Wind turbine planetary gear box fault diagnosis method based on ACGAN

Abstract

The invention relates to a wind turbine planetary gear box fault diagnosis method based on ACGAN. The method comprises the following steps: collecting a vibration signal of a planetary gear box as a diagnosis sample; dividing the diagnosis sample into a training set sample and a test set sample according to a set proportion; inputting the training set sample into the ACGAN for adaptive training, obtaining parameters of a discriminator network and a generator network in the ACGAN until the ACGAN reaches Nash equilibrium, and storing the trained parameters of the discriminator network and the generator network in the ACGAN; taking the trained ACGAN as a fault diagnosis model, inputting the ACGAN into a test set sample, generating a realistic sample by a generator network, and adding the realistic sample into a diagnosis sample; and enabling the discriminator network to output a gearbox fault diagnosis result. Compared with the prior art, the method has the advantages that the original data training network can be directly used, the feature vectors are automatically extracted, and the accuracy of model recognition and classification is high; meanwhile, the generalization ability is extremely high, and fault diagnosis can be effectively carried out on the planetary gear box of the wind turbine generator.

Description

Technical field [0001] The present invention relates to the field of control, monitoring and diagnosis of power system equipment, and in particular to an ACGAN-based fault diagnosis method for planetary gearboxes of wind turbines. Background technique [0002] Wind energy is one of the most promising new energy sources at present. As an important transmission device of wind turbines, planetary gearboxes are composed of planetary gears, sun gears, ring gears and planetary carriers. They can obtain high torque in a compact space. Compare. Due to its complex vibration transmission path, multi-tooth meshing effect, signal non-stationarity, and large working background noise, its fault diagnosis has its own characteristics and difficulties. It is difficult to analyze it in the time domain or frequency domain using traditional methods. Extract valid fault information. With the development of Internet technology and Internet of Things technology, data acquisition and storage have...

AI Technical Summary

Problems solved by technology

However, the traditional shallow machine learning algorithm has insurmountable shortcomings, such as low computational efficiency of the algorithm itself, difficult to meet the diagnostic accuracy requirements, poor generalization ability of the model, etc.

Images