Development of quality control systems in the manufacturing industry based on the application of computer vision technology

[Đ. R. Mijailović]

III, 123 lista

Primena tehnologije koju obuhvata Industrija 4.0. pruža mogućnost rešavanja velikog broja izazova iz oblasti proizvodne industrije. U okviru ove doktorske disertacije izvršeno je istraživanje u oblasti kontrole kvaliteta u proizvodnim procesima. Razvijen je sistem na osnovu realnih potreba za kontrolu kvaliteta proizvoda koji je zasnovan na primeni tehnologije kompjuterske vizije. Na osnovu istraživanja zaključeno je da sistem treba biti fleksibilan izazovima proizvodne industrije u skladu sa primenjenom tehnologijom. Osnovni cilj je razvoj sistema zasnovanog na kompjuterskoj viziji za identifikaciju neusaglašenosti proizvoda u proizvodnoj industriji. Svrha rada se zasniva na implementaciji razvijenog sistema za kontrolu kvaliteta radi automatizovane identifikacije neusaglašenosti proizvoda u proizvodnom procesu, što doprinosi u pogledu efikasnosti i preciznosti identifikacije i unapređenju kvaliteta proizvoda. Glavni tehnološki fokus je na primeni tehnologije kompjuterske vizije dok veliki doprinos imaju i neuronske mreže, algoritmi identifikacije internet stvari, skladištenje podataka u oblaku kao i ugrađeni sistemi. Ishodi rada se zasnivaju na efikasnoj identifikaciji neusaglašenosti proizvoda, integracija sa postojećim proizvodnim industrijskim sistemima, brza obrada podataka u realnom vremenu. Pre svega neophodno je izvršiti akviziciju podataka, obučiti sistem za definisanu neusaglašenost, implementirati neophodan hardver na proizvodnu jedinicu, obezbediti komunikaciju proizvodne jedinice i softverskog dela sistema. Sistem dalje na osnovu algoritama za identifikaciju analizira prikupljene podatke u realnom vremenu i skladišti rezultate o nousaglašenostima proizvoda. Matlab analizom podataka u oblaku dolazi se do ukupne efikasnosti proizvodne jedinice. Na osnovu razvijenog sistema zaključeno je da svaka proizvodna jedinica zahteva jedinstveno rešenje koje je specifične namene i isključivo je namenjeno za upotrebu na toj proizvodnoj jedinici. Sa aspekta primene sistem je razvijen za rad u okviru proizvodne industrije u kolaboraciji sa postojećim proizvodnim mašinama kao i za rad u okviru centra za kvalitet. Sistem je zasnovan na primeni algoritama identifikacije, prenosivom hardveru, fleksibilnoj obuci sistema, što čini sistem modularnim. Neophodno je težiti razvoju fleksibilnih rešenja koja će omogućiti brzu interakciju korisnika kao i brzu promenu mesta primene. Ovo svakako dovodi do toga da tehnologija može pružiti velike mogućnosti za dalja istraživanja i razvoj.

Application of technology covered by Industry 4.0. provides the possibility of solving a large number of challenges in the area of the manufacturing industry. As part of this doctoral dissertation, research was carried out in the field of quality control in production processes. A system was developed based on real needs for product quality control, which is based on the application of computer vision technology. Based on the research, it was concluded that the system should be flexible to the challenges of the manufacturing industry in accordance with the applied technology. The main goal is to develop a computer visionbased system for identifying product nonconformities in the manufacturing industry. The purpose of the work is based on the implementation of the developed quality control system for the automated identification of product non-conformities in the production process, which contributes to the efficiency and precision of identification and improvement of product quality. The main technological focus is on the application of computer vision technology, while neural networks, Internet of Things identification algorithms, data storage in the cloud, and embedded systems also make a major contribution. The results of the work are based on the efficient identification of product non-conformities, integration with existing production industrial systems, fast processing of data in real time. First of all, it is necessary to perform data acquisition, train the system for the defined non-conformity, implement the necessary hardware on the production unit, ensure communication between the production unit and the software part of the system. Based on identification algorithms, the system further analyzes the collected data in real time and stores the results on product non-conformities. Through Matlab data analysis in the cloud, the overall efficiency of the production unit is reached. Based on the developed system, it was concluded that each production unit requires a unique solution that is of specific purpose and is exclusively intended for use on that production unit. From the aspect of application, the system was developed for work within the production industry in collaboration with existing production machines, as well as for work within the quality center. The system is based on the application of identification algorithms, portable hardware, flexible system training, which makes the system modular. It is necessary to strive for the development of flexible solutions that will enable quick user interaction as well as a quick change of the place of application. This certainly means that the technology can provide great opportunities for further research and development.

Industrija 4.0, Kompjuterska vizija, Kontrola kvaliteta, Internet stvari, Matlab.

Industry 4.0, Computer vision, Quality control, Internet of things, Matlab.

62:004.652(043) 004.652

Industrija

Tekst