Post Your Answer
2 years ago in Prognostics and Health Management (PHM) , Systems Theory By Divya
"How can one predict the remaining useful life (RUL) of a used aeroengine and its components‑ "
Our current practice relies heavily on individual sensor thresholds, which leads to conservative, part-by-part replacements and doesn't capitalize on system-level degradation patterns. For my PhD, I'm surveying hybrid models that combine physics with data. I'm curious about the expert consensus on balancing model complexity with operational practicality, especially when dealing with intertwined failure modes across components like blades, bearings, and the combustion chamber.
Â
All Answers (1 Answers In All)
By Nisha Ali Answered 1 year ago
From my time in aerospace PHM, I've seen the most success with a hybrid, modular approach. I would recommend starting with physics-based models for known failure modes (e.g., creep, fatigue) at the component level to establish "strain zones." Then, fuse real-time sensor data (vibration, temperature, oil debris) using a machine learning layer often a recurrent neural network or particle filter to update these models and capture unexpected interactions. Crucially, you need a system-level Bayesian belief network that propagates updated component reliabilities to compute the overall engine RUL, providing a practical and trustworthy confidence interval for maintenance scheduling.
Â
Reply to Nisha Ali
Related Questions