Aircraft Skin Damage Detection And Life Cycle Analysis Method Based On Machine Vision

In order to ensure the safety of flight, the aircraft skin should be checked out at fixed period toensure structural damage should be found in time and repaired to restore to design structural strengthbefore structural strength decreases to residual structural strength.The main research of this thesis is as followed:Firstly, the development history and current research status of nondestructive testing fordetecting damage of aircraft skin were reviewed, and the advantages and defects of all kinds ofnondestructive testing were also analyzed.Second, the machine vision system based on wireless sensor network was established, on thisbasis,the aircraft skin images were collected and transmited to ground platform, pre-processspecifically at aircraft skin images was developed to eliminate noise to make the information includedin images would be much more approximate to that we need, according to the aircraft structural repairmanual and the different textures with different damages, the aircraft skin images were classified intonormal, crack, corrosion,impact and scratch in the image library of aircraft skin damages.Third, the wavelet packet coefficients (WPC) and gray-level co-occurrence matrixes(GLCM) ofthe images were extracted in turn as the feature vector, then the30-dimensional feature vector wasreduced to14dimension based on distance measurements.Finally, the support vector machine (SVM) algorithm was introduced, and the deficiencies ofSVM in the process for multi-class problems classification was analysed, the samples’ separatingcapacity could be improved by defining the fuzzy connectivity, then the Fuzzy Support VectorMachine (FSVM)based on sample affinity method has been used to classify the damages, whichconsiders the distance measuring characteristics between sample classes in Euclidean space. Theexperiment results showed that the proposed method is a very effective method.The geometriccharacteristics of the the crack damage were extracted, makging research on the fatigue life predictionof structures about aircraft skin cracks.The research of this thesis is of great significance for the automated detection of aircraft skin,and ensuring the safety of flight.