An integrated degradation modeling framework considering model uncertainty and calibration. (1st March 2022)

• Record Type: Journal Article
• Authors: Lin, Yan-Hui; Ding, Ze-Qi
• Is Part Of: Mechanical systems and signal processing. Volume 166(2022)
• Publication Date: 2022-03-01
• Subjects: Degradation data analysis -- General path model -- Stochastic process model -- Wavelet density estimation -- Model calibration -- Model uncertainty; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2021.108389
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: An integrated degradation modeling framework based on the wavelet density estimation is proposed. Assumption of zero mean normally distributed errors of general path models is released. Model uncertainty for stochastic process models is dealt with. The effectiveness and feasibility of the proposed method are illustrated through a numerical example and a case study. Abstract: General path models and stochastic process models are two widely applied categories of probabilistic degradation models. The former explains the randomness of degradation data as normal distributed errors with zero mean. The latter describes degradation with stochastic processes such as Wiener process, Gamma process and Inverse Gaussian process. For general path models, a limitation is the assumption of normally distributed errors. For stochastic process models, model uncertainty with respect to the available stochastic processes should be considered, but the widely-applied model selection methods in consideration of model uncertainty are unable to warn when all the candidate models fit data poorly. Therefore, an integrated degradation modeling framework based on wavelet density estimation is proposed, which can calibrate the distribution of errors for general path models and deal with model uncertainty for stochastic process models. The proposed framework can select the best stochastic process if certain stochastic processes fit the degradation data well. Otherwise, all the candidate stochastic processes can be calibrated, which overcomes the drawback of model selection methods. The effectiveness and feasibility of the proposed framework are illustrated through a case study and a numerical example.