Non-Intrusive Monitoring of Machining Processes for In-Process Product Health Prediction based on Machine Learning

Abstract

Intelligent monitoring systems for machining processes have been gradually developed for various scenarios, such as tool condition monitoring, chatter detection and product health prediction. Existing approaches of monitoring machining processes for quality assurance typically rely on intrusive sensor systems, such as dynamometers and spindle accelerometers, to obtain informative signals for training an algorithm, thus limiting their widespread adoption in industry. This paper presents a non-intrusive machining process monitoring method for in-process product health prediction with uncertainty information using Gaussian Process Regression (GPR). The performance of the proposed method is demonstrated on the prediction of dimensional deviations of a milling process.