Chatter during machining operations is extremely detrimental to cutting tool life and the surface quality of the workpiece. Online monitoring methods are becoming more relevant for chatter detection and prediction in automated machining operations. While the utilization of vibration signals for the direct detection of chatter has been extensively researched, there are limited studies that explore the use of certain other sensor signatures, such as force signals. The present work aims to expand on the area of chatter detection in the turning of difficult-to-machine materials through Machine Learning (ML) using force signals acquired through multi-component tool dynamometer. An experimental setup has been established to collect force signature during dry turning operation under various process parameters selected on the basis of analytical Stability Lobe Diagrams (SLDs). Statistical features of Standard Deviation (SD) & Moving Window Standard Deviation (MWSD) extracted from the time-domain force signatures for different machining conditions are used to build statistical models using the Decision Tree (DT) Algorithm. The feasibility and performance of classifier models using the feature of MWSD are studied in the present paper. The DT classifier trained using MWSD of window size 100 has been found to provide a classification accuracy of 97.511%

Keywords: Chatter, Orthogonal Turning, CART Decision Tree, Tool Dynamometer, Moving Window Standard Deviation