Multilevel Seismic Damage Quantification for Inclined RC Piers: Integrating Fiber‐Based Simulation and Machine Learning Prediction

Inclined reinforced concrete (RC) piers, common in mountainous highway bridges, can exhibit heightened seismic vulnerability. We investigate inclination–damage mechanisms using a fiber‑based OpenSees model combined with a hierarchical damage quantification framework. Pushover and nonlinear time‑history analyses (NTHA) are synthesized with a bidirectional LSTM to estimate material‑, section‑, and structure‑level damage indices with quantified uncertainty. We observe a bimodal damage pattern for inclinations between ~ 0.5° and 1.0°, and find that displacement‑ductility limits should be progressively reduced as pier inclination increases from 0° to 2°. For earthquakes with peak ground velocity (PGV) ≳0.5 m/s, the LSTM‑based monitoring shows promise for real‑time assessment. Given the single configuration, single recorded motion, and single network used here, results should be interpreted cautiously. The proposed framework has the potential to evaluate across different structural configurations and seismic environments.