Research On Fault Diagnosis Of Wind Turbine Drive System

Our wind power capacity has been at the forefront of the world, however, with the increase in installed capacity of wind turbines and commissioning time, the wind power system equipment are also emerging problems. Frequent accidents or equipment failure would cause huge losses, resulting in a serious impact on the economic benefits of wind power. On the basis of introducing current development and the basic structures of wind turbines and its common faults, this article select the main wind turbine drive system components to do fault diagnosis research, such as gearboxes, generators and main bearings.For the non-stationary, nonlinear characteristics of the rotating machinery vibration signal in the wind turbine, this article uses the latest time-frequency analysis method as Local Mean Decomposition(LMD) for the extraction of the vibration signal characteristic information and fault diagnosis research. Local mean decomposition method can adaptive decompose a complex multi-component FM signal into several PF components which is multiplying by the envelope signal and pure FM signal.Put forward using the method that chirp z transforming(CZT transform) the envelope signal and obtaining its spectrum, analyze the vibration signal of rolling bearing, to determine whether it appears fault characteristic frequency, in order to achieve the purpose of vibration signals fault diagnosis research. With the method of using spectral envelope for vibration signal characteristic frequency comparison, verify the validity of the method. Using the envelope spectrum analysis and spectrum analysis method for diagnosis of gear fault vibration signal characteristics. Using the method of envelope spectrum analysis and spectrum analysis in combination for the diagnosis of the characteristics of gear fault vibration signal.Put forward a kind of method based on LMD decomposition and singular value decomposition for the feature extraction and combine with the Fuzzy C-Means Clustering Algorithm(FCM) for recognition and classification the fault diagnosis. First, singular value decomposing the PF component of vibration signals that decomposed by LMD method, and think of the singular value matrix as feature vectors. Then based on the FCM clustering method as fault classifier, to identify the classification and diagnosis of different fault types. The method can realize the classification of a large number of fault diagnosis data at a time.