Ball screw fault diagnosis based on continuous wavelet transform and two-dimensional convolution neural network

Due to extreme operating conditions such as high-speed and heavy loads, ball screws are prone damages, that affect the accuracy operational safety of mechanical equipment. As strong background noise weak fault characteristics, it is difficult capture inherent state only depending on time-domain or frequency-domain information from vibration signal. In this paper, a diagnosis method for screw based continuous wavelet transform (CWT) two-dimensional convolutional neural network (2DCNN) proposed. The noise-reducing signal obtained via CWT. time-frequency graph reduction can more comprehensively reflect screw. used input train test 2DCNN. Finally, results different types faults reveal proposed CWT-2DCNN achieve an average recognition rate 99.67%. Compared with one-dimensional (1DCNN) traditional BP network, has fast convergence high accuracy. Time-frequency graphs noise-reduced features classification effectively avoid problem uncertainty due manual extraction features. application potential in field pair diagnosis.