Keynote
Reliability analysis on dynamical systems using manifold-autoregressive surrogate models
Please login to view abstract download link
Reliability assessment of complex engineering systems is often hindered by the high computational cost of simulations and the inability of standard meta-models to handle non-stationary behavior, such as progressive damage. This work addresses these challenges using the mNARX+ framework, a data-driven approach that offers the possibility to integrate "damage trackers" directly into the modeling process. By jointly reconstructing system responses and internal state indicators, mNARX+ effectively captures evolving properties like fatigue accumulation and plastic damage. This approach offers a computationally efficient and robust solution for forecasting the behavior of damaging structures over time.
