Design of experiments informed deep learning for modeling of directed energy deposition process with a small-size experimental dataset

Abstract

In order to address the high throughput data generation challenges in the directed energy deposition (DED) process development, a design of experiments (DOE) informed deep learning (DL) model is developed for modeling of laser powder-based DED process. A small-size experimental dataset is obtained according to DOE, by which a large-size dataset is augmented via the DOE regression model and then used to pre-train the DL model. A subset of experimental data is employed to fine-tune the DL model. The presently developed DOE-informed DL model is validated via single-track deposition of stainless steel 316L, in which the cross-section dilution shape (including depth) and the geometrical characteristics of beads, including the width, height, area, and wetting angle are predicted accurately. The prediction of the porosity and hardness are acceptable for single-track deposition, since variations of both the experimental porosity (from 0.04 % to 0.30 %) and hardness (from 168 HV to 182 HV) are quite small for the single-track deposition, especially predicted ranges for the porosity (from 0.05 % to 0.25 %) and hardness (from 170 HV to 180 HV) are the subset of the experimental results. The DOE-informed DL model developed in this study is based on a single-track deposition dataset; in future work, the DOE-informed DL model will be extended to multi-layer deposition.