Abstract
Process twins provide real-time representations of entire production processes. Unlike asset-based digital twins, they capture how process steps interact, which can lead to efficiency gains across the entire process. However, developing a process twin is costly, requiring accurate modeling of the entire production process, including process steps, equipment and product-specific settings, and process variations. The resulting model must then be bound to live operational data.
We present FacProcessTwin, a system that leverages a large language model (LLM) to reduce development time by building a process twin from a plant's process documentation and natural-language input from an operator. FacProcessTwin generates a complete process model and automatically binds its process steps to live operational data. The generated model and its data bindings are rendered as an interactive process diagram, allowing manufacturing personnel to monitor and correct the system's autonomous decisions, such as resolving uncertainty at safety-critical binding steps.
We evaluate FacProcessTwin through a real-world case study of an Australian food manufacturer, covering 16 production process flows across chilled, frozen, and aseptic shelf-stable product categories, including process variations within the same product. The results show that FacProcessTwin generates these process models accurately (a mean F1 of 95.2% against ground truth) and builds each twin in roughly a sixth of the manual time. Its human-in-the-loop governance keeps the safety-critical bindings correct: at ambiguous tags where a single-pass baseline silently mis-binds 75.0% of the time, FacProcessTwin defers to the operator and mis-binds none.
Blogger's Review: FacProcessTwin showcases a remarkable integration of LLM with real-time data binding, significantly enhancing efficiency in manufacturing processes. Its potential in safety-critical decision-making through human-in-the-loop governance demonstrates a tremendous opportunity for collaborative technologies. This innovation not only reduces development time but also improves model accuracy, making it worthy of broader application across various fields.