Knowledge Discovery by Decision Tree Using Experimental Data in High-Speed Turning of Steel with Ceramic Tool Insert

Abstract

The manufacturing industry is of immense importance. Turning is one of the most basic operations performed across all manufacturing industries till date. Process parameter optimization and modeling in this field, which is very complex, have been investigated by many past researchers. Various methods like statistical techniques, and finite element-based and soft computing-based approaches were used to predict the machinability parameters like flank wear based on the given input cutting conditions like cutting speed, feed rate, depth of cut, etc. Nevertheless, a very few work was done in the area of knowledge discovery with the experimental data. In this work, efforts have been made to extract knowledge automatically using decision tree from the raw experimental data while turning EN24 steel with Cr₂O₃-doped zirconia toughened alumina (Cr-ZTA) ceramic tool insert. After that, the extracted knowledge in the forms of set of fuzzy rules was fed into a custom-made fuzzy logic control (FLC) system developed for predicting flank wear. The results of predictions are validated with experimental test data, and the capability of the system is stated with scope for improvements.