Aerodynamic data of aircraft can be obtained through methods such as numerical simulations, wind tunnel tests, and flight tests. However, obtaining high-fidelity aerodynamic data with low uncertainty at a low cost remains a challenging problem. In deep learning methods, multi-layer neural networks can represent and combine features in higher dimensions, effectively integrating relevant features from different data sources to improve the model's prediction accuracy and reliability. However, the existing methods still face numerous technical challenges when fusing features from different datasets. To address this issue, we propose a multi-source data fusion method based on Radial Basis Function Generative Adversarial Network (RBFGAN)—RGAN-MSFM. This method can fuse aerodynamic datasets from different sources, mining the complementarity and correlation between multi-source aerodynamic data features. The proposed method first uses RBFGAN to fuse complementary state features from different data sources to generate a new fused dataset. Then, functional and contextual networks are constructed to explore the correlation among the data sources, enabling fusion learning of multi-source data features. We designed two sets of simulation experiments based on datasets containing CFD and wind tunnel test data, as well as CFD and flight test data. The experimental results show that after data feature fusion using this model, the accuracy of the datasets improved by an average of 30.2% and 67.5%; compared with traditional data fusion methods, the uncertainty was reduced by 23.9% and 27.6%, respectively.

Compared to traditional aerodynamic modeling methods, the deep-learning-based aerodynamic modeling has higher modeling efficiency and precision. However, previous deep learning methods with fully connected network (FCN) or convolutional neural network (CNN) ignored the influence of input data discrepancies. The shape characteristic parameters and flow state parameters of aircraft are greatly discrepant with different types, which will cause different degrees of impacts on the predicted results. When these two types of parameters are used to predict aerodynamic characteristics together, if we ignore the data discrepancy between them, the accuracy of prediction will be lost. Inspired by the multi-task learning method, we propose Large-Discrepancy Multi-task Learning Network(LD-MTL). Our method firstly divides the dataset into multiple tasks, and then splits the whole learning network into multiple clusters to learn relevant knowledge of the predicted aerodynamic performance according to different tasks. Finally, the relevant knowledge learned from individual clusters are then fused into the final prediction result. Through comparative tests, it is shown that our method can better reflect and quantify the influence of data discrepancies and have a higher prediction accuracy when modeling with largely discrepant aerodynamic data.