Nondestructive Testing Method Of General Aircraft Surface Based On Hyperspectral Imaging

Periodic inspection of general aircraft(GA)has always been an important part of field operation and maintenance.Because GA is exposed of complex and severe external environment for a long time,the skin-coating system on its surface is prone to damage.At present,the surface damage detection technology of GA used in the outfield is still mainly manual visual detection,which has problems such as low efficiency,low reliability and high labor cost.In recent years,people have vigorously developed various kinds of automatic nondestructive testing techniques based on spatial images which can be used in the outfield.Because this kind of technology depends on a large number of high-resolution spatial images,its processing efficiency is poor and it is difficult to meet the needs of the actual operation of the outfield,so it is urgent to develop new nondestructive testing technologies.Hyperspectral imaging is a cutting-edge technology which can directly obtain the spatial image and spectral information of the target simultaneously.It has been gradually applied in the field of rapid nondestructive testing.At present,there are few GA surface nondestructive testing researches based on hyperspectral imaging at home and abroad.Therefore,we have built two sets of near-infrared hyperspectral imaging damage detection systems indoor and outdoor.With the help of the indoor detection system and the use of various spectral matching algorithms and spectral feature algorithms,damage identification research was conducted on self-made simulated samples and trainer skin cut samples,respectively.On this basis,a GA surface nondestructive testing method based on hyperspectral imaging,which can be used in the outfield,is proposed.Finally,an outfield verification experiment is designed and carried out using the outdoor detection system.The main research findings of this dissertation include:1.With Image-λ-N17E hyperspectral imager is the core equipment,and two sets of non-destructive testing systems based on near-infrared hyperspectral imaging have been designed and built,indoor and outdoor.2.The classification and recognition effect of damage was measured based on total classification accuracy P_a and Kappa coefficient.Among several spectral matching algorithms(SA,MD,SID,SCC)and their combination algorithms,the accuracy of the combination algorithm is significantly better than any of them,but the computational complexity and time consumption of the former will also significantly increase;In terms of the experimental results of a single algorithm,both the skin sample and the reference sample are the most sensitive to the spectral correlation coefficient(SCC)parameter.3.In the study of spectral feature algorithms,three spectral feature variables were established,namely ratio variable r,quadratic fitting variable q,and cubic fitting variable t.The experimental results on skin samples show that using variable q to identify damage has a good effect,while based on the joint feature variable r-q-t,the recognition effect is the best.4.A nondestructive testing method for GA surface in outfield based on hyperspectral imaging technology is established.The validation experimental results indicate that the hyperspectral imaging detection method established in this dissertation can effectively identify skin samples and wing surface damage under the outfield conditions,so the method can realize the classification and recognition of surface damage of GA.