Open AccessA remaining useful life prediction method for complex systems based on multi-index fusion with MC and HSMM
Author(s) -
D. F. Zhang,
Yang Ran,
G. B. Zhang,
Canghong Jin,
S. L. Li
Publication year - 2021
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/1043/4/042028
Subject(s) - computer science , reliability engineering , key (lock) , degradation (telecommunications) , reliability (semiconductor) , markov chain , index (typography) , data mining , machine learning , engineering , telecommunications , power (physics) , physics , computer security , quantum mechanics , world wide web
Remaining useful life (RUL) estimation is the core and basic of system Prognostic and Health Management (PHM) and also a challenging for complex systems. It is necessary to use performance indicators that are closely related to system status for analysis, due to the Multi-indicator different characterization change of system degradation, different detectability and the degree of correlation caused by system coupling. As the system status degradation shows certain scientific laws in macro, their would be certain random relationship between the system status degradation and RUL. To address these problems, the concept of imperfect condition monitoring followed by the concept of key performance indicators in order to reduce the blindness of analysis object selection. The condition degradation probability index is proposed to represent the status degradation degree of the system, whose future trend is fitted by Markov matrix obtained by the improved algorithm as a implementation of mapping condition monitoring data to CDPI. Finally, the system RUL estimation method at time t combined the hidden semi-Markov model with improved forward variable is given to realize the mapping CDPI to RUL. Experiments are carried out to validate the key concepts of the developed methods, and results suggest the effectiveness.