Radiotekhnika
Publishing house Radiotekhnika

"Publishing house Radiotekhnika":
scientific and technical literature.
Books and journals of publishing houses: IPRZHR, RS-PRESS, SCIENCE-PRESS


Тел.: +7 (495) 625-9241

 

Neural network methods for integral structure non-destructive testing

DOI 10.18127/j19998554-201902-06

Keywords:

A. Yu. Viryasova – Master, Bauman Moscow State Technical University
E-mail: virnastya@yandex.ru
A. I. Vlasov – Ph.D. (Eng.), Associate Professor, Department of Design and Production Technology of Electronics, Bauman Moscow State Technical University
E-mail: vlasovai@bmstu.ru
A. A. Gladkikh – Ph.D. (Eng.), Associate Professor, Department of Design and Production Technology of Electronics, Bauman Moscow State Technical University
E-mail: a.gladkikh@inforion.ru


The article considers the issues of automating the classification of defects in the topology of integral structures, which are represented by the results of optical control in the form of pixel images. In order to solve this problem, three main classification algorithms have been considered: K-nearest neighbors algorithm, feedforward neural network and convolutional neural network. It has been shown that the disadvantage of the K-nearest neighbors method is the need to keep the huge base of standards and to compare the input image with each one of the databases, together with the high susceptibility of this method to rotations, object’s position on the classifier and scaling. The disadvantages of feedforward neural networks include a large number of parameters to be taught, low learning rate, inability to parallelize computations and to implement network learning algorithms on graphics processors and susceptibility to the object position offset in the input data.
It is worth paying attention to the fact that the problem of classification of defects is narrowed to the classification and search for features of images, for the solution of which this paper proposes using of the machine learning method.
As a result of the research, it has been established that, in order to ensure the high quality of integrated structures defects classification at their input check, it is advisable to use convolutional neural networks, since convolutional neural networks provide partial immunity to scaling, offsets, rotations, angle changes and other distortions.
Despite considerable attention to the problems of non-destructive testing and non-destructive testing of integral structures, a number of issues remain unresolved. Further research subjects will focus on assessing the completeness and depth of the formation of knowledge bases of reference topological images of integral structures basic elements that could be used to solve classification problems (standardize the defect’s image size, viewing angle and focus).

References:
  1. Khartov V.V. Kosmicheskie problemy elektroniki: pered upotrebleniem – vzboltat. Elektronika NTB. 2007. № 7. S. 22–25. [in Russian]
  2. Vlasov A.I., Gridnev V.N., Konstantinov P., Iudin A.V. Neirosetevye metody defektoskopii pechatnykh plat. Elektronnye komponenty. 2004. № 8. S. 148–155. [in Russian]
  3. Buianov A.I., Vlasov A.I., Zagoskin A.V. Primenenie neirosetevykh metodov pri defektoskopii pechatnykh plat. Neirokompiutery: razrabotka, primenenie. 2002. № 3. S. 42–70. [in Russian]
  4. Vlasov A.I., Gridnev V.N., Panfilova S.P., Chervinskii A.S. Beskontaktnyi teplovoi kontrol elektronno-vychislitelnykh sredstv. Tekhnologiia i konstruirovanie v elektronnoi apparature. 2007. № 6 (72). S. 42–49. [in Russian]
  5. Vlasov A.I., Gridnev V.N., Panfilova S.P., Chervinskii A.S. Beskontaktnyi teplovoi kontrol izdelii elektronnoi tekhniki. Proizvodstvo elektroniki. 2007. № 3. S. 25–30. [in Russian]
  6. Alekseev M.A., Arabov D.I. Lineinye modeli raspoznavaniia v sistemakh mashinnogo zreniia. Sb. dokl. Vosmoi Vseros. konf. molodykh uchenykh i spetsialistov «Budushchee mashinostroeniia Rossii». M.: MGTU im. N.E. Baumana. 2015. S. 366–370. [in Russian]
  7. Grigorev V.P., Kamyshnaia E.N., Nesterov Iu.I., Nikitin S.A. Primenenie metodov iskusstvennogo intellekta v SAPR tekhnologi¬cheskikh protsessov proizvodstva elektronnoi apparatury. M.: Izd-vo MGTU im. N.E. Baumana. 1998. [in Russian]
  8. Grigorev V.P., Nesterov Iu.I., Cherepanov D.V. Informatsionnye tekhnologii v konstruirovanii i tekhnologii mikroelektroniki. Razdel «Primenenie imitatsionnogo modelirovaniia dlia prognozirovaniia i otsenki rabotosposobnosti izdeliia pri razrabotke mikroelektronnykh integralnykh struktur». M.: Izd-vo MGTU im. N.E. Baumana. 2000. [in Russian]
  9. Nazarov A.V., Fomin A.V., Dembitskii N.L. i dr. Avtomatizatsiia proektirovaniia matrichnykh KMOP BIS / Pod red. A.V. Fomina. M.: Izd-vo MAI. 1991. [in Russian]
  10. Nazarov A.V. Mnogokomponentnoe 3D-proektirovanie nanosistem / Pod red. V.A. Shakhnova. Ser. Biblioteka «Nanoinzheneriia». Kn. 4. M.: Izd-vo MGTU im. N.E. Baumana. 2011. [in Russian]
  11. Dembitskii N.L., Nazarov A.V. Primenenie metodov iskusstvennogo intellekta v proektirovanii i proizvodstve radiotekhnicheskikh ustroistv. Ser. Nauchnaia biblioteka. M.: Izd-vo MAI. 2009. [in Russian]
  12. Markelov V.V., Kabaeva A.S. Upravlenie kachestvom elektronnykh sredstv. Ser. Biblioteka «Konstruirovanie i tekhnologiia elektronnykh sredstv». T. 2. M.: Izd-vo MGTU im. N.E. Baumana. 2014. [in Russian]
  13. Markelov V.V., Vlasov A.I., Kamyshnaia E.N. Sistemnyi analiz protsessa upravleniia kachestvom izdelii elektronnoi tekhniki. Nadezhnost i kachestvo slozhnykh sistem. 2014. № 1 (5). S. 35–42. [in Russian]
  14. Vlasov A.I., Markelov V.V., Zoteva D.E. Upravlenie i kontrol kachestva izdelii elektronnoi tekhniki. Sem osnovnykh instrumentov sistemnogo analiza pri upravlenii kachestvom izdelii elektronnoi tekhniki. Datchiki i sistemy. 2014. № 8 (183). S. 55–66. [in Russian]
  15. Rowley H.A., Baluja S., Kanade T. Neural network-based face detection. IEEE Trans. on Pattern Analysis and Machine Intelligence. 1998. V. 20. P. 23–38. DOI: 10.1109/34.655647
  16. Juell P., Marsh R. A hierarchical neural network for human face detection. Pattern Recognition. 1996. V. 29. № 5. P. 781–787.
  17. Lin S.-H., Kung S.-Y., Lin L.-J. Face recognition/detection by probabilistic decision-based neural network. IEEE Trans. on Neural Networks. 1997. V. 8. P. 114–132.
  18. Balukhto A.N., Bulaev V.I., Buryi E.V. i dr. Neirokompiutery v sistemakh obrabotki izobrazhenii. Ser. Biblioteka zhurnala «Neirokompiutery: razrabotka, primenenie». T. 7. M.: Radiotekhnika. 2003. [in Russian]
  19. Balukhto A.N., Galushkin A.I., Kovalchuk D.V., Nazarov L.E., Tomashevich N.S. Neirokompiutery v prikladnykh zadachakh obrabotki izobrazhenii. Ser. Biblioteka zhurnala «Neirokompiutery: razrabotka, primenenie». T. 8. M.: Radiotekhnika. 2003. [in Russian]
  20. Shakhnov V.A., Vlasov A.I., Poliakov Iu.A., Kuznetsov A.S. Neirokompiutery: arkhitektura i skhemotekhnika. Prilozhenie k zhurnalu «Informatsionnye tekhnologii». № 9. M.: Mashinostroenie. 2000. [in Russian]
  21. Lawrence S., Giles C.L., Tsoi A.C., Back A.D. Face recognition: a convolutional neural network approach. IEEE Trans. on Neural Networks. Special Issue on Neural Networks and Pattern Recognition. P. 1–24.
  22. Uossermen F. Neirokompiuternaia tekhnika: Teoriia i praktika. M.: Mir. 1992. [in Russian]
  23. Prokhorov V.G. Ispolzovanie svertochnykh setei dlia raspoznavaniia rukopisnykh simvolov. Institut programmnykh sistem NAN Ukrainy. 2004. [in Russian]
  24. CS231n convolutional neural networks for visual recognition: Image classification [Elektronnyi resurs] / URL: http://cs231n. github.io/classification/
  25. Rysmiatova A.A. Ispolzovanie svertochnykh neironnykh setei dlia zadachi klassifikatsii tekstov. M.: Moskovskii gosudarstvennyi universitet imeni M.V. Lomonosova. 2016. [in Russian]
May 29, 2020

© Издательство «РАДИОТЕХНИКА», 2004-2017            Тел.: (495) 625-9241                   Designed by [SWAP]Studio