varying nonstationary fault characteristic components are reset using the generalized demodulation transform.Furthermore,the reset fault characteristic components are quantitative expressed based on the fast Fourier transform.Finally,integrated spectra for determining localized faults are obtained,where the spectra are calculated by repeating the above demodulation and filtering processes based on the phase functions of the harmonics.The phase function set and frequency points containing fault indexes of gears and bearing are constructed based on the instantaneous dominant meshing multiple function and mechanical parameters,and the time-varying instantaneous dominant meshing multiple can be extracted from the time-frequency representation of the raw signal.Both the simulation and experiment results show that the proposed method can effectively separate and extract fault features of gearbox and bearing under variable speed.In addition,it is free from band filtering,angular resampling and rotational speed measurement,and also can achieve higher performance than that of the band-pass filtering algorithm.