The Research Of Fault Diagnosis System For Wind Turbine Generator Gearbox Based On Data Mining

Wind turbine is the key equipments of the wind farm, the distribution area is wide,the number is large, the working condition is poor, and is aways far from themonitoring center, in the event of failure, caused by the loss and consequences isincalculable. According to statistics, the failure of the wind turbine is mainlyconcentrated in the gearbox, generator, blades and others, the gearbox has the highestfailure rate of the unit, and the failure rate is increasing year by year. Therefore, inorder to ensure the safe and stable operation, it is necessary to research the effectivecondition monitoring and fault diagnosis for the wind turbine gearbox.At present, there are many researchs about the monitoring system and its variouslinks’ improved of wind turbine. And there are some general condition monitoringsystems applicate to the field of wind power, but most of the analysis and diagnosticcapabilities of the existing monitoring system are relatively weak. According to set thecollection point threshold and collection data trend analysis, only to see its overalllevel in the period of time, when certain values exceed the alarm limits, and it only canmake an initial fault diagnosis.As the wind turbine produces vast amounts of datas in the daily monitoringprocess, these datas imply a large number of potential rules, and meet the necessaryconditions for data mining. Based on the properties reduction capability of rough setand rapid classification of C4.5decision tree algorithm, this paper presents animproved rough decision tree model, combined with SQL Server2005and VB and usethe modular design ideas united developed a wind turbine gearbox fault diagnosissystem. Through comparative anaysis, the results show that the improved model notonly can effectively reduce the workload of obtaining the characterization datas andcan quickly and accurately realize the fault identification, with the more engineeringpracticality than using the C4.5decision tree algorithm directly.The fault diagnosis system based on rough decision tree of the wind turbinegearbox developed in this paper realize the three-dimensional comprehensivemanagement, not only can management the system’s databases, but also realize thefault monitoring and diagnosis rapidly.