Durable Test Bench Vibration Fault Diagnosis Method Based on Convolutional Neural Networks
Currently,common methods for vibration fault diagnosis involve signal feature extraction and the application of relevant signal processing algorithms.However,these methods have high hardware requirements and limitations associated with different algorithms.To address vibration fault diagnosis,this study proposes a signal processing and image recogni-tion-based approach.Firstly,the collected vibration signals are denoised to eliminate noise.Then,the wavelet transform is employed to obtain the wavelet transform spectrogram.The obtained spectrograms of different fault types are used for neu-ral network training,resulting in the generation of a fault diagnosis model file.Experimental results demonstrate that the trained model performs well in terms of vibration fault diagnosis accuracy and recognition speed,achieving an accuracy rate of 98%.Compared to traditional fault diagnosis algorithms,this research offers certain advantages.
NETL
NSTL
万方数据
