Application of artificial neural networks for identification of non-normative errors in measuring instruments for controlling the induction soldering process
Скачать файл:
URI (для ссылок/цитирований):
https://www.sgem.org/documents/programme/_Publish_Programme_PUBLISH.pdfhttps://elib.sfu-kras.ru/handle/2311/110516
Автор:
Тынченко, В. С.
Милов, А. В.
Тынченко, В. В.
Бухтояров, Владимир Викторович
Антамошкин, О. А.
Коллективный автор:
Институт нефти и газа
Институт космических и информационных технологий
Гуманитарный институт
Кафедра технологических машин и оборудования нефтегазового комплекса
Кафедра информатики
Кафедра информационных технологий в креативных и культурных индустриях
Дата:
2018-05Журнал:
International Conference on GeoinformaticsКвартиль журнала в Scopus:
без квартиляБиблиографическое описание:
Тынченко, В. С. Application of artificial neural networks for identification of non-normative errors in measuring instruments for controlling the induction soldering process [Текст] / В. С. Тынченко, А. В. Милов, В. В. Тынченко, Владимир Викторович Бухтояров, О. А. Антамошкин // International Conference on Geoinformatics: Geoinformatics and Remote Sensing. — 2018.Текст статьи не публикуется в открытом доступе в соответствии с политикой журнала.
Аннотация:
The paper deals with the problem of the influence of non-normative errors on the process of induction soldering of space vehicles waveguide paths. The application of the method of neural network modeling is proposed to solve the problem posed, which is a problem of classifying the errors of measuring instruments. As a tool, Deductor Academic software is used, which allows to conduct a wide range of analytical studies. The work presents the development of the structure of an artificial neural network for the classification of errors, as well as the choice of the artificial neural network training method. As input neurons of the network, it is suggested to use retrospective values of the temperature difference in two parts of the heated piece immersed in the lag space. For the experimental studies, a sample of 463 implementations of the induction soldering process, divided into the training and test sets, was used. As a result, a classifier was obtained that allows to detect with an accuracy of 91% the non-normative errors of measuring equipment. It allows to improve the control quality of induction soldering process of the space vehicles waveguide paths. In the future, it is supposed to use artificial neural networks not only for identification, but also for solving the problem of compensating the non-normative errors of measuring instruments.