10.14489/td.2025.02.pp.023-028

2025, 02 Februar

DOI: 10.14489/td.2025.02.pp.023-028

Poletaev V. A., Palamar I. N., Volkov D. I., Gagarina A. I.
SURFACE QUALITY EVALUATION SYSTEM AFTER SHOTBLASTING BASED ON A DEEP BIMODAL CLASSIFIER
(pp. 23-28)

Abstract. The problem of improving the accuracy of assessing the surface quality of a part after shotblasting is solved. A system for evaluating the surface quality of a part based on the criterion of texture uniformity when a given roughness of the treated surface is achieved based on machine learning is proposed. Surface image analysis is usually performed visually using sets of photo etalons for various materials and using qualitative criteria with verbal descriptions. The use of a deep classifier of surface images based on a convolutional neural network provides a low accuracy of binary classification of the order of 0.65 due to the small volume of the training sample and the difficult to formalize surface texture. The idea of the study is to use a roughness profile to assess the surface quality, since the primary profile is measurable, and the information in the signal taken by the profiler is related to the surface relief. A new architecture of a deep bimodal classifier is proposed with the introduction of two convolutional neural networks for image and signal analysis, as well as a method for generating an output feature vector. Two datasets were used to train the classifier: surface images and corresponding roughness profiles. It was experimentally obtained that the accuracy of the bimodal classifier increased by 12…17 %, depending on the selected processing modes and the type of material.

Keywords: surface quality evaluation, surface image, roughness profile, shot blasting, deep bimodal classifier, machine learning system.

+ - About the Authors Click to collapse

V. A. Poletaev, I. N. Palamar, D. I. Volkov, A. I. Gagarina (State educational institution higher education "Soloviev Rybinsk State Aviation Technical University" (RSATU), Rybinsk, Russia) E-mail: Данный адрес e-mail защищен от спам-ботов, Вам необходимо включить Javascript для его просмотра. , Данный адрес e-mail защищен от спам-ботов, Вам необходимо включить Javascript для его просмотра. , Данный адрес e-mail защищен от спам-ботов, Вам необходимо включить Javascript для его просмотра. , Данный адрес e-mail защищен от спам-ботов, Вам необходимо включить Javascript для его просмотра.

+ - References Click to collapse

1. Tolkachev A. V., Volkov D. I. (2020). Monitoring the results of the shot peening process in production conditions. Vestnik RGATA im. P. A. Solov'eva, 54(3), 69 – 72. [in Russian language]
2. Bezyazychniy V. F., Ryabov A. N., Sutyagin A. N. et al. (2015). Methodology for studying residual stresses in the surface layer of gas turbine engine blades after shot peening. Vestnik RGATA im. P. A. Solov'eva, 32(1), 104 – 108. [in Russian language]
3. Tolkachev A. V., Volkov D. I. (2019). Parameters of the shot peening process that determine the result of the treatment and require mandatory control. Uprochnyayushchie tekhnologii i pokrytiya, Vol. 15 180(12), 542 – 545. [in Russian language]
4. Preparing the steel surface before applying paints and related products. Visual assessment of surface cleanliness. Part 1. The degree of oxidation and the degree of preparation of the uncoated steel surface and the steel surface after complete removal of previous coatings. (2014). Standard No. GOST R ISO 8501-1‒2014. Moscow: Izdatel'stvo standartov. [in Russian language]
5. Palamar' I. N., Pervov M. L. (2019). Formalization of verbal assessments of the quality of shot blasting based on the analysis of images of the metal surface. Uprochnyayushchie tekhnologii i pokrytiya, Vol. 15 170(2), 65 – 69. [in Russian language]
6. Basharina T. A., Demchenko A. A., El'tsov I. S. (2023). Convolutional neural network in the problem of identifying and classifying defects on cylindrical surfaces. Uprochnyayushchie tekhnologii i pokrytiya, Vol. 19 227(11), 526 – 528. [in Russian language]
7. Vasil'ev M. E., Kos'kin A. V., Shalimov A. S. (2024). Automated defect detection on product surfaces based on convolutional neural networks. Vestnik komp'yuternyh i informatsionnyh tekhnologiy, Vol. 21 237(3), 30 ‒ 36. [in Russian language] DOI: 10.14489/vkit.2024.03.pp.030-036
8. Ignat'ev M. A. (2024). Application of machine learning and data mining in an automated system for non-destructive eddy current testing of the surface layer of bearing parts. Vestnik Astrahanskogo gosudarstvennogo tekhnicheskogo universiteta. Seriya: Upravlenie, vychislitel'naya tekhnika i informatika, (3), 26 ‒ 34. [in Russian language]
9. Zaharova O. I., Kulyas O. L., Loshkarev A. S. et al. (2024). Selecting a neural network architecture for searching for defects on the inner surface of pipes. Pribory i sistemy. Upravlenie, kontrol', diagnostika, (1), 51 ‒ 56. [in Russian language]
10. Harchevnikova A. S., Savchenko A. V. (2020). Extracting user preferences based on methods for automatically generating text descriptions of photo album images. Komp'yuternaya optika, 44(4), 618 – 626. [in Russian language] DOI: 10.18287/2412-6179-CO-678

+ - Purchase digital version of a single article Click to collapse

This article is available in electronic format (PDF).

The cost of a single article is 700 rubles. (including VAT 20%). After you place an order within a few days, you will receive following documents to your specified e-mail: account on payment and receipt to pay in the bank.

After depositing your payment on our bank account we send you file of the article by e-mail.

To order articles please copy the article doi:

10.14489/td.2025.02.pp.023-028

and fill out the form