MACHINE LEARNING FOR DEFECT DETECTION IN PRODUCT MANUFACTURING

UDC 004.032.2
DOI: 10.36871/2618-9976.2023.06.004

Authors

Viktor V. Erokhin,
Doctor of Technical Science, Docent, Moscow state institute of international relations (university), Ministry of Foreign Affairs of the Russian Federation, Moscow, Russia
Arkady R. Koryakin,
Graduate Student, Moscow State Technical University named after N.E. Bauman, Moscow, Russia

Abstract

In the production of products, the detection of defects in the process of their manufacture ensures a reduction in the cost of manufacturing a product and an increase in labor productivity. Modern product defect detection methods often rely on simple defect detection methods and manual validation of data from sensors that monitor the quality of the product and the manufacturing process. To improve the quality of a product manufacturing line, it is necessary to develop machine learning tools for more reliable and accurate detection of product defects. The use of machine learning to detect product defects in the construction of reference cycles, the use of clustering and time series forecasts to detect product defects is presented. A method for automated detection of product defects has been developed by combining the methods of cluster analysis and time series forecasting. The study of the detection of marriage is carried out on a production line for the manufacture of products of the construction industry of mass use from composite materials.

Keywords

Artificial neural networks, Clustering, Production process, Modeling, Forecasting