结构性能退化的评估和贝叶斯动态预测

 结构性能的退化影响因素复杂、时间跨度长、不确定性大，其预测不可避免地要考虑具体结构的实际情况，即在一般规律认识的基础上，不断利用在具体结构生命周期中得到的新信息，更新对结构未来性能退化的预测，进而得到更确切合理的可靠度分析和决策。基于这一思路，引入贝叶斯动态模型理论，设计模型结构，用于结构性能退化的预测。同时，根据结构性能退化的多阶段特点和目前工程实践可获取的结构检测信息的实际情况，提出状态指标的概念，用于描述结构性能退化。计算实例验证贝叶斯动态预测的优越性及信息更新对结构性能退化预测和寿命评估的影响，同时说明状态指标具有简便实用的特点。

EVALUATION AND BAYESIAN DYNAMIC PREDICTION OF DETERIORATION OF STRUCTURAL PERFORMANCE

The deterioration of structural performance is a time-variant process with large amounts of uncertainties and incompleteness. It seems to be inevitable that the prediction of the deterioration of the structural performance should be based on a philosophy of information updating. A new system for structural performance prediction is introduced based on Bayesian dynamic models theory. The system aims to incorporate information from any relevant source，leading to amended and updated model structures. Intervention and monitoring techniques are also designed to ensure the stability of the system. A specific structure of Bayesian dynamic model is defined for structural performance prediction. So-called condition indexes are also defined to evaluate the structural performance. Using condition indexes，the deterioration condition of the structural performance can be described by visual inspection information in current engineering practice. Case study illustrates the advantages of Bayesian dynamic prediction. The information updating has favorable effects on reliability analysis and life prediction. The condition indexes are simple and convenient.