| Popis: |
A novel integrated prognosis and diagnosis model for airframe structural health management (SHM) is developed to support the United States Army’s Condition Based Maintenance (CBM) project. In the project, the damage of the structure requires detection and diagnosis in real-time. Since the damage that grows over time is unobservable and unknown, the damage size should be estimated by a probability distribution. While the work presented in this paper describes the development and verification of the prognostic and diagnostic model, the developed model can be further used on a helicopter. The status of the structure is measured by a low-fidelity binary indicator (hit/miss) in real time. The uncertainty of the diagnostic system is quantified by the probability of detection (POD) of the damage. The uncertainty of the initial crack size is assigned to a probability distribution that can be determined by the manufacturer’s tolerance or worstcase scenarios. The distribution of the estimated damage is updated as the crack propagates over time using a physical model (for example, Paris’s law for fatigue crack propagation) and a temporal Bayesian approach. The model was verified using data from a fatigue test of a titanium alloy. The results demonstrated that this approach efficiently provides accurate diagnostic and prognostic estimations of the final crack size and its uncertainty bounds. |
