Coupled prediction method for assembly precision and performance of composite structures based on a hybrid saint-venant’s principle and neural network approach

Abstract

The application of composite materials and interference fit technology in aerospace products presents new challenges to assembly quality requirements: specifically, the demand for higher assembly precision and reduced assembly stress, as these factors directly impact the aerodynamic performance and service life of the product. Consequently, a large number of assembly deviation and stress predictions are necessary during the aircraft structure design process. To meet the requirements for prediction accuracy and efficiency under the constraints of large data volumes and high computational costs, this study proposes an innovative method for the rapid prediction of assembly precision and performance coupling (CPAP) in composite structures. This method combines Saint-Venant’s principle with finite element analysis (FEA) to create an efficient sample generation technique that can quickly provide key data on assembly deviations and stress around the interference fit holes (SAH). Additionally, dimensionality reduction techniques are incorporated into the metamodel (MM), effectively capturing the nonlinear relationships between assembly process parameters and both assembly precision and performance. This results in a predictive coupling model with statistical analysis capabilities. Case studies demonstrate that the method proposed in this study significantly improves prediction efficiency compared to traditional approaches. Furthermore, the results highlight the substantial influence of interference fit process parameters on the assembly accuracy and performance of single longitudinal splicing (SLS) joint structures. This research offers an effective tool for controlling the assembly quality of aerospace products, contributing to technological innovation and advancements in the aerospace industry.