| Abstract: | The deterioration of structural performance is a time-variant process with a large amount of uncertainties and incompleteness of load and environmental effects. It seems inevitable that the prediction of structural deterioration should be based on a philosophy of information updating. In the present paper, a new model system for structural performance prediction is introduced based on Bayesian dynamic linear model (DLM) theory. The system can implement Bayesian updating on the prediction of the time series process of structural deterioration. It can effectively incorporate useful information through the deterioration process of structures to update the prediction. Intervention and monitoring techniques are also designed to ensure the stability of the model system. This paper also defines an indicator of the so-called ‘condition index’ to evaluate the structural performance. Using condition indexes, the qualitative condition rating of structural performance based on visual inspection in current engineering practice can be integrated into the Bayesian DLMs. Case studies validate the advantages of Bayesian DLMs. The information updating has a favourable effect on reliability analysis and life prediction. The condition indexes are simple and convenient used in practice. |
