Development of a methodical approach for uncertainty quantification and meta-modeling of surface hardness in white layers of longitudinal turned AISI4140 surfaces

Abstract

The formation of thermally and mechanically induced near-surface microstructures in the form of white layers leads to different hardness properties in these areas. Therefore, this paper conducts systematic surface hardness measurements and uncertainty quantification utilizing the Monte Carlo Method (MCM) in accordance with the Guide to the Expression of Uncertainty in Measurement (GUM). Furthermore, several meta-models describing the hardness course in relationship to the material depth are used to model this nonlinear relationship via machine learning. The evaluation and selection of the optimal model considers the trade-off between measurement uncertainty and prediction quality in terms of mean squared error (MSE). The resulting measurement uncertainty is to be used for the calibration of a non-destructive micromagnetic material sensor. This will then be implemented for in-process monitoring in the outer diameter longitudinal turning process. This should make it possible to detect white layers during machining and to avoid them accordingly by controlling the machine parameters. By means of a soft sensor, the corresponding target value is to be derived from the micromagnetic material sensor measurement.

Quantifying predictive uncertainty in damage classification for nondestructive evaluation using Bayesian approximation and deep learning

“…defect size and shape) and material property (e.g. hardness and strength) are considered as the aleatoric uncertainty in applications related to material characterization and damage detection [37][38][39][40]. On the other hand, the processing parameters related to simulation (e.g.…”