Fault Diagnosis Method With Considering The Usability Of Condition Monitoring Data For Rolling Bearings

Rolling bearing is an important part of high-speed train and rotating machinery.And it is easy to be damaged.It operates under the condition of high speed and heavy load for a long time.The contact stress makes it prone to failure.Once the bearing fault occurs,it may bring great economic losses or even safety accidents.And the rolling bearing failure is the main cause the failure of equipment.Therefore,monitoring the health status of rolling bearings is of great significance in ensuring the safe operation of high-speed train and improving the reliability of rotating machinery.The rolling bearing faults will increase with time.The incremental learning method is suitable for appling to the fault diagnosis of rolling bearing.Incremental learning method can update the fault diagnosis model to fit the new fault type.This paper focuses on fault diagnosis of rolling bearing based on incremental learning method.The main contents and innovation points are as follows:(1)The frequency,amplitude and phase of vibration signal collected at variable speed will be modulated by both of the rolling bearing failures and the rotary speed,which lead to the problem of low identification accuracy of rolling bearing fault diagnosis based on incremental learning at variable speed.This paper solves this problem from the perspective of data preprocessing by adopting four different data preprocessing methods: rotary speed normalization,Z-Score normalization,uniform angular sampling and power spectrum density.Finally,it is found that the power spectrum density can improve the identification accuracy of rolling bearing fault diagnosis based on incremental learning at variable speed.(2)The data used by specific fault diagnosis method must contain a large amount of information fit to the fault diagnosis method.In order to solve the problem that the signal acquried by abnormal signal acquisition system is unusable for the specific fault diagnosis method,the concept of data usability based on the fault of the signal acquisition system is proposed.The incremental learning method is successfully applied to fault diagnosis of signal acquisition system by designing and carrying out relevant experiments,and a high identification accuracy is obtained.(3)After the fault diagnosis of the signal acquisition system is completed,the system is combined with the fault diagnosis of the rolling bearing.The usability of data collected is concideried before falut diagnosis of rolling bearing.The results verifies the importance of fault diagnosis of the signal acquisition system before the fault diagnosis of the rolling bearing.