Abstract:Aiming at the problems of low speed of large Turbine Bearings, unstable and scarce experimental samples, and fault diagnosis is limited to artificial diagnosis, a new fault diagnosis method based on correlation function of weighted fusion algorithm and improved Hilbert-Huang Transform (HHT) algorithm is proposed. Firstly, a weighted fusion algorithm based on correlation function with dynamic adaptability and anti-disturbance was adopted to deal with the collected data, then the fault feature vector was constructed based on the improved HHT. Back Propagation (BP) neural network was used to identify fault types of feature layer, and to determine the type of failure of the wind turbine at the last. The experimental results show that this method can improve the reliability of fault diagnosis effectively.