Offline planning optimization and formation prediction of laser directed energy deposition process

Research question

Can offline planning reduce corner bulging in laser deposition while predicting the resulting layer height?

Method

CTAR corner detection guides laser-power adjustments, while an adapted Mask R-CNN extracts melt-pool geometry. A stacking ensemble combines process and pool features to predict layer height in 316L stainless-steel deposition.

Main findings

Corner-specific power adjustment removed the local bulging observed in the blade-model experiment. On the paper's held-out layer-height data, the stacking model achieved a mean absolute percentage error of 2.3697%.