Adaptive staged remaining useful life prediction method based on multi-sensor and multi-feature fusion. (March 2023)

• Record Type: Journal Article
• Authors: Ta, Yuntian; Li, Yanfeng; Cai, Wenan; Zhang, Qianqian; Wang, Zhijian; Dong, Lei; Du, Wenhua
• Is Part Of: Reliability engineering & system safety. Volume 231(2023)
• Publication Date: 2023-03
• Subjects: Remaining useful life prediction -- Adaptive staged prediction -- Multi-sensor and multi-feature fusion -- Parameters estimation; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2022.109033
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: This paper proposes a RUL prediction framework from data fusion to staged predict. A MSMFF method is proposed for the fusion of multiple degraded information. An ASP method is proposed for component staged prediction. Different parameter-solving methods are integrated to facilitate the solution of parameters. Two groups of experiments were completed to verify the effectiveness of the method. Abstract: The single sensor is difficult to acquire the complete degradation information of the component, and the degradation model cannot adaptively track the staged degradation process (DP) of the component, which lead to a decrease in the accuracy of the remaining useful life (RUL) prediction methods. Therefore, this paper proposes an adaptive staged RUL prediction (ASP) method based on multi-sensor and multi-feature fusion (MSMFF). Firstly, a MSMFF method is proposed, which uses information contribution rate and degradation indicator (DI) suitability to initially fuse vibration signals and features respectively. Updates the MSMFF technology with the prognosis information of different degradation indicators, so as to facilitate the construction of component final DI. Secondly, an ASP method is proposed, which adaptively makes different degradation models match the different degradation stages of the component. Based on the definition of the first hitting time, the probability density function of the ASP method is obtained for predicting the RUL of components. Then, a four-step method is proposed to estimate and update the unknown parameters in the model to solve the problem of parameter complexity. Finally, two different sets of experiments are carried out to verify the effectiveness and superiority of the proposed method.