Chip Morphology Prediction in Inconel 718 Milling through Machine Learning to Control Surface Integrity

Abstract

A nickel-based aerospace superalloy, Inconel 718 presents machining challenges because of its hardness and strength. Monitoring and predicting chip morphology during milling is essential for early defect detection and process optimisation. This study examines the correlation between sensor signals with surface roughness and chip morphology in milling Inconel 718 using machine learning (ML). Due to progressive tool wear and heat generation, the surface roughness varies in addition to the chip exhibiting different morphologies, such as continuous, discontinuous, and oxidised chips. AE signals were analysed in the time and frequency domains to identify chip morphology transitions. An accelerometer captured cutting vibration signals that showed higher instability during discontinuous chip formation. Chip colour due to oxidization varies with milling forces as a result of tool wear. Based on multiple sensor data fusion, a random forest model predicts better chip morphology from different machining parameters. The integrated ML system enables real-time monitoring of chip morphology mechanisms through diverse signals. This permits early diagnosis of surface integrity and chip morphologies indicating imminent tool wear. The approach enhances process stability and tool life when milling difficult-to-machine alloys. It demonstrates the viability of relating sensor signals to fundamental mechanisms through AI for intelligent machining.