A Study on a Deep Learning-Based Flight Parameter-Strain Correlation Model
摘要
To thoroughly understand the structural performance of imported aircraft and achieve independent mastery of structural maintenance and modification capabilities, establishing a correlation model between flight parameters and structural strain during the determination of aircraft load spectra is critical for expanding data samples, obtaining detailed stress states of structural components, and evaluating structural lifespan. This study takes flight parameter data and synchronized structural strain response data of a specific aircraft model as research objects. Based on a GRU-based sequential prediction framework, this work systematically examines the influence of critical model parameters—including optimizer configuration, nonlinear activation functions, input strain feature selection, and the number of GRU layers—on fitting performance. Through comparative analysis of these parameters’ effects, an optimal correlation model is ultimately established to quantify the relationship between flight parameters and structural strain. This approach enables the prediction of local structural strain based on flight parameters, providing data support for compiling load spectra of critical components for individual aircraft.