Fault Based On The Analysis Of EMD&RVM Wind Turbine Online Monitoring Method Of Research

With the rapid development of wind power currently, installed capacity of wind turbineare also growing and wind turbine in the course of its operation in the event of serious fault,may cause damage to the wind turbines themselves, can also lead to the entire wind turbinepower supply interruption, even affect the local or a power grid operation. Therefore, it needstudy wind turbine online monitoring system to understand real-time of wind turbineoperation condition and timely find wind turbine hidden fault trouble.Firstly, this article in view of the direct drive wind turbines bearing vibration signal isusing the Fourier transform (FFT) and empirical mode decomposition (EMD) to carry on theanalysis comparison, and then combine with the relevance vector machine (RVM) on normaland bearing fault classification. The results of experiment show that this method is more rapidand effective than using RVM method for bearing signal classification. The main researchcontents are as follows:(1) According to the simulation of wind turbine bearing vibration signal, the normaloperation and abnormal operating conditions of bearing, respectively, by using Fouriertransform (FFT) and empirical mode decomposition method (EMD) carries on the analysiscomparison. The analysis results show that the EMD method of spectrum can reflect theactual operation of the unit.(2) According to the simulation of wind turbine bearing vibration signal, respectively,the research uses the RVM method for wind turbine bearings in data analysis. Analysis resultsshow that RVM can classify the data between normal and abnormal of the bearing.(3) The method of EMD extracts the vibration of the bearing characteristics of windturbine information, and the bearing characteristics of the normal working state can be used asa benchmark for training in the RVM signal. According to the condition monitoring guideline,selecting of1.25times the benchmark as a threshold, can accurately classify unit bearing vibration signals to judge.Aiming at1.5direct drive wind turbine bearing vibration signal is stationary nonlinearsignal. This paper proposes a based on the research of the EMD with RVM method. Using theMATLAB software simulation, the results of research show that the method can quickly andeffectively judge bearing state of operation condition. This kind of analysis method of windturbine online monitoring provides a new train of thought.