Detection of Grinding Burn Fault in Bearings by Squeeze Net.

Grinding burn is a significant problem encountered in the bearing manufacturing process. In the case of burns in the grinding bearing material, the surface quality deteriorates, and the material’s internal structure and mechanical properties are adversely affected. Therefore, detection of grinding burn is important. Acoustic emission (AE) technique and artificial intelligence-based methods are also used to detect this manufacture defect. In this study, data collected with AE sensors from a bearing grinding machine are classified in the Squeeze Net network based on Convolutional Neural Networks (CNN) by transfer learning. The recommended data combination and network parameters are detected with 100% accuracy in the weighted Squeeze Net network.