Image Based Detection of Coating Wear on Cutting Tools With Machine Learning

Abstract

Wear of cutting tools is known to affect the surface integrity of the workpiece and significantly contributes to machine downtime. To establish wear-resistant cutting tools, several coating strategies have been introduced. It has been shown that the wear rate increases dramatically once the coating is worn through. Detecting coating layer loss is therefore a good indicator of the remaining useful life of the cutting tool. Based on cutting experiments conducted with a TiN/AlTiN-coated cutting tool, an image dataset was generated and pre-processed using multiple algorithms, such as Canny edge detection and Hough line transforms. For the classification task four machine learning Algorithms consisting of Random Forests, Decision Trees, Support Vector Machines and a Feed Forward Neural Network were implemented. The results demonstrate that all four algorithms lead to a good classification performance, with Decision Trees showing the best performance with a F1-score of 0.95. Therefore, this research provides an efficient data processing and classification framework for detecting coating wear on cutting tools.