Matching And Synchronous Strategy Based Time-Frequency Representation Enhancement For Fault Diagnosis Of Critical Rotating Components Of Machinery

Rotating equipments such as gearbox,shaft box and wheelset are the critical components in trail vehicles and automobiles,whose operating conditions are directly related to the running quality and safety of vehicles.The mechanical equipment of rotating machinery and vehicles contains key rotating components such as bearings and gears.The service environment is harsh and the operating conditions are time-varying,which in turn makes failures occur frequently.According to statistics,30%of the failures of rotating machinery and 44%of the failures of large asynchronous motor are caused by bearings;according to incomplete statistics,gearboxes of high-speed trains operating in WuhanGuangzhou,Beijing-Shanghai have accumulated multiple cases such as bearing fracture because of overheating and gear tooth surface pitting problems before 2014.Therefore,it is of great significance to carry out health monitoring and fault diagnosis research on the key rotating parts in machinery and vehicle equipment for its operational safety guarantee and intelligent operation and maintenance.Fault diagnosis based on vibration signal processing is one of the most effective ways.Due to the time-varying operating conditions,the vibration signal often exhibits strong nonstationarity.Time-frequency(TF)analysis(TFA)is an effective method to deal with nonstationary vibration signals in that it can reveal TF domain information simultaneously.However,the weak fault related features are easy to be annihilated by strong background noise and interferences,leading to problems such as ambiguity in time-frequency representation(TFR),poor readability and fault characteristics cannot be identified.Therefore,from the perspective of improving the TF aggregation level to improve the readability of the TFR,and for the purpose of extracting weak fault features to perform fault diagnosis of the key components of the equipment,this thesis deal with the principle of matching and synchronous based TF enhancement.From the perspective of matching the nonlinear changing pattern of the frequency trajectories simultaneously and refining the instantaneous frequency(IF)ridges,the nonstationary signal processing methods based on generalized demodulation(GD),linear chirplet transform(LCT)and time-frequency reassignment methods are further studied,aiming to improve TFA performance in addressing TF clustering,signal reconstruction and synchronous extraction of weak fault features to successfully perform fault diagnosis of key rotating components.Funded by the National Natural Science Foundation of China ’Research on the Feature Extraction and Diagnosis of Bearing Faults under Variable Speed Operating Conditions’(No.51605319),the specific research contents are summarized as follows:The demodulator construction in generalized stepwise demodulation transform(GSDT)needs to first extract the instantaneous frequency of the signal in advance.Then instantaneous frequency synchronized generalized stepwise demodulation transform(IFSGSDT)is proposed.When the estimated angle can accurately describe the trend of frequency changing pattern,the stepwise demodulation transform can first demodulate the signal energy forward to a constant frequency(paralleling to the time axis)and has the highest time-frequency aggregation level.Vibration signals often contain a high level of noise.In order to improve the performance of the proposed method against heavy noise,the improvements are:(1)The time-efficiency of the algorithm is improved without considering the change of the window length;(2)The IF is iteratively estimated based on the obtained TFR to correct the estimated angle parameter to further obtain a successively accurate TFR.In addition,considering the readability of the TFR,on the basis of matching linear chirplet transform(MLCT),timefrequency reassignment is performed on the TFR with enhanced features in the frequency dimension so that the TF energy is more concentrated around the IF ridge.The final TFR is still suitable for processing harmonic multi-component signals,and the energy is more concentrated,which is closer to the ideal TFR.Starting from the frequency modulation property of strongly time-varying signals,matching its changing trend and constructing improved matching instantaneous frequency estimator,making it closer to the true frequency of the signal,iterative matching synchrosqueezing transform is then proposed.It has been proved that the successive instantaneous frequency estimation results after N iterations are more close to the true instantaneous frequency as the iteration goes.And then TF distributions reassignment according to the synchrosqueezing transform method,and simultaneously has the reversible nature and the high TF aggregation of the traditional TF reassignment method.In summary,this article takes the key rotating parts of the equipment such as bearings and gears as the object,and studies the matching and synchronous enhancement strategy for the purpose of health monitoring based on signal processing.According to the extracted features,fault diagnosis can then be performed.Aiming at dealing with the nonstationary characteristics of the vibration signals of the key rotating parts of the equipment,following matching and synchronous enhancement strategy,the adaptive construction of the demodulator,the adaptive matching of the chirp-rates and the strategy of the IF estimator are studied,which aim to match and enhance the nonstationary features of the analyzed signal.Faced with the "multi-component and heavy noise" problem of the vibration signal of the key rotating parts of the equipment,synchronous enhancement strategy is then proposed,where the expansion of the kernel function and the iterative matching synchrosqueezing strategy are studied via synchronous enhancement of multi-component fault characteristics and optimization of IF estimator under heavy noise.These two strategies have their own characteristics and cooperate with each other.Finally,the enhancement and extraction of the nonstationary fault features of the key rotating parts of the equipment under variable speed conditions and heavy noise are established,and the fault diagnosis is then realized.This content provides a theoretical basis for the monitoring of equipment health status and the implementation of intelligent operation and maintenance.