Wind power variable pitch system fault diagnosis method based on CEEMDAN-BNs

Abstract

The invention provides a wind power variable pitch system fault diagnosis method based on CEEMDAN-BNs, and the method comprises the steps of: adding white noise into a sample time sequence signal in a fault state, and carrying out CEEMDAN mode decomposition, thus obtaining IMF and margin signals of a plurality of mode component signals; selecting a mode component with a high signal-to-noise ratio and a large correlation coefficient in the IMF signal, and constructing an energy characteristic matrix through Hilbert transform; performing interval division on the energy characteristic matrix; constructing a Bayesian network topological structure according to the divided energy characteristic matrix and expert priori knowledge; searching an optimal network topology structure of the Bayesian network through a hill climbing algorithm; and training the Bayesian network, and inputting the to-be-diagnosed signal into the trained Bayesian network to obtain a fault diagnosis result. According to the method, the Bayesian network is constructed through feature extraction in combination with the CEEMDAN algorithm, and while the accuracy is improved, simplification of the Bayesian network structure and improvement of the network training speed are realized.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis method for a wind variable pitch system based on CEEMDAN-BNs. Background technique [0002] Wind turbines have the characteristics of huge equipment, high cost, and complex structure, and are generally located in relatively remote areas such as Gobi desert, plateau, or sea. Therefore, once a subsystem is damaged and fails during operation, it cannot be repaired in time. This will lead to aggravated faults, and may even lead to failures of adjacent systems, resulting in large economic losses. Therefore, fault diagnosis and status detection of wind turbines are very important for daily operation and maintenance. The wind variable pitch system is an important subsystem to maintain the safe and stable operation of wind turbines. It is one of the high-frequency fault components of wind turbines. It ensures the safe operation of wind turbines by adjusting the ...

AI Technical Summary

Problems solved by technology

Classical Bayesian networks usually build Bayesian network topology based on expert knowledge and mechanical system structure, but the establishment of Bayesian networks for large-scale mechanical systems is complicated, and the conditional probability of Bayesian networks due to the complex network structure The distribution is difficult to obtain, which limits the application of Bayesian networks in fault diagnosis

Images