The Research On The Early Fault Diagnosis Of Wind Power Gear Box Based On EMD And SVM

Wind energy is an abundant natural resources. It is developing with a high speed in the world for the reasons that it is renewable, low cost and no pollution. In China wind power generation technology is started lately, but developing rapidly. Previous wind turbines all rely on import, since the overall design and manufacturing technology of foreign advanced wind turbines is introduced into China in the 90 years, we have made the optimization and innovation based on the digestion and absorption. In recent years, the failures of the major components in statistics are the gearbox, generator and blade and the gearbox failure incidence increases year by year, the percentage is more than 60% which is the highest rate of unit failure occurred in parts. Gearbox condition monitoring and fault diagnosis should be made.The main forms for the early failure of Wind turbine are gearbox-wear, tooth-bonding, contact- fatigue and broken-teeth. And the fault diagnosis method can be used with domain method, frequency-domain method, envelope analysis, order analysis, cepstrum, three-dimensional holographic maps and spectra. These methods have their own advantages, but none of them can be a comprehensive and efficient way to judge the fault signal.This article discusses how to use the EMD and SVM method to achieve the early gearbox fault identification, and the experiment is made to verify the feasibility of the method. The first step of the experiment is collecting data in the normal state, the slightly wear condition and the mild wear state that are separately under the conditions of 10Hz, 15Hz, 20Hz and 25Hz,.secondly we calculate the EMD based on MATLAB platform to gain the sensitive component and then calculate the RMS value of the sensitive ,at last we use the support vector machines to classify the faults. From the experimental results, the correct classification rate is very high, the method is feasible.To be able to combine with the actual project, this paper uses a combination of EMD and the SVM method to develop a set of fault diagnosis and identification system based on LABVIEW, thus a new way of thinking is introduced for the fault diagnosis .