Process optimization and quality prediction of laser aided additive manufacturing SS 420 based on RSM and WOA-Bi-LSTM

Abstract

Laser aided additive manufacturing (LAAM) is a promising method for surface modification, repair, and direct fabrication 3D components. Nevertheless, the high cost and time-consuming nature of trial-and-error based process development hinder the progress of this technology. Consequently, machine learning has garnered substantial attention from both the academia and industrial sectors due to its ability to expedite process study for both research and industrial applications. This paper presents a study on the LAAM of 420 stainless steel on the surface of Q235 steel using the combined Response Surface Methodology (RSM) and Whale Optimization Algorithm-Bidirectional Long Short-Term Memory (WOA-Bi-LSTM) model. The results show that the process parameters can be effectively optimised to improve the prediction accuracy using the multi-objective whale optimisation algorithm for the fitted equations obtained in response surface analysis. At the same time, with the support of a large number of data, the WOA-Bi-LSTM model can accurately predict the quality of the LAAMed material, which is better than the response surface analysis method.