Prediction of Machining Quality and Tool Wear in Micro-Turning Machine Using Machine Learning Models

Industry 4.0 machines are intelligent enough to respond to changes in the machining environment during operation. Prediction of workpiece quality and cutting tool wear is an essential aspect in forecasting the performance of the micro-manufacturing machine tool in the field of smart manufacturing. Micro-structures of titanium alloys are widely used in next-generation vascular stents, drug-eluting stents, micro-opto-electromechanical systems, microfluidics and bio-micro-electromechanical systems. This study presents the development of machine learning (ML)-based models to predict the condition of the micro-turning machine tool. Coated carbide (CCMT060204LF KC5010) cutting inserts are readily used to machine the titanium Ti6Al4V alloy. To monitor vibration signals, a vibration sensor is positioned at the tooltip. The tool flank wear and the surface roughness of the machined component have been measured for each machining pass. The contribution of various parameters affecting surface roughness and flank wear has been studied. The experimental data consisting of vibration signal, speed, feed and depth of cut is used to train multiple ML models and thereby predict the surface roughness and tool flank wear. The Random Forest (RF) model is more accurate but takes longer to computations, whereas the Radial-based function is less accurate but takes less time. The Regression tree has a 7% lower accuracy than the (RF) model, but it is two times quicker. Given the importance of computation time and accuracy associated with various machine learning models, the regression tree performs better in predicting machine tool condition.