@Article{cmes.2020.08867,
AUTHOR = {Shenkun Zhao, Chao Jiang, Zhe Zhang, Xiangyun Long},
TITLE = {Robust Remaining Useful Life Estimation Based on an Improved Unscented Kalman Filtering Method},
JOURNAL = {Computer Modeling in Engineering \& Sciences},
VOLUME = {123},
YEAR = {2020},
NUMBER = {3},
PAGES = {1151--1173},
URL = {http://www.techscience.com/CMES/v123n3/39309},
ISSN = {1526-1506},
ABSTRACT = {In the Prognostics and Health Management (PHM), remaining useful
life (RUL) is very important and utilized to ensure the reliability and safety of
the operation of complex mechanical systems. Recently, unscented Kalman filtering (UKF) has been applied widely in the RUL estimation. For a degradation system, the relationship between its monitored measurements and its degradation
states is assumed to be nonlinear in the conventional UKF. However, in some special degradation systems, their monitored measurements have a linear relation
with their degradation states. For these special problems, it may bring estimation
errors to use the UKF method directly. Besides, many uncertain factors can result
in the fluctuations of the estimated results, which may have a bad influence on the
RUL estimation method. As a result, a robust RUL estimation approach is proposed in this paper to reduce the errors and randomness of estimation results
for this kind of degradation problems. Firstly, an improved unscented Kalman filtering is established utilizing the Kalman filtering (KF) method and a linear adaptive strategy. The linear adaptive strategy is used to adjust its noise term
adaptively. Then, the robust RUL estimation is realized by the improved UKF.
At last, three problems are investigated to demonstrate the effectiveness of the
proposed method.},
DOI = {10.32604/cmes.2020.08867}
}