如何科学地评估服役桥梁结构的性能是工程领域遇到的难题之一，而基于桥梁静动力实验基础上的结构识别是解决这个难题的最好的途径。由于传统的单模型结构识别理论在面临多种误差来源时存在着局限性，成功的桥梁结构识别例子不多且应用前景并不为桥梁管理部门所看好。本研究将在分析结构识别中误差来源的基础上用多模型的方法进行桥梁结构识别研究，用来探索提高桥梁结构性能评估及决策的可靠性问题。本项目主要研究内容包括，基于桥梁"模型碎片库"的交互式编程的建模研究，基于最大熵原理的多模型结构传感器最优布设方法，基于数据挖掘技术的多模型分类和甄选，以及基于贝叶斯统计理论的多模型结构反应预测。构造的科学问题是通过融合建模、传感器优化布置、数据挖掘和贝叶斯统计方法，利用多模型结构识别方法科学地评估桥梁性能并提供决策支持。该研究对阐明多模型结构识别原理和提高结构识别的可靠性有着重要意义。

How to scientificlly evaluate the in-service bridge performance is one of the difficulties in the field of engineering, and the structural identification (St-Id) of the bridge based on the static and dynamic tests is the best approach to solve this difficult problem. Due to the limitation of the single model St-Id theory when dealing with different kinds of uncertainty sources, there are limited successful bridge St-Id examples and the bridge management authority has no confidence on the prospect of the St-Id applications. The proposed research is to introduce multimodel St-Id stratagy into bridges analysis based on analyzing different kinds of uncertainty sources, and the purpose is to explore how to improve the reliability for structural performance evaluation and decision-making. The proposal includes the following main contents, the application programming interface modeling research based on bridge modeling fragment library, the optimal sensor instrumentation research based on maximum entropy theory, multiple model classification and selection method based on data mining technique, and the bridge response estimation by utilizing multimodel based on Bayesian statistics theory.The framed scientific problem is to scientifically evaluate the bridge performance and to assist decision-making using multiple model St-Id approach by intergrating modeling, optimal instrumentation, data mining and Bayesian statistic theory. The research has important significance to explain the multiple model St-Id theory and to improve the reliability of St-Id.