Deep learning enabled cutting tool selection for special-shaped machining features of complex products. (July 2019)

• Record Type: Journal Article
• Authors: Zhou, Guanghui; Yang, Xiongjun; Zhang, Chao; Li, Zhi; Xiao, Zhongdong
• Is Part Of: Advances in engineering software. Volume 133(2019)
• Page Start: 1
• Page End: 11
• Publication Date: 2019-07
• Subjects: Cutting tool selection -- Special-shaped machining features -- Complex products -- Residual networks -- Deep learning; Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2019.04.007
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: Deep learning based cutting tool selection approach for special-shaped machining features. There is a correspondence between each special-shaped machining feature and each tool. Transforming the problem of cutting tool selection into a feature recognition problem. The accuracy of cutting tool selection under the constructed dataset is 99.68%. Abstract: Each complex product contains many special-shaped machining features required to be machined by the specific customized cutting tools. In this context, we propose a deep learning based cutting tool selection approach, which contributes to make it effective and efficiency for and also improves the intelligence of the process of cutting tool selection for special-shaped machining features of complex products. In this approach, one-to-one correspondence between each special-shaped machining feature and each cutting tool is first analyzed and established. Then, the problem of cutting tool selection could be transformed into a feature recognition problem. To this end, each special-shaped machining feature is represented by its multiple drawing views that contain rich information for differentiating each of these features. With numbers of these views as training set, a deep residual network (ResNet) is trained successfully for feature recognition, where the recognized feature's cutting tool could also be automatically selected based on the one-to-one correspondence. With the learned ResNet, engineers could use an engineering drawing to select cutting tools intelligently. Finally, the proposed approach is applied to the special-shaped machining features of a vortex shell workpiece to demonstrate its feasibility. The presented approach provides a valuable insight into the intelligent cutting tool selection for special-shaped machining features of complex products.