Research on Real-Time Stress Prediction Method for Symmetrical Vertical Tail Structures During Transportation Based on Metabolic Deep Learning Models

The transportation safety of aviation structural components directly impacts equipment performance and mission success rates, constituting a critical link in modern aviation industry that cannot be overlooked. Traditional methods relying on numerical analysis or structural health monitoring techniques analyze structural stress during transportation to ensure safety. However, they suffer from low computational efficiency, inability to perform real-time online monitoring, or limited coverage to a few measurement points. To address these challenges, this study proposes a metabolic deep learning model based on Gated Recurrent Units (GRU) for predicting stresses in symmetrical vertical tail (SVT) structures during transportation. First, a refined finite element model of the symmetrical vertical tail structure is established using a model ensemble analysis strategy. By integrating modal analysis with transient analysis, the maximum stress loads for different transportation processes are calculated, generating maximum stress data samples under multiple acceleration history conditions. Second, by establishing a mathematical description and deep learning framework, the GRU establishes a mapping relationship between acceleration history and maximum stress, enabling prediction of maximum stress loads associated with different acceleration histories. This method effectively resolves the challenge of exponential mesh growth in complex assembly simulations. This research enables real-time structural stress warnings for drivers during highway transportation, triggering early alerts when stress approaches allowable limits to ensure structural safety and reliability.