Special Issue on Physics-Informed Machine Learning for Advanced Manufacturing

Journal of Manufacturing Science and Engineering

Submit Paper

Modeling of manufacturing processes, machines, and systems is the everlasting theme of advanced manufacturing. Compared to conventional physics-based analytical or numerical models, data-driven methods such as machine learning (ML) have been shown to be an alternative approach to prediction and optimization. However, the inherent “black box” nature of data-driven ML techniques such as those represented by neural networks has often presented a challenge to interpret ML outcomes, particularly when the training dataset may be limited or noisy. On the other hand, manufacturing physical laws (e.g., ordinary or partial differential equations) and domain knowledge are not effectively utilized to develop data-efficient ML algorithms.

To leverage the advantages of ML and physical laws of advanced manufacturing, this call focuses on physics-informed machine learning (PIML) by integrating ML techniques and physical laws to improve the accuracy, efficiency, and generalization of predictive models. PIML can be particularly useful for advanced manufacturing where the underlying physics is well understood, or it is difficult or impractical to develop a comprehensive model that captures all of the relevant features.

---

Topic Areas

THE SCOPE OF THIS ISSUE INCLUDES BUT IS NOT LIMITED TO:

Forward PIML models Inverse and data assimilation problems Data-driven discovery of physical models Neural network architectures for physics PIML-based reduced-order models

Uncertainty quantification and Bayesian optimization PIML for process or system optimization PIML for machine dynamics PIML for quality control PIML for predictive maintenance