Physics-Informed Machine Learning for Smarter Design and Manufacturing Part II

Abstract Manufacturing systems are undergoing a fundamental transformation driven by the convergence of rich data, cyber-physical infrastructure, and artificial intelligence. At the heart of this transformation lies the challenge of embedding physical knowledge into data-driven models to achieve both predictive power and engineering interpretability. Physics-informed machine learning (PIML) has emerged as a paradigm to integrate conservation laws, constitutive equations, and domain constraints with neural architectures. Part I of this special issue addressed process monitoring, fault diagnosis and prognostics, and dynamic production planning and scheduling. Part II turns to the design and planning stage, the manufacturing process stage, and the product and supply management stage, where physical knowledge enters as geometric structure, mechanism equations, or operational constraints. The second part of this special issue brings together six contributions that address these challenges through novel PIML models. These works show how integrating physical knowledge can reduce data requirements, improve generalization, and keep predictions consistent with the underlying physics across design, manufacturing, and supply management applications.