A physics‐informed Bayesian framework for characterizing ground motion process in the presence of missing data

Abstract A Bayesian framework to characterize ground motions even in the presence of missing data is developed. This approach features combination seismological knowledge ( a priori ) with empirical observations (even incomplete) via inference. At its core neural network model that probabilistically learns temporal patterns from motion data. Uncertainties are accounted for throughout framework. Performance has been quantitatively demonstrated various scenarios. provides general solution dealing records by providing forms representation probabilistic manner, allowing it be adopted numerous engineering and applications. Notably, compatible versatile Monte Carlo simulation scheme, such stochastic dynamic analyses still achievable Furthermore, serves as complementary current ground‐motion models data‐scarce regions under growing interests PBEE (performance‐based earthquake engineering), mitigating data‐model dependence dilemma due paucity data, ultimately, fundamental limited problem scarce regions.