Intelligent Monitoring of Machining Processes using Gaussian Process Regression and On-Machine Comparator Measurement

Abstract

This paper presents an intelligent machining process monitoring approach with emphasis on On-Machine Comparator Measurement (OMCM) and the effect of remastering on dimensional accuracy. Comparator measurement applies the comparator principle by referencing each measurement to a calibrated master part. In this approach, a mastering procedure is first performed by measuring the calibrated master part to establish a reference. By directly comparing the test part with the master part under repeatability conditions, constant systematic errors in the measurement system are effectively cancelled when determining deviations from the master part. However, the accuracy of this method depends on the time interval between mastering and subsequent production measurements. An experimental study is conducted on a vertical milling centre using non-intrusive sensing. Gaussian Process Regression (GPR) is employed to model Coordinate Measuring Machine (CMM) measured diameter deviations and the associated uncertainty. OMCM is implemented, and the effect of remastering is discussed. The results demonstrate that measurement accuracy deteriorates when remastering is not performed prior to significant system drift, whereas timely remastering improves accuracy and reduces uncertainty, highlighting the critical role of remastering in maintaining measurement reliability with OMCM.