Bayesian Hierarchical Model-Based Information Fusion for Degradation Analysis Considering Non-Competing Relationship

Degradation analysis methods are increasingly used in reliability assessment of long-life products. The small sample size problem is worthy of attention in the degradation analysis. Bayesian hierarchical method is employed for fusing multi-source degradation information to handle small sample proble...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2019, Vol.7, p.175222-175227
Main Authors: Guo, Junyu, Zheng, Hualin, Li, Binglin, Fu, Guo-Zhong
Format: Article
Language:English
Subjects:
Analytical models
Bayes methods
Bayesian analysis
Bayesian hierarchical model
Data integration
Degradation
degradation analysis
exoskeleton robot
Exoskeletons
Hydraulic equipment
Parameter estimation
Reliability
Reliability analysis
Spool valves
uncertainty
Valves
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Degradation analysis methods are increasingly used in reliability assessment of long-life products. The small sample size problem is worthy of attention in the degradation analysis. Bayesian hierarchical method is employed for fusing multi-source degradation information to handle small sample problems. The competing relationship multiple degradation models have been widely researched and applied. However, there is the non-competing multiple degradation situation exist in mechanical products. In this paper, a Bayesian hierarchical model-based information fusion degradation method with gamma process model is introduced for describing the non-competing relationship multiple degradation process. Therefore, the Bayesian method and the MCMC method are employed for parameters estimation. A numerical example of spool valve of hydraulic system in exoskeleton robot is provided to demonstrate the proposed Bayesian hierarchical model-based information fusion degradation method.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2019.2955932