What question did this study set out to answer?

June 10, 2026Open Access

Research on Density Prediction of Laser Powder Bed Fusion Process Parameters for IN718 Nickel-Based Superalloy Based on Machine Learning

Key Points

The aim is to model the relationship between process parameters and relative density in selective laser melting of IN718 superalloy using machine learning methods.
Developed a data-driven prediction framework integrating data augmentation and feature engineering.
Used fourteen experimental data sets while applying augmentation strategies like RBF interpolation, GAN, and KNN.
Trained models including SVR, RF, and ANN with hyperparameter optimization via grid search and leave-one-out cross-validation.
ANN achieved the highest average R2 of 0.9269, outperforming both SVR (0.9148) and RF (0.8393).
E_vol contributed 51.58% to the predictive power, confirming physics-derived features enhance model performance.
LOO-CV results indicated an improvement of 0.0246 in average R2 with the introduction of physics-derived features.

Abstract

This study addresses the challenge of modeling the complex non-linear relationship between process parameters and relative density in selective laser melting (SLM) of IN718 nickel-based superalloy under small-sample conditions. A data-driven prediction framework integrating data augmentation, physics-informed feature engineering, machine learning, and model interpretability analysis was developed and systematically validated. Fourteen sets of experimental data covering both vertical and horizontal building directions were collected by varying laser power (P), scan speed (v), and hatch spacing (h). To overcome the small-sample limitation, three augmentation strategies—radial basis function (RBF) interpolation, generative adversarial network (GAN), and K-nearest neighbors (KNN) —were systematically compared under unified physical constraints combining local perturbation and volumetric energy density (Eᵥol) filtering, with Pearson correlation coefficient consistency used to select the optimal strategy. Eight physically meaningful input features were constructed, including Eᵥol and line energy density (Eₗine), explicitly embedding SLM process physics into the learning framework. Support vector regression (SVR), random forest (RF), and artificial neural network (ANN) models were trained and their hyperparameters were systematically optimized via exhaustive grid search combined with leave-one-out cross-validation (LOO-CV), ensuring robust model selection under small-sample constraints. A physics-based baseline model (Eᵥol quadratic fitting, LOO-CV average R2 = 0. 2534) was established to quantify the gain of machine learning over empirical formulas. LOO-CV results show that ANN achieves the highest average R2 of 0. 9269, followed by SVR (0. 9148) and RF (0. 8393), all of which substantially outperform the physical baseline. Feature importance analysis reveals that Eᵥol accounts for 51. 58% of the predictive power, and ablation experiments confirm that introducing physics-derived features improves the average R2 by 0. 0246 compared with raw process parameters alone. To further elucidate the predictive mechanism of the optimal ANN model, Partial Dependence Plot (PDP) analysis was conducted for all eight input features, visualizing their marginal effects on predicted density and confirming physical consistency with SLM mechanisms. This framework provides a reliable, interpretable, data-driven solution for intelligent SLM process optimization with limited experimental data.

Read Full Paperexternally

اسأل الذكاء الاصطناعي

Bookmark

View Full Paper