Application Of Spectral Kurtosis Based On Choi-Williams Distribution In Gearbox Fault Diagnosis

This paper proposes a gearbox fault diagnosis study based on the Choi-Williams distribution of spectral kurtosis(CWD-SK)and hidden Markov model(HMM),which realizes the feature extraction of the initial gear failure in the gearbox.The paper uses CWD-SK as a feature,which is applied to the HMM.Then classify the gear fault features.Finally,the actual data is used for simulation to verify the effectiveness of the algorithm.The main content of this article includes vibration signal preprocessing,fault extraction and fault status classification.The specific work is as follows:1.Firstly,the basic structure and common failure types of ZS-65 type gearbox are introduced in detail,including the common failure modes of gears and diagnosis method based on vibration signals.Firstly,local mean value decomposition(LMD)is performed on the collected gear vibration signals to obtain several PF component signals.By analyzing each PE component,the PF component with obvious fault characteristic information is selected.2.Perform CWD-SK calculation on the selected PF component,identify the initial fault as early as possible.The initial fault information is not easily found in the original signal.When the CWD-SK value is greater or equal to 3,it is considered that the device has started to have an initial failure.Compared with the original algorithm,it reduces the amount of calculation,improves the difficulty of selecting the window function,and has the disadvantages of cross-interference.3.Through the gearbox fault experiment,based on CWD-SK for fault feature extraction,HMM is used as a recognizer.Furthermore,the fault feature sequence extracted by CWD-SK is input into the HMM in turn for fault identification.Among them,the maximum likelihood estimation is used to distinguish the types of faults generated during the operation of the equipment.The method is compared with the support vector machine(SVM)fault classification model.Finally,the effectiveness of the combination of HMM and CWD-SK to extract the fault feature sequence is fully verified,which can better classify the fault conditions in the gearbox accurately.