Diaz-Gonzalez, Abel, Coursey, Austin, Quinones-Grueiro, Marcos, & Biswas, Gautam. (2024). A flexible data-driven prognostics model using system performance metrics. In Proceedings of the 12th IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes (SAFEPROCESS 2024), Ferrara, Italy, June 4-7, 2024, Volume 58, Issue 4, Pages 222-227. https://doi.org/10.1016/j.ifacol.2024.07.221
This study focuses on improving the prediction of remaining useful life (RUL) in systems, which is critical for maintenance and reliability management. Traditional data-driven methods for RUL prediction often result in rigid models that cannot easily adapt to changing user requirements or performance thresholds. To address this limitation, the authors propose a new prediction architecture that draws inspiration from physics-based solutions.
Their approach allows the RUL model to adjust its level of conservatism—essentially how cautious or aggressive the prediction is—without the need to retrain the model when the performance criteria are changed. This flexibility is particularly valuable in practical applications where system performance specifications can vary over time. The method was tested on a full-wave rectifier with multiple degrading components, and the results show that the model can accurately adapt to different performance scenarios, demonstrating the potential for broader application in predictive maintenance across various systems.
