인공지능 기반 머시닝 공구 마모 예측에 관한 연구

키워드

확장요약문

Carbon Fiber Reinforced Polymers (CFRPs) are considered as difficult-to-machine materials due to the highly abrasive nature of carbon fibers. Diamond coated and polycrystalline diamond (PCD) tools are commonly used in CFRP machining because of their high wear resistance. However, their high hardness and relatively low toughness may result in cutting edge fracture or chipping. This study investigates image-based chipping detection for CFRP milling tools with different cutter geometries. Canny edge detection was applied to microscope images of the tooth flank face before convolutional neural network (CNN) classification to highlight cutting edge features and reduce irrelevant image information. The CNN was trained using non-chipped and chipped images from a serrated diamond coated carbide tool. The trained model was then tested on two tools with different geometries: 1) a standard helical diamond coated carbide endmill and 2) a PCD tool, neither of which was included in the training data. The model correctly classified chipping conditions for both tool geometries, indicating that the Canny edge-based CNN model can detect cutting edge chipping in previously unseen tool geometries.