Machine Learning Applications in Additive Manufacturing

Additive manufacturing (AM), often referred to as three-dimensional (3D) printing, has revolutionised modern production by enabling the layer-by-layer fabrication of parts with intricate geometries and minimal material waste. Machine learning (ML) techniques have become central to overcoming the inherent complexity of AM processes, in which dozens of interdependent parameters govern build quality, mechanical properties and microstructure. By mining large experimental and in-situ sensor datasets, ML models uncover hidden correlations among laser power, scan speed, powder feed, thermal histories and environmental conditions. Convolutional neural networks and ensemble learners are deployed for real-time defect detection—identifying porosity, delamination or lack of fusion from photodiode signals and thermal images—and for closed-loop control strategies. Generative algorithms facilitate design optimisation by proposing lattice or topology variations that satisfy stiffness, weight and functional requirements. Data-driven frameworks also integrate heterogeneous inputs from mechanical testing, microstructure characterisation and energy consumption metrics, supporting comprehensive decision-making in research and industry. Together, these advances are driving improvements in print reliability, part performance and resource efficiency across aerospace, biomedical, automotive and energy sectors.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Machine Learning Applications in Additive Manufacturing publication trend

The graph below shows the total number of articles in machine learning applications in additive manufacturing across all publications each year (not limited to Nature Index journals).

Technical terms

Additive Manufacturing (AM): Layer-wise fabrication of parts from digital models, enabling complex geometries and reduced waste.

Machine Learning (ML): Data-driven algorithms that learn patterns from datasets to make predictions or decisions without explicit programming.

Powder-Bed Fusion (PBF): AM process using a heat source to selectively fuse powder particles on a build platform.

Directed Energy Deposition (DED): AM technique that feeds powdered or wire material into a focused energy beam to build or repair components.

In-Situ Monitoring: Real-time acquisition of process signatures (e.g. thermal, acoustic, photodiode) during fabrication for quality assessment.

Data Fusion: Integration of heterogeneous data sources to produce more accurate, reliable and comprehensive predictive models.

Ensemble Learning: ML paradigm that combines multiple models (e.g. random forests, gradient boosting) to improve prediction robustness.

References

1. Progress and Opportunities for Machine Learning in Materials and Processes of Additive Manufacturing. Advanced Materials (2024).
2. Task-driven data fusion for additive manufacturing: Framework, approaches, and case studies. Journal of Industrial Information Integration (2023).
3. Process monitoring and machine learning for defect detection in laser-based metal additive manufacturing. Journal of Intelligent Manufacturing (2023).