Research On Detection Technology Of Composite Materials Based On Planar Array ECT

Composite materials have many excellent properties,such as corrosion resistance,fatigue resistance,high temperature resistance,good damping performance and so on,so they are widely used in aerospace,military and other fields.Due to the influence of the production environment in the manufacturing process,or the impact and other external forces in the use process,the structure and properties of the composites will change,resulting in cracks,delamination and other damage,which may increase the risk of aircraft flight.Therefore,it is necessary to detect the damage and defects of the composites.Electrical Capacitance Tomography(ECT)is a new type of non-destructive testing,which has the advantages of fast,intuitive,economic and environmental protection,and has been more and more widely used.The working principle of ECT is based on the change of different dielectric constant of different substances,which makes the capacitance between the measuring electrodes change,so as to visually detect the parts that may be damaged by external forces in a composite object.In this paper,the comprehensive analysis and defect detection of honeycomb composite materials are carried out based on the aviation composite material detection technology of plane electrical capacitance tomography.A damage detection model of aviation composite materials based on convolutional neural network is constructed to detect the region and coordinate location of defects,and a new deep neural network based on multi-scale residual coding and decoding path is proposed for image reconstruction.The main work is as follows:(1)The imaging principle of planar ECT nondestructive testing technology is introduced from the forward and inverse problems.Planar ECT technology is used to detect the defects of honeycomb composite materials.At the same time,a three-dimensional simulation model of typical defects of honeycomb composites based on COMSOL multi-physical field simulation software is established.(2)Three-dimensional sensitivity distribution imaging and layered imaging of aeronautical honeycomb composites are studied.Different traditional image reconstruction algorithms are compared,and the imaging effects are compared by comparing the image correlation coefficient and the image relative error,and applied to the subsequent construction of composite defect detection network based on multi-scale fusion.(3)The damage models of honeycomb composites in different positions are set up and processed as the training sample data of one-dimensional convolution neural network.Based on one-dimensional convolution neural network,a honeycomb composite damage detection model is constructed to detect the region and coordinate location of defects.The test results show that the constructed network can better identify the location and area of damage,which provides technical support for aviation honeycomb composite damage detection.(4)A deep neural network Ms RED based on multi-scale residual coding and decoding path is proposed and verified,and higher image correlation coefficient and lower image relative error are obtained,which verifies the effectiveness of Ms RED imaging algorithm.