In-process surface roughness estimation model for compliant abrasive belt machining process

Abstract

Surface roughness inspection in robotic abrasive belt machining process is an off-line operation which is time-consuming. An in-process multi-sensor integration technique comprising of force, accelerometer and acoustic emission sensor was developed to predict state of the surface roughness during machining. Time and frequency-domain features extracted from sensor signals were correlated with the corresponding surface roughness to train the Support vector machines (SVM's) in Matlab toolbox and a classification model was developed. Prediction accuracy of the classification model shows proposed in-process surface roughness recognition system can be integrated with abrasive belt machining process for capping lead-time and is reliable.