Abstract
Rotating machinery is a large class of important equipment widely used in nuclear power plants (NPPs), and its reliability and stability are crucial to the operation of NPPs. To accurately identify the operation state of rotating machinery, a fault diagnosis method based on wavelet transform and deep residual neural network is proposed. Firstly, observation data is obtained through a plurality of vibration signal measurement sensors arranged at different positions of the rotating machinery, so as to analyze the operation state of the rotating machinery more comprehensively. Secondly, continuous wavelet transform is used to extract time-frequency features from multi-channel vibration signals, and the multi-channel time-frequency features are fused into time-frequency images. On this basis, the deep residual neural network is applied to adaptively extract the fault features contained in the timefrequency images to achieve accurate rotating machinery fault diagnosis. In this study, the motor experimental device and the rotor experimental device are used as test platforms to provide data support for the condition monitoring of rotating machinery.
The experimental results show that the proposed method can accurately identify the operation state of rotating machinery, and the effectiveness of the method is verified. The advantages of the multi-sensor monitoring strategy are further illustrated by comparison with single-sensor diagnostic effects. In addition, the method studied in this paper is compared with several current mainstream intelligent fault diagnosis methods. The comparison results show that the overall diagnosis effect of the proposed method is the best, which shows the potential application value of the method in the fault diagnosis of NPPs rotating machinery.
