Prediction and Optimization of Surface quality and Microhardness using Machine learning in Selective laser melting of SS316L Biomedical alloy

Laser additive manufacturing based selective laser melting (SLM) technique have attracted a lot of attention due to the rising need of high performance materials in aerospace, automotive, and biomedical applications. However, because of the intricate relationships between several parameters including laser power, scanning speed, hatch spacing, and layer thickness, optimization of the process parameters for SLM is a tedious task. Machine learning (ML) technique can handle a variety of data sets and can accurately predict complicated and non-linear relationships in SLM. In this paper, three tree based ML models such as Random forest, Gradient Boosting, and XG boost Regressor are used for prediction of surface roughness (R _a ) and microhardness (MH) of SLM fabricated parts for improved part quality and longevity. The efficacy of the ML models is evaluated in terms of prediction accuracy and computational efficiency after training and testing to predict optimal process parameters for minimum R _a and maximum MH, respectively. The average error of XG boost model for prediction of R _a and MH is 0.1217% and 1.73%, respectively which is significantly lower as compared to Random forest and Gradient boosting methods. Therefore, XG boosting showed better accuracy in prediction of R _a and MH values as compared to Random forest and Gradient boosting methods. This is because of its better data handling capacity and efficient capturing of complex data sets. A 29.64% decrease in R _a and 14.73% increase in MH values are achieved at optimized settings for performance improvement of SLM fabricated parts. The maximum and minimum porosity in SLM fabricated parts is found to be 0.987% and 0.249% at different energy densities after image processing by Image J software. This work will be useful in implementation of ML technique in SLM fabrication for better process control, reduction in trial-and-error, and to improve the functionality and reliability of finished parts.