Abstract: Aircraft structural strength test is the most important verification method of aviation structure at present, and it is an indispensable and important link in the aircraft development process. At present, the measurement method of aircraft structural strength test is rela-tively simple, and only limited and discrete response data can be obtained, and it is difficult to obtain information during the whole process and the whole field of the test, which limits the comprehensive analysis and processing of data. In view of this, this paper proposes a data-driven virtual-real fusion algorithm to construct a digital twin model suitable for structural strength test by fusing simulation data and test data, so as to achieve high-precision prediction of the mechanical properties of test objects. The algorithm is divided into two stages: pre-training and real-time prediction. In the pre-training stage, the simulation data is used to train the particle swarm optimization random forest model (PSO-RF). In the real-time prediction stage, based on the error between the trained PSO-RF model and the experimental data, the radial basis function multi-fidelity surrogate model (RBF-MFS) is trained. Finally, by fus-ing the PSO-RF model and the RBF-MFS model, a digital twin model of the test object is constructed. The results show that the accuracy of the model on the test set is basically about R2=0.97, and the calculation time on the 330,000 grid nodes is only 0.8s, and the prediction error of the model is less than 10% for the danger area of the wing box segment, which meets the needs of practical engineering applications and provides a reference for the digitization of aircraft structural strength tests.

Key words: digital twin, virtual-real fusion, structural strength test, random forest, multi-fidelity surrogate model