基于无人机视觉的飞机蒙皮损伤智能检测系统

打开文本图片集

中图分类号：TP18；V267 文献标识码：A 文章编号：2096-4706（2025）21-0098-05

Intelligent Detection System for Aircraft Skin Damage Based on UAVVision

AOLiangzhong，LIQingman （College ofAviation Engineering，Civil Aviation FlightUniversityofChina，Chengdu ，China）

Abstract：Theskinisanimportantpartoftheaircraft structure，anditshealth status isdirectlyrelatedtoflightsafetyand operational eficiency.Traditional detectionreliesonmanualvisualandnon-destructivetesting methods，whichhaveproblems suchaslowefiecyhighfalsetectioatendstrongsubjectiviyanditisdifult toeettheqirementsofe detection foraccuracyandcomplexity.Aiming at theabove problems，an inteligent damage detection system based o UAV visionandDeep Learing is developed.Theintroductionof UAV technology solves the problem thatthe surfacearea of the aireraftiscomple，theaeaisuge，anditisdifculttofulloverthemanual.Withthehelpoftetargetdetectionalgorit， theautomaticecogitionabilityanddetectionaccuracyoftheskindamageareimproved，andthemissddetectiondfalse detectionarereduced.Atesametime，adetectionsystemwithreal-timeandhighefciencyasthecoreisdesignedtometthe actual nedsofrapidresponseinaviationmaintenance.Theexperimentalresultsshowthatthesystemcancoverareasthatare difficult to detect manually. The model recognition accuracy reaches 87% ，and the mAP @0.5 is 87.3% . In practical applications， the efficient real-time detection effect of real-time detection during flight is realized.

Keywords：Deep Learning;damage detection;aircraft skin; UAV

0 引言

传统的飞机蒙皮损伤检测主要依赖人工目视或接触式手段，存在因蒙皮区域复杂、面积庞大而导致人工难以全面覆盖的问题。（剩余5455字）