Application of Convolutional Neural Network to Defect Diagnosis of Drill Bits

Drilling, one of the most used machining processes, has wide application in different industrial fields. Monitoring system health and operation status drilling process is essential for maintaining production efficiency. In this study, a convolutional neural network (CNN), deep-learning method, applied to defect diagnosis drill bits. Four bits with conditions were holes an aluminum block, vibration sensor collected signals. Vibration spectrograms generated using short-time Fourier transform 2D CNN algorithm, they then reconstructed into 1D data set algorithm. The input size was reduced significantly compared raw after data-reconstruction process. As result, shows diagnostic accuracy 97.33%. On other hand, provides 96.6%, but it only requires 2/3 computational time required by CNN.