Detection and Identification of Defects in 3D-Printed Dielectric Structures via Thermographic Inspection and Deep Neural Networks

Abstract

In this paper, we propose a new method based on active infrared thermography (IRT) applied to assess the state of 3D-printed structures. The technique utilized here—active IRT—assumes the use of an external energy source to heat the tested material and to create a temperature difference between undamaged and defective areas, and this temperature difference is possible to observe with a thermal imaging camera. In the case of materials with a low value of thermal conductivity, such as the acrylonitrile butadiene styrene (ABS) plastic printout tested in the presented work, the obtained temperature differences are hardly measurable. Hence, the proposed novel IRT method is complemented by a dedicated algorithm for signal analysis and a multi-label classifier based on a deep convolutional neural network (DCNN). For the initial testing of the presented methodology, a 3D printout made in the shape of a cuboid was prepared. One type of defect was tested—surface breaking holes of various depths and diameters that were produced artificially by inclusion in the printout. As a result of examining the sample via the IRT method, a sequence of thermograms was obtained, which enabled the examination of the temporal representation of temperature variation over the examined region of the material. First, the obtained signals were analysed using a new algorithm to enhance the contrast between the background and the defect areas in the 3D print. In the second step, the DCNN was utilised to identify the chosen defect parameters. The experimental results show the high effectiveness of the proposed hybrid signal analysis method to visualise the inner structure of the sample and to determine the defect and size, including the depth and diameter.

Deep learning modeling strategy for material science: from natural materials to metamaterials

“…The modeling strategy is effective for modern complex materials (such as composite materials). The use of DCNNs has been proven to be able to identify the defect parameters of 3D printing materials [130], and the use of RNN models can also predict the plastic deformation of materials with unique geometric shapes (such as 3D materials) [52].…”

“…The use of hybrid models can also solve the optimization design of multilayer coatings in perovskite solar cells [129] and evaluate the status of 3D printing structures [130]. A more complex hybrid model is used to predict the properties of the grain boundary (GB).…”

Identification of Grain Oriented SiFe Steels Based on Imaging the Instantaneous Dynamics of Magnetic Barkhausen Noise Using Short-Time Fourier Transform and Deep Convolutional Neural Network

“…Due to their properties, CNN networks find many applications in computer vision and image recognition [ 41 ]. Applications in object recognition and tracking [ 41 ], medicine—for imaging diagnosis of diseases [ 52 ] or in technology—for the assessment of the state of structures [ 42 , 45 , 53 , 54 , 55 , 56 ] or design [ 57 ], and testing the properties [ 53 ] of materials, comprise a special area. For example, in [ 52 ], the effectiveness of detecting carious lesions was on average between 82% and 89%.…”

“…The CNN are also used in the field of non-destructive testing (NDT). In [ 54 ] and [ 55 ], the CNNs have been used for the automatic defect identification, respectively, in steel elements based on magnetic imaging and in 3D printouts by thermal examination, achieving an overall accuracy of over 90%. The CNN are also used to detect changes in dynamic systems and classify their states based on STFT spectral characteristics [ 58 , 59 , 60 ].…”

An Evaluation of 3D-Printed Materials’ Structural Properties Using Active Infrared Thermography and Deep Neural Networks Trained on the Numerical Data

“…Extracting such a collection from the same data set, which is then assessed by a trained network, is the most often employed approach. In our previous work, we have shown that, for this purpose, the convolutional neural networks may be used [ 40 ], and other researchers have worked on a similar problem [ 41 ]. This results in the obvious issue of a network adjusting to a single−use case, rendering it very inflexible and unsuited for jobs requiring some automation.…”