A framework for generative intelligent process design of aviation riveting based on transformer-based multimodal information fusion

As a core process in aircraft structural assembly, aviation riveting quality directly determines the safety and reliability of aircraft. However, amid the application of new materials and parallel production scenarios of multiple aircraft models, it faces significant bottlenecks, including multimodal data silos, poor parameter adaptability, and inadequate dynamic response. Conventional manual experience-based methods can no longer meet the high requirements of advanced aviation manufacturing. To address these issues, this study proposes a Transformer-based multimodal fusion generative intelligent framework: a structured system unifies multimodal data features; a cross-modal attention mechanism captures inter-modal dependencies; an incremental learning-integrated model realizes knowledge transfer and dynamic parameter generation; deep reinforcement learning optimises parameters. Experiments show it achieves a 94.8% F1-score, which is 5.8% higher than that of traditional models, and a minimum optimisation time of 28.6 s, representing a 10.1% reduction compared with traditional methods. It outperforms traditional schemes, providing a new technical pathway for intelligent riveting design and enhancing aviation manufacturing digitalisation.