Research Of Diagnosis System For Synchronous Wind Turbine Based On Support Vector Machine

With the global energy tension and energy crisis,wind power has attracted extensive attention world wide with its advantages of cleaness and safety.Wind power industry has also grown fast.Wind turbine fault diagnosis technology can effectively improve it’s reliability, reduce operating costs and improve the efficiency of wind farms,which has a great significance for easing global energy crisis.The structure and working principle of wind turbine is introduced at first.The common failure and mechanism of wind turbines is enumerated.Some typical methods about caodition monitioring of wind turbine is summarized.A fault diagnosis program for direct-driven synchronous wind turbine by current signal analysis is presented, according to the wind turbine research and development direction.By collecting the imbalance fault current signal of the experimental model and using wavelet packet analysis tool the feasibility of current signal analysis for direct-driven synchronous wind turbine fault diagnosis is proved.The advantages of signal analysis and processing is introduced then.The advantages and disadvantages among expert system,BP neural network and support vector machine have been analyzed and compared.At last, according to the system design requirements and the advantages of wavelet packet analysis and support vector machine(SVM), a fault diagnosis system is designed,which is based on current signal analysis. It focuses on SVM inference module of the fault diagnosis system, using MATLAB software to modify parameters of support vector machine module and obtain the simulation results of wind turbine fault diagnosis system. The results show that support vector machine as the inference module of ditect-driven synchronous wind turbine fault diagnosis system has a good diagnostic result.Researchs in this paper have important reference value of ditect-driven synchronous wind turbine fault diagnosis.Some questions about direct-driven wind turbine fault diagnosis system remain to perfect and futher study is summarized.