Gear Fault Feature Extraction And Diagnosis Based On Vibration Information

As an important part of rotating machinery,gears with the advantages of high transmission efficiency and precise transmission ratio are widely used in engineering fields.Due to the long-term work in variable load environments,the gear teeth are prone to cracks or to be broken under the long-term stress.Once the gear fails,the transmission system will be paralyzed,and serious problems will affect the safe operation of the equipment,causing major economic losses even threatening life.Therefore,it is of great economic and academic value to be able to detect whether a gear fails in time and diagnose the corresponding failure mode.In order to achieve gear fault diagnosis,mainly research of the thesis are as follows:(1)Pretreatment of gear vibration signalsThe noise of the environment and vibration interferes with the collected gear signals,which makes it difficult to extract and diagnose the fault features of the gears.In response to this problem,the thesis proposes a noise reduction method of CEEMD combined with SSA.The CEEMD method is used to decompose the gear vibration signal.The decomposed IMFs are filtered and reconstructed based on the autocorrelation coefficient method.SSA is used to perform noise reduction on the reconstructed signal.The noise reduction of the gear vibration signal is realized,and experiments have verified that the method has a good noise reduction effect.(2)Extraction of Gear Fault FeaturesAiming at the problem of gear fault feature extraction,the thesis uses a time-frequency analysis method based on continuous wavelet transform to perform time-frequency analysis on the gear signal vibration signal after noise reduction and generate a time-frequency figure.The time-frequency figure can show the fault feature of the gear signal.(3)Diagnosis method of gear failureAiming at the problem of gear fault diagnosis,the thesis uses deep sparse auto encoder to diagnose and identify the extracted time-frequency features.Experimental results show that the application of deep sparse auto-encoder in gear fault diagnosis has good diagnostic results.