Hybrid fault prognostics for nuclear applications : addressing rotating plant model uncertainty

Blair, J and Stephen, B and Brown, Blair David and Forbes, A and McArthur, S; (2022) Hybrid fault prognostics for nuclear applications : addressing rotating plant model uncertainty. In: PHM Society European Conference. Prognostics and Health Society European Conference, ITA, pp. 58-67. ISBN 9781936263363 (https://doi.org/10.36001/phme.2022.v7i1.3321)

[thumbnail of Blair-etal-PHSC-2022-Hybrid-fault-prognostics-for-nuclear-applications-addressing-rotating-plant-model-uncertainty]
Text. Filename: Blair_etal_PHSC_2022_Hybrid_fault_prognostics_for_nuclear_applications_addressing_rotating_plant_model_uncertainty.pdf
Final Published Version
License: Creative Commons Attribution 3.0 logo

Download (2MB)| Preview


Nuclear plant operators are required to understand the uncertainties associated with the deployment of prognostics tools in order to justify their inclusion in operational decision - making processes and satisfy regulatory requirements. Operational uncertainty can cause underlying prognostics models to underperform on assets that are subject to evolving impacts of age, manufacturing tolerances, operating conditions, and operating environment effects, of which may be captured through a condition monitoring (CM) system that itself may be degraded. Sources of uncertainty in the data acquisition pipeline can impact the health of CM data used to estimate the remaining useful life (RUL) of assets. These uncertainties can disguise or misrepresent developing faults, where (for example) the fault identification is not achieved until it has progressed to an unmanageable state. This leaves little flexibility for the operator’s maintenance decisions and generally undermines model confidence. One method to quantify and account for operational uncertainty is calibrated hybrid models, employing physics, knowledge or data driven methods to improve model accuracy and robustness. Hybrid models allow known physical relations to offset full reliance on potentially untrustworthy data, whilst reducing the need for an abundance of representative historical data to reliably identify the monitored asset’s underlying behavioural trends. Calibration of the model then ensures the model is updated and representative of the real monitored asset by accounting for differences between the physics or knowledge model and CM data. In this paper, an open-source bearing knowledge informed machine learning (ML) model and CM datasets are utilized in an illustrative bearing prognostic application. The uncertainty incurred by the decisions made at key stages in the development of the model’s data acquisition and processing pipeline are assessed and demonstrated by the resultant impact on RUL prediction performance. It was shown that design decisions could result in multiple valid pipeline designs which generated different predicted RUL trajectories, increasing the uncertainty in the model output.