Predicting mechanical properties of RX4E electric aircraft wing composite panels using deep learning

The building block verification system of the RX4E electric aircraft exhibits complexity and diversity in operational conditions at the foundational coupon level, accompanied by issues of experimental complexity and resource intensity. This paper investigates the effects of various environmental conditions (temperature, humidity) and layup methods through a series of experiments. On this basis, a feature-based multi-condition coupled mechanical performance prediction method (FMCPM) is proposed, which can extract spatiotemporal features from multi-condition data and establish the relationship between features and predicted outputs. In addition, features from room temperature data can be extracted and used to predict stress–strain curves under extreme conditions. Results indicate that compressive strength increases in cryogenic environments but decreases in high-temperature and high-temperature, humid conditions. Significant variations in mechanical properties are observed in various layup methods. The proposed model effectively predicts stress–strain curves under two coupling conditions and across three extreme environments and accurately estimates residual and failure loads. This research contributes to the foundation and methodology for real-time health assessment of future wing mechanical properties.