A reliability estimation method based on combination of failure mechanism and ANN supported wiener processes
Di Liu,
Yajing Qiao,
Shaoping Wang,
Siming Fan,
Dong Liu,
Cun Shi,
Jian Shi
Affiliations
Di Liu
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China; State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou 310027, China; Tianmushan Laboratory, Xixi Octagon City, Yuhang District, Hangzhou 310023, China; Key Laboratory of Flight Techniques and Flight Safety, CAAC, Guanghan 618307, China
Yajing Qiao
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China
Shaoping Wang
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China; Tianmushan Laboratory, Xixi Octagon City, Yuhang District, Hangzhou 310023, China
Siming Fan
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China
Dong Liu
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China
Cun Shi
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China
Jian Shi
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100191, China; Corresponding author.
For some engineering application, accurately estimating reliability only depend on the history data or failure mechanism is difficult to implement, due to the lack of data and imperfect theory of failure mechanism. Namely, both history data and failure mechanism should be utilized to improve the reliability estimation accuracy for engineering applications. Hence, we construct a reliability estimation method by fusing the failure mechanism and artificial neural network (ANN) supported Wiener processes for utilizing both history data and failure mechanism. ANN and failure mechanism are integrated into Wiener process with random effects, respectively. Bayesian model averaging (BMA) method is adapted to combine the failure mechanism with ANN supported Wiener processes, as well as to update the model parameters by fusing data. Based on a typical aviation hydraulic pump's actual dataset, we illustrate the advantages of our approach by comparing to Wiener process supported only by ANN or failure mechanism in engineering practices. The proposed method shows superiorities on reliability estimation considering the estimation accuracies comparing the other two models.
