Research On Wind Turbine Gearbox Fault Diagnosis Based On Support Vector Machine

With the rapid of developing of science and technology in China, modern equipment is also toward the development of highly integrated and precision, so the safety of equipment become the key to the normal production. Gear box is a key components of the wind turbines, the gearbox is installed at a narrow place, once the malfunction, it is very difficult to repair it. For the wind turbine gearbox vibration signal exist a problem of small sample, and it is nonlinear and non-stationary, this paper combined with the algorithm of support vector machine, research and analysis the wind turbine gearbox fault diagnosis pattern recognition, model parameter selection and other aspects.This paper first introduces the basic theory and method of Support Vector Machine, and analyzes the basic principles and algorithm of the Least Squares Support Vector Machine in the classification problem. Then combed the characteristics of Support Vector Machine in the pattern recognition problems, the wavelet analysis and Least Squares Support Vector Machine are applied to the fault pattern recognition of wind turbine gearbox.The parameters of classification model with Support Vector Machine play a role in the classification effective. The genetic algorithm makes use of its global search ability to put forward a parameter selection method, and it is used in the classification model of Least Squares Support Vector Machine. And then the classification model is applied to the wind turbine gearbox vibration signal pattern recognition. The experimental results show that the Support Vector Machine for wind turbine gearbox fault diagnosis with good results, shown in its superiority in handling of small sample problem.