Research On Lubrication Condition Monitoring And Fault Diagnosis Of Large Hydro-generator Thrust Bearing

With the rapid growth of the national economy,the demand for electricity is also increasing,and a stable and reliable power generation system is becoming more and more important.China is rich in hydropower resources,the installed capacity of hydropower continues to increase,and the unit capacity of hydropower units continues to expand.It is of great significance to improve the operation reliability of large hydro-generator,reduce the maintenance cost of large hydro-generator,and avoid the occurrence of major accidents.As one of the main components of large hydro-generator,thrust bearings are particularly important for their status monitoring and fault warning.In recent years,with the continuous development of online oil monitoring technology,the use of oil monitoring technology for fault diagnosis of equipment has been applied.In order to realize the condition monitoring and intelligent remote operation and maintenance of the thrust bearing of the large hydro-generator,the online oil monitoring sensor is used to collect the data of the thrust bearing lubricating oil,and the online oil condition monitoring system is constructed to realize the thrust bearing fault warning and fault identification.The main research contents of this article are as follows:First,based on the theory of hydrodynamic lubrication,the oil film pressure and oil film thickness distribution of the thrust bearing under normal operating conditions were calculated,the characteristics of the minimum oil film thickness change under different speeds and loads were analyzed,and the approximate range of the oil film thickness was determined.The oil film thickness monitoring scheme based on the eddy current displacement sensor is studied,and verifing its effectiveness and reliability.Subsequently,an online oil data collection box was designed to monitor the physical and chemical indexes and pollution degree of the lubricating oil of the thrust bearing in real time.The overall structure of the large hydro-generator thrust bearing oil online monitoring system was formulated,and the installation of eddy current sensors and online oil data collection box was completed.Secondly,the fault detection and early warning methods of thrust bearing lubricating oil are studied.For the problem of uneven distribution of online oil monitoring data,the time series model and principal component analysis method are used to model the data during normal operation and the model residuals are constructed.The process control charts of residual and square prediction error SPE and Hotelling T~2 integrate three control charts for fault detection and early warning,which can effectively realize anomaly detection and reduce the false detection rate,and can identify abnormal variables.Once again,a thrust bearing friction and wear test bench was designed to perform wear tests on the thrust bearing under different working conditions and lubrication conditions,and the changes of the thrust bearing thrust pad temperature and friction coefficient during operation was analyzed,and various parameters were obtained for the thrust bearing.The effect of performance summarizes the relationship between different lubrication states and thrust bearing failures.In order to estimate the wear life of the thrust bearing,a periodic start-stop wear test was conducted to study the relationship between its life and the number of start-stops.Finally,the fault characteristics and fault mechanism of the thrust bearing of the large hydro-generator are analyzed,the common fault types of its thrust bearing are summarized,and the oil fault data set is collected and classified.The feature extraction method of online oil monitoring data is studied.In view of the shortcomings of the K-nearest neighbor algorithm,the K-nearest neighbor algorithm based on genetic algorithm is used for fault diagnosis.A fault recognition ball set is constructed to improve the accuracy and calculation speed of fault recognition.The self-adaptability and scalability of the diagnosis method are guaranteed,so that the system can identify new faults.Using Python and Qtcreator to develop the thrust bearing lubricating oil condition monitoring software system,it has completed the functions of thrust bearing condition monitoring,fault detection and early warning,and fault identification,and realized the large hydro-generator thrust bearing lubrication status monitoring and intelligent remote operation.