Physics‐informed neural network (PINN)‐embedded digital twinning for manufacturing sustainability: A framework for intelligent and adaptive process systems

Abstract Advanced manufacturing comes into the digital age. Among strategic digital technologies, digital twin (DT) technology has been increasingly applied in industry. A more recent advancement in DT technology is the integration of physics‐informed neural networks (PINNs) in digital twinning, which enables the construction of a robust virtual plant. By leveraging real‐time data from a connected physical plant, the PINN‐based DT supports sustainability assessment, optimizes design and operation, predicts sustainability challenges, and identifies best possible technical solutions. In this paper, a PINN‐embedded modular DT framework is introduced for sustainable manufacturing. The PINNs are used to dynamically adjust key parameters in DT modules. This allows the resulting digital plant to be self‐evolving and capable of accurately mirroring, predicting, and improving the plant's sustainability performance. It also facilitates reconfiguring processes, where sustainability metrics are incorporated into process design. The methodological efficacy is demonstrated by developing and operating sustainable cleaning‐rinsing systems in electroplating plants.