Bearing fault diagnosis method based on EEMD-AR model and DBN

Abstract

The invention provides a bearing fault diagnosis method based on an EEMD-AR model and a DBN. The bearing vibration signal changes nonlinearly. Firstly, the signal is decomposed into multiple IMF components by the ensemble empirical mode decomposition (EEMD) method, and the first several IMF components are selected to establish an autoregressive (AR) model. The autoregressive coefficient and the corresponding variance are calculated for each AR model as the input parameters of the deep belief network (DBN). The low-order features of the signal are extracted by EEMD and AR preprocessing. Then the representative high-order features are extracted from the input parameters by using the powerful hierarchical feature extraction and generalization ability of the DBN network for fault diagnosis. The method has higher accuracy than that of the conventional support vector machine and the artificial neural network fault diagnosis method.

Description

technical field [0001] The invention relates to the technical field of wind power, in particular to a bearing fault diagnosis method based on EEMD-AR model and DBN. Background technique [0002] Wind energy is one of the fastest-growing renewable energy sources today, and wind power occupies an overwhelming advantage in the installed capacity of renewable energy generation in the world. With the construction, promotion and operation of large-scale wind turbines, the failure of wind turbines will increase year by year, and the resulting economic losses will become more and more serious. Therefore, timely and effective fault prediction for key components such as wind turbine bearings has become an important way to improve the reliability of wind turbine operation and reduce maintenance costs. [0003] In view of the above problems, literature [1] preprocesses the high-speed bearing signals of wind turbines by using EMD (empirical mode decomposition), and extracts the amplitud...

AI Technical Summary

Problems solved by technology

[0007] Aiming at the problem that the current traditional wind turbine bearing fault signal has non-stationary and nonlinear characteristics, and the fault feature extraction is difficult, the present invention provides a bearing fault diagnosis method based on the EEMD-AR model and DBN to realize the high-precision fault of the wind turbine bearing diagnosis

Images