Research On Inside Defect Detection Of Multilayer Metal Lattice Structure Based On CT Images

The lattice structure is a new type of ordered porous material with structure-function integration,which has the characteristics of ultra-light,high specific strength,high efficiency heat dissipation and heat insulation.The lattice structure prepared by Additive Manufacturing(AM)is of great significance and promotion for the development of aerospace technology and aerospace power.In the process of additive manufacturing,a large amount of residual stress is easily generated to cause structural defects,which greatly reduces the performance of the prepared material and has a serious impact on the life and reliability of the structure.Therefore,the non-destructive testing of metal lattice structures prepared by additive manufacturing technology is of great significance.In this paper,multi-layer metal lattice structure components fabricated by Selective Laser Melting(SLM)technology are studied,and a new method for detecting and identifying internal defects of lattice structures is studied.Firstly,aiming at the complex internal structure of metal lattice structure and its difficulty in direct detection,a method for detecting internal defects of metal lattice structure based on industrial CT scanning is proposed.The image reconstruction of the metal lattice structure is performed using the high permeability and attenuation characteristics of industrial CT rays.Secondly,according to the periodic distribution of the metal lattice structure and the distribution characteristics of the pixel gray value in the CT tomogram,a defect detection and recognition method based on the gray value distribution of the CT tomogram pixel gray image is proposed.The variation of the difference between the gray value of the pixel set of a certain size is used as the characteristic of a typical defect of the lattice structure.The theoretical analysis gives the corresponding discriminant conditions to realize the identification and location of the defect.Finally,combined with the excellent feature extraction ability of deep learning,an automatic recognition method for internal defects of lattice structure based on deep learning is proposed.Combined the convolutional neural network,a feature learningnetwork is designed based on the Faster R-CNN network architecture.Some defects in the metal lattice structure are translated,flipped,reduced and enlarged to prepare internal defect samples of the metal lattice structure.The defect detection model is obtained by training the defect training set.And from the two aspects of recognition accuracy and algorithm running speed,the two automatic defect recognition methods proposed in the paper are compared and analyzed to verify the effectiveness of each algorithm in identifying the internal defects of a multi-layer complex lattice structure.