Predicting Surface Roughness of Dry Cut Grey Cast Iron Based on Cutting Parameters and Vibration Signals from Different Sensor Positions in CNC Turning

抄録

During the turning process, cast iron is directly shattered to become particles. This mechanism means the surface roughness cannot be predicted using the kinematic equation. This paper provides surface roughness predictions using two methods, the multiple regression model (MRM) and artificial neural network (ANN). Cutting parameters and vibration signals are considered input variables in both methods. This work also overcomes the common sensor position limitation (tool shank) and provides a safe and efficient solution. The prediction values from MRM and ANN show accurate results compared to the measured surface roughness, with the average error of less than 8%. Furthermore, the proposed sensor position, at the turret bed, also exhibits similar prediction accuracy to a sensor at the tool shank, hence promising feasible industrial application.