Diagnostics and Control of Gas Valve Electric Drive Based on Machine Learning Methods

Автор: Golovenko E.A., Kinev E.S., Pavlov E.A., Shalaev P.O., Lukyanov E.N., Litovchenko A.V., Bryzgova X.A., Pomozov E.I., Smirnaya A.A.

Журнал: Журнал Сибирского федерального университета. Серия: Техника и технологии @technologies-sfu

Рубрика: Исследования. Проектирование. Опыт эксплуатации

Статья в выпуске: 6 т.18, 2025 года.

Бесплатный доступ

The article presents the results of a study of the operating modes of an automated electric drive in liquid and gas transport control systems, in which trouble-free operation of transport lines and units requires the use of reliable shut-off and control equipment and efficient control algorithms. The project sets one of the objectives of developing a test program for 50 products to determine a comprehensive reliability indicator based on a full factorial experiment and identifying a comprehensive reliability indicator in terms of GOST R 27.102–2021 “Object Reliability” based on the use of statistical research methods, failure-free operation, maintainability and recoverability, and storability. The main suppliers that provide their customers with not only services, but also entire ecosystems for condition diagnostics are IBM (USA), Oracle Corporation (USA), Microsoft Corporation (USA), SAP SE (Germany), AMAZON (USA) and others. There is no data on domestic solutions. Based on the ranking, by the number of triggers for certain parameters, it is possible to draw a conclusion with a high degree of probability about the cause causing such effects. The growth of triggers over the measurement interval will show the most probable cause of system failure in the future, taking into account the constant operating time per shift.

Еще

Pipeline valves, automated electric drive, full factorial experiment, reliability, machine learning, dataset labeling

Короткий адрес: https://sciup.org/146283189

IDR: 146283189   |   УДК: 681.5