Uncertainty Management and Differential Model Decomposition for Fault Diagnosis and Prognosis
Dongzhen Lyu,Guangxing Niu,Enhui Liu,Tao Yang,Gang Chen,Bin Zhang
DOI: https://doi.org/10.1109/tie.2021.3086706
IF: 7.7
2022-05-01
IEEE Transactions on Industrial Electronics
Abstract:The performance of fault diagnosis and prognosis (FDP) relies greatly on the model that describes the fault behavior and dynamics under study. In the past decades, many researches focused on establishing reliable, robust, and accurate models to describe fault growth dynamics. However, uncertainties in the modeling process severely affect the accuracy and precision of FDP results. A major challenge in FDP is how to deal with the uncertainties that affect the performance of the system. For commonly used state-space model, uncertainties exist in both state transition equation and the measurement equation. In this article, the influence of uncertainties on FDP results is studied and compared in terms of accuracy and precision. Furthermore, this article proposes a differential model decomposition approach, which can automatically acquire appropriate model parameters and uncertainties for the state-space model. The effectiveness and advantages of the proposed method are verified and demonstrated with three different application scenarios, including comparison studies on public data and online experimental verification.
automation & control systems,engineering, electrical & electronic,instruments & instrumentation