The Method of Diagnostics of Radio-Electronic Means Based on the Analysis of Shock Effects by Means of Machine Learning Algorithms

• Published in: 2021 International Seminar on Electron Devices Design and Production (SED) pp. 1 - 6
• Main Authors: Kondrashov, Daniel E., Bushmeleva, Kiya I., Uvaysov, Saygid U., Bushmelev, Petr E.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27-04-2021
• Subjects: Artificial neural networks; automated system; defects; Electric shock; impact effects; machine learning; Machine learning algorithms; neural network; Printed circuits; Production; radio-electronic means; Seminars; technical diagnostics
• DOI: 10.1109/SED51197.2021.9444499

The content of the article focuses on the problem of insufficient effectiveness of existing methods of diagnostics of radio-electronic means, in particular, aimed at identifying structural defects. The defects of mechanical nature are not given sufficient attention, so the development of a diagnostic system aimed at identifying structural defects is an urgent task of practical significance. To solve this problem, it is proposed to use modern information technologies, including machine learning algorithms, which in turn will be implemented in the developed methodology for diagnosing radio-electronic means based on the analysis of impact impacts and which in turn will involve artificial neural networks based on the recognition of images of structural defects obtained after mechanical impacts. The paper describes in detail the specified diagnostic method, determines the configuration of the artificial neural network and calculates its accuracy, which in turn allows us to conclude that the chosen approach is promising. The paper presents a project of an automated system for diagnosing radio-electronic means based on the analysis of shock effects using neural networks, defines its architecture, designs the main modules, and determines the types of software.