An Intelligent Machine Learning Based Method for Tool Wear Estimation in the Vertical Broaching Process

Abstract

This work describes a machine vision system workflow to automatically estimate the broaching tool wear. The proposed system offers the possibility to evaluate the evolution of wear under different machining conditions and to decide when a tool should be replaced, guaranteeing the quality of the machined part and avoiding catastrophic tool breakage. In addition, the paper discusses the advantages of the proposed method over the traditional and widely used ISO 3685:1993 based methods, which are highly influenced by the operator. The proposed method uses a novel wear area segmentation technique based on Machine Learning artificial intelligence, generating highly reproducible values, saving technicians labor-intensive tasks, and obtaining values with high accuracy. The results show a strong relationship between the values obtained by the proposed automatic method and the experimental ones, with errors below 0.17% and 2.88% corresponding to the MSE and MAE respectively.