CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
A DATA-DRIVEN METHOD FOR REMAINING USEFUL LIFE PREDICTION OF MULTIPLE-COMPONENT SYSTEMS
 |
Read | Sai Van Cuong, Shcherbakov Maksim V. A DATA-DRIVEN METHOD FOR REMAINING USEFUL LIFE PREDICTION OF MULTIPLE-COMPONENT SYSTEMS // Caspian journal : management and high technologies. — 2019. — №1. — pp. 33-44. |
/33-44.jpg)
Sai Van Cuong - post-graduate student, Volgograd State Technical University, 28 Lenin AvРµ., Volgograd, 400005, Russian Federation, svcuonghvktqs@gmail.com
Shcherbakov Maksim V. - Doct. Sci. (Engineering), Professor, Volgograd State Technical University, 28 Lenin AvРµ., Volgograd, 400005, Russian Federation, maxim.shcherbakov@vstu.ru
Supporting the operation of the equipment at the operational stage with minimal costs is an urgent task for various industries. Classical approaches to maintenance of systems lose their effectiveness in modern conditions. The article describes a data-driven method of proactive maintenance of equipment based on predicting the remaining useful life (RUL). The main purpose of this research is to develop a method of predicting the RUL aimed at minimizing operating costs in the maintenance of equipment. Contributions of the paper are: i) a new architecture of real-time predictive maintenance system; ii) the study of the effectiveness of different approaches (both deep neural networks and typical algorithms of machine learning) in RUL predicting; iii) a new hybrid CNN+LSTM model based on the combination of convolutional neural networks (CNN) and long short-term memory networks (LSTM), superior analogues in solving the problem of predicting the RUL using Turbofan Engine Degradation Simulation Data Set from NASA
Key words: интернет вещей, прогностическое обслуживание, остаточный ресурс (RUL), машинное обучение, глубокое обучение, CNN, LSTM, XGBoost, SVM, Random Forest, internet of thing, predictive maintenance, remaining useful life, machine learning, deep learning, CNN, LS