Fault Feature Extraction Method And Its Application On Drive Train Of Wind Turbines Based On Fractal Dimension

Wind turbine is currently one of the main renewable energy power generation. It relies on the blades that transform the wind kinetic energy into mechanical to drive a generator producing electricity. The operation condition of the wind turbine is very poor, and its condition monitoring and fault diagnosis is a key role to the security and efficiency.In this paper, the operation principle and main structural components of wind turbine are introduced, and the common faults and failure mechanism of the key components are described. Based on this, the method of fault feature extraction on vibration monitoring signal is studied and mainly include the following three aspects:(1) Based on the fractal dimension, the fault characteristics of the wind turbine generator is studied. The correlation dimension of the vibration signal is calculated in the MATLAB environment. The results are verified by the simulation and the measured wind turbine vibration data, then analyze the influence of the noise on the correlation dimension.(2) The combination of Kalman filter and correlation dimension is studied. Firstly, filter the vibration signal, and reduce the influence of interference noise, and then calculate the correlation dimension of the data which is excluded from other non-fault factors.(3)The combination of wavelet scale domain filtering and correlation dimension is studied. First of all take the measured vibration signal wavelet decomposition get the wavelet coefficients. After filtering the wavelet coefficients using Kalman, reconstruct the signal to calculate the correlation dimension, which can eliminate the influence of interference noise to get fault information. The practical application shows that the method can effectively characterize the fault state of the wind turbine.