Abstract:

Excessive damage to aircraft braking moving discs will directly threaten flight safety. In order to solve the problems of low detection efficiency and strong subjectivity in the evaluation of damage degree of braking moving disc based on manual experience, an intelligent evaluation algorithm for damage degree and replacement need of aircraft braking moving disc based on image segmentation was proposed. Firstly, the classification and feature information of braking moving disc damage were analyzed. Then, based on the U-Net model, semantic segmentation was performed on the braking moving disc damage image dataset. A quantitative analysis model for the proportion of braking moving disc damage area and a calculation model for the maximum radial width of the fall off area were constructed. Finally, the model was validated using image data of the braking moving disc of Cessna525. The results indicate that the U-Net model has a good image segmentation effect on the types of braking moving disc damage, with average accuracy, average recall, average pixel accuracy, and average intersection to union(IoU) of 90.75%, 91.25%, 90.25%, and 87.25%, respectively. The evaluation results of braking moving disc are basically consistent with the evaluation conclusions of experienced mechanical engineers, and the evaluation results are objective, accurate, and highly visualized, which proves the rationality and feasibility of the algorithm proposed in this paper.

Key words: braking moving disc, damage degree, replacement need, intelligent evaluation, image segmentation, U-Net model