An Operating State Prediction Model for Aircraft Assembly Lines Based on Temporal Graph Neural Network

With the increasing demand for aviation market orders, accurately monitoring the operational status of assembly lines, and achieving sustainable optimization are critical to ensuring on-time product delivery. The complexity and heterogeneity of assembly operations, resources, and materials result in the assembly line’s operational state being influenced by multiple interacting factors. In recent years, data-driven approaches have been widely applied to assembly line operational state prediction. However, most existing methods primarily rely on instantaneous features at specific time points, failing to fully capture temporal dependencies, which may lead to inaccurate predictions. To address these limitations, this study proposes a temporal graph neural network (TGNN)-based approach for aircraft assembly line operational state prediction, incorporating both the temporal characteristics of the assembly process and the interactions among production factors. First, the structural requirements and modeling considerations are analyzed to formulate a descriptive framework for the assembly process. Then, a temporal dataset incorporating event nodes, timestamps, and other essential elements is constructed, and the assembly process is represented as a graph structure, serving as input to the TGNN-based prediction model. Additionally, a computational methodology for deriving operational state indicators is also proposed. Finally, the effectiveness of the proposed method is verified through an industry case derived from a real aircraft assembly line.