Neural network prediction of thermal field spatiotemporal evolution during additive manufacturing: an overview

• Published in: International journal of advanced manufacturing technology Vol. 134; no. 5-6; pp. 2107 - 2128
• Main Authors: Chike, Onuchukwu Godwin, Ahmad, Norhayati, Faiz Wan Ali, Wan Fahmin
• Format: Journal Article
• Language: English
• Published: London Springer London 01.09.2024
• Subjects: CAE) and Design; Computer-Aided Engineering (CAD; Critical Review; Engineering; Industrial and Production Engineering; Mechanical Engineering; Media Management; Thermal field; Deep neural network; Additive manufacturing; Spatiotemporal evolution; Physics-based modeling; Machine learning
• ISSN: 0268-3768; 1433-3015
• DOI: 10.1007/s00170-024-14256-6

This paper provides an overview of the application of machine learning (ML) techniques for predicting the spatiotemporal evolution of thermal fields during additive manufacturing (AM) processes. AM, also known as three-dimensional printing, has gained significant attention in various industries due to its potential for rapid prototyping and customized production. However, accurately predicting and controlling the thermal behavior during the AM process is crucial for ensuring the quality and reliability of the printed components. Traditional physics-based models (PBM) often face challenges in capturing AM’s complex dynamics and inherent uncertainties. In recent years, ML algorithms, particularly neural networks (NNs), have shown promising results in predicting the thermal field evolution. This paper reviews the existing literature and highlights the critical methodologies and recent advancements in NN-based predictions. It explores novel perspectives by discussing the hybrid modeling approaches, including the combination of PBMs with NNs. This overview highlights the evolving landscape of predictive techniques in the context of AM and underscores the potential for enhancing accuracy and efficiency in thermal field prediction. The paper also discusses the challenges and outlines future directions for enhancing the accuracy and efficiency of thermal field prediction in AM. By synthesizing current research, this overview will guide researchers and practitioners toward leveraging NNs effectively for optimizing thermal management in AM processes. The insights presented underscore the transformative potential of NN predictions in advancing AM capabilities.