Optimal process parameter determination in selective laser melting via machine learning-guided sequential quadratic programing

Abstract

Selective Laser Melting (SLM) is a widely used additive manufacturing method with critical processing parameters such as Laser Power, Scanning Speed, Hatch Distance, and Laser Beam Diameter, which significantly impact process quality and mechanical properties of fabricated parts like Relative Density, Hardness, and Surface Roughness. This study investigates an effective approach to modeling the SLM process for predicting and optimizing processing parameters using a combination of Machine Learning (ML) and Sequential Quadratic Programing (SQP). The proposed method aims to predict the mechanical properties of SLM-produced AlSi10Mg parts using Machine Learning techniques and then transform the process into a constrained multiple-objective optimization model employing the SQP algorithm to determine optimal process parameters. Artificial Neural Networks (ANN), Gaussian Process Regression (GPR), and Support Vector Machine (SVM) algorithms are applied to establish correlations between process parameters and mechanical properties, represented as mathematical functions. A cost function is formulated using these functions and minimized using the SQP method. Statistical analysis is implemented to validate the influence of processing parameters on mechanical properties accordingly. The outcomes of the method are verified through additional experiments and analyses using Micro-CT scanning techniques. The results demonstrate that processing parameters have a significant impact on part properties across different scales, with Micro-CT scanning proving instrumental in validating mechanical properties. Overall, this study establishes that multiple objectives can be efficiently optimized through the integration of ML and SQP methods, complemented by a judicious number of experiments.