Сomparative analysis of two methods for detecting anomalies using the example of industrial equipment operation

Echkina, E.; Nazarov, A.; Sushko, A.

Авторы: Echkina E., Nazarov A., Sushko A.
Журнал: Journal of Physics: Conference Series
Год издания: 2024
Издательство: IOP Publishing
Местоположение издательства: [Bristol, UK], England
Аннотация: In factories and industries, most equipment is equipped with sensors that record data about the performance of the equipment and its technical condition. Data can be recorded at different frequencies: from once per minute to several measurements per day. In one year, up to tens of thousands of measurements are collected for each equipment and they contain rich information about the performance and condition of the equipment. Currently, most data is analyzed only by comparison with threshold values or manually. Such a comparison is made only at the time of data recording and conclusions about the state of the equipment can only be drawn at the moment. An additional complexity in the correct interpretation and processing of data lies in the need to know the domain area of the equipment’s operation and the principle of operation of a particular equipment. This article proposes methods for assessing the future condition of an equipment, predicting potential failures, and analyzing anomalies in equipment operation based on historical data. For this, trend analysis is used in conjunction with anomaly detection using decision trees and the k-means method. The article provides comparisons of various methods for detecting anomalies and predicting equipment operation, as well as comparisons of the effectiveness and quality of the methods.
Добавил в систему: Ечкина Евгения Юрьевна

	ИСТИНА	Войти в систему Регистрация
	ИПМех РАН
	Главная Поиск Статистика О проекте Помощь

ИСТИНА

ИПМех РАН

Сomparative analysis of two methods for detecting anomalies using the example of industrial equipment operationстатья Исследовательская статья

Прикрепленные файлы