Research On Fault Early-warning Method In Gearbox Of Wind Turbine Based On SCADA Data

There are many faults caused by unstable operating conditions of wind turbine,among which gear box fault is more common in wind turbine with double feeder gear box and will bring huge economic losses to the turbine every year.In order to find faults in advance and effectively reduce the economic losses caused by faults,it is necessary to predict the fault trend of gearbox.According to the fault trend,the early warning analysis is carried out to realize the early detection of the fault of the unit.It is necessary to evaluate the running state of gearbox and make the maintenance decision according to the evaluation results for the maintenance of the failure indicated by the early warning analysis.In the selection of maintenance decision,the importance order of various maintenance schemes is carried out,which contributes to making effective maintenance decision and reducing the loss of unit failure shutdown.In this paper,the typical faults of wind turbine gearbox occurred during the actual operation of UP 1500 double-feed wind turbine in Longyuan Electric Power Tianjin wind farm are studied,in order to reduce the shutdown of fan alarm fault,the paper studies the typical failure of wind turbine gearbox in combination power production.Improve the unit economy to make a contribution.In this paper,the relevant parameters of the unit are described in detail,the fault mechanism of the gear box is analyzed,and the knowledge carding and fault tree analysis of the typical faults are carried out.Secondly,taking the typical fault mode gear box oil temperature high fault as an example,based on the actual operation of the unit scada(supervisory control and data acquisition)record data every 10 minutes to extract the characteristic parameters.Eight scada parameters with strong correlation with gearbox oil temperature are used as input parameters to build the prediction model.Step-by-step regression model,grey model and temporal neural network model were used to predict oil temperature.In order to improve the prediction accuracy,the information fusion prediction with weighted combination of the three prediction models is carried out,which effectively improves the accuracy of the prediction results and effectively reduces the error of the prediction values.then the threshold value of the fault early warning is determined according to the control parameter set by the unit and based on the statistical process control method spc(statistical process control),and the fault early warning analysis of the gear box is realized according to the predicted data,And the validity of the early warning analysis is verified through the normal operation data and the fault data in the actual running data of the unit.In order to determine whether the fault of early warning prompt needs to be repaired immediately,this paper analyzes the running state of gearbox based on fuzzy comprehensive evaluation,and provides reference for maintenance work according to the result of evaluation.Finally,in order to make the maintenance work more targeted and timely,the improved analytic hierarchy process(AHP)based on the quantitative analysis method of fault tree is used to analyze the typical faults of the unit.Based on the improved analytic hierarchy process(AHP),which combines the traditional tomographic analysis method and the expert experience method,this paper analyzes the importance of the bottom events that cause the gearbox fault tree,and provides the most effective maintenance decision for the maintenance personnel in the event of failure.It effectively improves the efficiency and scientificity of maintenance.