Prediction of Tool Wear Using Machine Learning Approaches for Machining on Lathe Machine

Abstract

: In manufacturing industries, removal of material from the workpiece is the prime processes that convert raw material into finished product. During removal processes the cutting tool are incessantly deteriorated in health, which can be stated as perks and drawbacks of process. The precision and roughness of the material are directly related to the condition of the tools during the machining process. Machining analysis depends on numerous of cutting conditions when it is being performed. The likelihood of wearing increases with repeated use. So, by implementing the proposed approach for tool wear prediction can improve the quality as well as reduce the machining time. However, to maintain the healthy tool's conditions for prolong time is a major challenge for the scientific community. Hence, as a component of industry 4.0, this study explored the possibilities to monitor the tool condition using the machine learning techniques. So, an endeavor has been made to present a solution of this problem without hampering the productivity losses in terms of time, material, and tool, consequences in high productivity. For the proposed work, machine learning techniques such as k-NN, Random forest, Adaboost, k-Star, and Decision Tree are implemented and there accuracy of prediction is demonstrated. Furthermore, WEKA, open source software has been used to employ several tool learning algorithms for better understanding. The investigation noticed that the random forest algorithm has a higher accuracy of 97.30% and a root mean square error value of 0.144 among all other algorithm.