Feasibility Study of the Recurrent Neural Network for Modeling and Predicting Laser Polished Surface Quality

Surface quality improvement by a laser polishing (LP) process is a new innovative technology enabling value-adding functionalities, such as improving visual appearance, wettability, friction, and others through the control and reconfiguration of the surface topography. However, the resultant surface is dependent upon many process parameters which makes selecting optimal process parameters to achieve desired surface topography difficult and unrepeatable. It was proposed and demonstrated that recurrent neural network (RNN) can reliably model the LP of H13 tool steel and predict the laser polished surface topography parameters such as areal waviness and roughness with a probability of 99% and 79%, respectively. © 2021 Her Majesty the Queen in Right of Canada, as represented by the National Research Council of Canada; equal contribution of all co-authors