| Summary: | The structural integrity of a structure is dependent on all of its internal components that connected together to provide support to various loadings. However, due to the existence of errors in structural properties, reliability updating has always been a crucial field of study in structural health monitoring. These inaccuracies may lead to over or under prediction of structural behaviors. Bayesian updating are applied to compute the posterior distribution of uncertain model parameters and model outputs in general, using strong vibration data. However, a large and complex building structure will always have high dimensional parameters in reality. Significant amount of computational efforts has to be involved to compute the integration of multivariate. Therefore, an algorithm based on subset simulation is proposed as an enhancement of the classic rejection sampling algorithm for Bayesian updating, which the number of random variables does not affect its efficiency. As a further matter, due to the extremely low failure probability required for modern buildings, enormous number of samples need to be generated so that the accuracy of the model could be achieved. By utilizing subset simulation, the original extremely rare event could be converted into a series of frequent series problems that are easier to solve.
|
|---|