Damage localization with modeling errors and uncertainties

摘要

Modeling errors and uncertainties are major obstacles to model-based structural damage localization.When an inappropriate model is chosen for the purpose of damage detection,the identified damage locations and severities can be misleading.This paper focuses on the issues of modeling errors and uncertainties in structural damage localization.A probabilistic framework is proposed to resolve the issues of modeling errors and uncertainties.Two levels of uncertainties will be addressed:uncertainties in model parameters and uncertainties in model selection.The former addresses the issue of the uncertainties in damage localization if the chosen model is appropriate,while the latter addresses the issue of choosing the most appropriate model among several candidate models.A novel stochastic simulation method will be developed to update the above two levels of uncertainties based on monitoring data.Examples will be given to demonstrate the use of the new approach.