Bayesian updating and model class selection of deteriorating hysteretic structural models using recorded seismic response

作者: Matthew M. Muto , James L. Beck

DOI:

关键词:

摘要: Identification of structural models from measured earthquake response can play a key role in health monitoring, control and improving performance-based design. System identification using data strong seismic shaking is complicated by the nonlinear hysteretic structures where restoring forces depend on previous time history rather than an instantaneous finite-dimensional state. Furthermore, this inverse problem ill-conditioned because even if some components structure show substantial yielding, others will exhibit nearly elastic response, producing no information about their yielding behavior. Classical least-squares or maximum likelihood estimation not work with realistic class it be unidentifiable based data. On other hand, combination Bayesian updating model selection provides powerful rigorous approach to tackle problem, especially when implemented Markov Chain Monte Carlo simulation methods such as Metropolis-Hastings, Gibbs Sampler Hybrid algorithms. The emergence these stochastic recent years has led renaissance across all disciplines science engineering high-dimensional integrations that are involved now readily evaluated. power handle system problems demonstrated recently-developed algorithm, Transitional Carlo, perform Masing relatively simple yet give responses loading. Examples given deteriorating building simulated

参考文章(0)