Research On Fault Diagnosis Method Of Variable Speeds Gearbox Based On Order Tracking

The gearbox is a key component of mechanical power transmission.The gears will produce distinct non-stationary vibration response during the meshing process under large speed fluctuations.Considering the influence of the signal transmission path,the modulation characteristics of the signal will be more complicated.Therefore,the diagnosis of mechanical faults under variable speed conditions is a hot topic of current research.By order tracking,the fault diagnosis technologies of variable speed gearbox are studied in depth with extracting vibration response characteristics and dynamic characteristics in this thesis.The main contents of the thesis are as follows:The dynamic model of gear transmission system is established,the meshing response of the variable speed gears and its modulation sideband characteristics are described.The gear fault classification is generally defined as the steady-type fault,the impact-type fault and the composite-type fault according to the geometric form or size of the fault,and the corresponding fault signal characteristics are briefly described.The vibration responses are simulated based on the calculated order tracking(COT).The characteristics of the composite-type fault are verified by experiments.In the fault diagnosis of complex gearbox system under the rotation frequency fluctuation,the meshing frequency components of the vibration signal measured interfere with each other mutually,so it is difficult to estimate gear meshing frequencies.The energy centrobaric correction method exhibits exact instantaneous frequency estimation(IFE)capability under the interference,but the energy centrobaric correction method is susceptible to strong noise.Based on peak search of the time-frequency distribution or discrete spectrum correction technique,two tacho-less order tracking techniques for gear fault diagnosis under the rotating frequency multi-linear fluctuation are proposed.Time varying filter is set based on short-time Fourier transform(STFT).Combining time-varying filtering with continuous wavelet transform(CWT),or combining time-varying filtering with energy center correction methods,to obtain smooth instantaneous frequency estimation,respectively.The instantaneous phase of reference shaft for angular resampling of the original vibration signal and the dynamics of gear meshing are obtained based on the Vold-Kalman filter(VKF).The order spectra obtained based on the measured vibration of the low-speed bearing housing by previous method and the original vibration added 1 dB Gaussian white noise by the latter method are substantially the same as the order spectra obtained by the COT respectively.The two proposed methods can be combined with other signal processing techniques to diagnose the health of the gearbox system.Consider the effect of the transmission path to the vibration signal,a fault feature extraction method based on multi-channel signals combining constrained independent component analysis(cICA),COT and Fast-Kurtogram is proposed.Rectangular wave as reference signal is established according to the instantaneous meshing frequency of the fault gear.Non-stationary signal with more obvious fault characteristics is extracted by cICA,namely the independent component of interest(ICI)in time domain,then converted to angular stationary signal.Then ICI in the time domain is converted into angular stationary signal.According to the Fast-kurtogram,the parameters of appropriate filter are selected and the ICI in the angular domain is filtered.The position of shaft,where the fault gear located,is determined by square envelope spectrum.Then evaluate health status of gearbox according to the order spectrum of ICI.Aiming at the signal generated by the low-speed broken gear under speed variations contains the impact,but the energy of the fault signal is lower.The time of the single impulse response peaks emerged within each rotation of the fault gear shaft,is determined directly with searching extreme value.It reduces the amount of calculation,as the modal parameters are calculated according to the shorter time series on the right side of the peaks time based on correlation filtering algorithm.According to the impact dictionary,the time domain impulse response component is extracted by matching pursuit.After removing the impact component from original vibration,residual signal is converted into the angular domain.The steady-type fault feature of the system is extracted from angular residual signal according to the quasi-steady modulation dictionary.Simulation and experiment show the time domain impact and angular domain steady-type fault components are reconstructed well.