What Lies Beneath: Surface Integrity

Accurate characterization and control of end-product material surface integrity is essential for the quality assurance process of subtractively manufactured components. The purpose of this presentation is to identify the metrics that define surface integrity, examine why they matter, and identify how predictive analysis of these metrics can be accomplished.

Traditional machine learning approaches have had measured success in recent literature at the predictive identification of several of the metrics of machined surface integrity, however these traditional approaches are not easily transferable to materials and cutting geometries outside of the architecture's trained region.

By incorporating energy based physical domain restrictions into the architecture and defining parameters based on lower dimensional standardized parameters such as ultimate yield strength of the material, and chip load- rather than higher dimensionalized parameters such as feed, speed, and depth of cut to reduce model complexity - it is theorized that a material and tool agnostic generalizable physics informed neural network can be established.