Recognition of defects in the blade apparatus of turbomachines using a neural network
Keywords:
convolutional neural networks, YOLOv8, model, defect, detection, blade apparatus, axial compressor, gas turbine engineAbstract
Correct, serviceable and high-quality operation of a turbomachine directly depends on the condition of its flow path. One of the reasons for the decrease in the efficiency and reliability of the turbine unit is the wear of the blade apparatus due to the formation of various defects on the edges and surface of the blade airfoil. This study develops an approach to automating the defect detection process of products using modern computer vision technologies. In this work, a program code was prepared in the Python programming language, a database of images of blade defects was created, and the YOLOv8 model was trained and tested. The achieved accuracy in determining the type of defect in the study exceeded 80%.
Metrics
References
ГОСТ
1. Коваль С. И. Методы диагностики технического состояния газотурбинных двигателей в процессе эксплуатации и технического обслуживания // XXI век: итоги прошлого и проблемы настоящего плюс. – 2019. – Т. 8, № 2(46). – С. 53-58. EDN CKPNZH
2. See J.E. Visual inspection: a review of the literature [Электронный ресурс] Albuquerque (USA): Sandia National Laboratories, 2012. – 77 p. URL: https://digital.library.unt.edu/ark: /67531/metadc835891/ (Дата обращения: 22.11.2023).
October 1, 2012
3. Aust J., Pons D. Bowtie Methodology for Risk Analysis of Visual Borescope Inspection during Aircraft Engine Maintenance // Aerospace. – 2019. Vol. 6(10). – P. 110. DOI: 10.3390/aerospace6100110.
4. Exploring impact and features of machine vision for progressive industry 4.0 culture / M. Javaid, A. Haleem, R.P. Singh et al. // Sensors International. – 2022. – Vol. 3(5). – P. 100132. DOI: 10.1016/j.sintl.2021.100132.
5. Neuhauser F.M., Bachmann G., Hora P. Surface defect classification and detection on extruded aluminum profiles using convolutional neural networks // International Journal of Material Forming. – 2020. – Vol. 13. – P. 591-603. [Электронный ресурс]. DOI: 10.1007/s12289-019-01496-1.
6. Цифровой подход к обнаружению дефектов лопаточного аппарата и оценке их влияния на характеристики турбомашин / В.Л. Блинов, О.В. Беляев, В.И. Брезгин, О.В. Комаров // Турбины и Дизели. – 2023. – № 3. – С. 38-44. EDN: NSZKFZ
APA
1. Koval, S. I. (2019). Metody` diagnostiki texnicheskogo sostoyaniya gazoturbinny`x dvigate-lej v processe e`kspluatacii i texnicheskogo obsluzhivaniya [Methods for diagnosing the technical condition of gas turbine engines during operation and maintenance]. XXI vek: itogi proshlogo i problemy` nastoyashhego plyus, 2(46) , 53-58. [In Russian]
2. See, J. E. (October 1, 2012). Visual inspection: a review of the literature. Sandia National Laboratories. https://digital.library.unt.edu/ark: /67531/metadc835891/.
3. Aust, J., & Pons, D. (2019). Bowtie Methodology for Risk Analysis of Visual Borescope Inspection during Aircraft Engine Maintenance. Aerospace, 6(10) , 110.
https://doi.org/10.3390/aerospace6100110
4. Javaid, M., Haleem, A., Pratap, R. Singh, Rab, S., & Suman R. (2022). Exploring impact and features of machine vision for progressive industry 4.0 culture. Sensors International, 3(5) , 100132. http://dx.doi.org/10.1016/j.sintl.2021.100132
5. Neuhauser, F. M., Bachmann, G., & Hora, P. (2019). Surface defect classification and detection on extruded aluminum profiles using convolutional neural networks. International Journal of Material Forming, 13, 591-603. https://doi.org/10.1007/S12289-019-01496-1
6. Blinov, V. L., Belyaev, O. V., Brezgin, V. I., & Komarov, O. V. (2023). Cifrovoj podxod k obnaruzheniyu defektov lopatochnogo apparata i ocenke ix vliyaniya na xarakteristiki turbomashin [Digital approach to detecting blade defects and assessing their impact on the characteristics of turbomachines]. Turbiny` i Dizeli, 3, 38-44. [In Russian]
Downloads
Published
How to Cite
Issue
Section
Categories
URN
License
Copyright (c) 2023 V. Blinov, Zh. Ivan
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
