The health of aircraft landing gear is related to flight safety. In order to real-time grasp the fatigue accumulation of landing gear and achieve maintenance according to the situation, it is necessary to accurately predict the landing load data of landing gear in real time. This study is based on actual test flight data of a certain aircraft model, and introduces the Yolo v11 algorithm model. CBAM attention mechanism, C3Chost module, and DyHead object detection head are used to improve it, forming a landing gear landing load prediction model based on the improved Yolo v11 algorithm. In terms of data processing, 220956 samples (209956 training sets and 11000 test sets) were cleaned and feature dimensionality reduced, and 13 flight parameters were selected as features using various methods. In the experiment, the improved model will be compared with the Yolo v11 and Yolo v11En models, and evaluated using metrics such as R ², MSE, and average relative error. The experimental results show that, based on the R ² metric, the three models perform similarly on the training and testing sets, without overfitting. In terms of MSE indicators, the Proposed Method model performed outstandingly in the test set, with MSEs of 0.3700 (Task 1 test set composite value) and 0.3607 (Task 2 test set composite value), respectively. Compared with the other two models, the MSE on the vertical load test sets of the left and right main landing gear decreased by more than 66%; The MSE difference between the training set and the test set is smaller, indicating stronger stability. In terms of average relative error, the proposed method model has an average prediction error of 1.21% and 1.19% for the vertical load on the left and right main landing gear, respectively, which is significantly better than the other two models. Research has shown that the landing gear load prediction model based on the Proposed method model has high prediction accuracy, strong stability, and good reliability, and has higher value and potential in practical applications.
