Data and Mechanism Modeling: Residual Life Start-End Determination for Systems with Stable Equilibrium State

• Published in: IEEE transactions on instrumentation and measurement Vol. 73; p. 1
• Main Authors: Xia, Qian, Yue, Jiguang, Chen, Jichang, Cui, Zhexin
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Computational modeling; Data models; DC/DC; Degradation; degradation modelling; digital twin; Digital twins; Fault diagnosis; Life prediction; Mathematical models; Measurement; Parameters; Predictive models; stable equilibrium state; Steady-state; System failures
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2024.3373047

Accurate fault diagnosis and remaining useful life (RUL) prediction are critical for health management before fatal system failures occur. Reasonable stage prediction of RUL can avoid the waste of computing power. However, the difficult-to-measure multiple metrics of real systems cause the starting and end point of RUL to be challenging to determine effectively. Based on the data distribution and digital twin technique, this paper proposes a method to collect the steady-state output signal to determine the beginning and end points of the stable equilibrium state system (SET) RUL. The combination of DT technology can accurately reproduce the actual operating state and measurable characteristics of the system, update the component degradation model by collecting the current component parameters in real-time after the starting point of RUL and inject the future parameter state of the component into DT. The DT is equivalent to a multi-indicator acquisition device, which solves the problem of the physical system's difficulty in acquiring multiple indicators determines the remaining life of the system. Achieving a reasonable starting point and accurate endpoint remaining life prediction under the dual constraints of data and mechanism. The Superbuck converter is used as an example for experiments, and the results prove that the method can reasonably determine the system's start and end points to predict the remaining life. The comparison of the three methods proves the superior performance of the method in this paper.