考虑温度效应的桥梁结构模态参数识别技术及损伤识别方法研究

Detecting and localizing damages for bridge have obtained worldwide attention. The changes in dynamic characteristics of structure are offen used as indicators because of its sensitivity to damage. However, temperature loads may also cause changes in these dynamic features, which can lead to a false alarm. The purpose of this research is to propose an effective damage identification strategy which can discriminate temperature effects. Firstly, the theoretical calculation model between modal properties and temperature effect is derived through modal analysis based on transient temperature distribution and finite element method; Secondly, an accurate temperature control and testing system is designed. The influence law of temperature on physical-mechanical parameters and modal properties are obtained through laboratory tests, respectively. The inherent mechanism that temperature can raise variation of modal properties is discussed based on the experimental results and multivariate correlation analysis of grey system. Thirdly, parameters that have greater impact on modal property considering temperature effect are selected as design variables and a model updating technique that can consider temperature effect is proposed. Finally, a reliable damage identification strategy that can reject the temperature effects, external vehicle load, noise and errors is demonstrated by using the overpass project as examples.

桥梁结构的损伤识别是世界各国面临的重大难题，利用模态参数对损伤的敏感性可以有效解决这个问题，然而温度荷载等容易引起结构模态参数的非正常变化，导致识别方法的失效。本研究针对考虑温度效应的桥梁结构损伤识别这一前沿课题，首先通过瞬态温度有限元分析及温度应力矩阵修正，进行结构模态分析，获取温度影响桥梁模态参数的理论计算模型；其次基于室内试验和灰色系统多变量关联度分析，阐明温度影响结构模态参数的内在机理；再次，选取温度荷载作用下对桥梁模态参数影响较大的因素构造模型修正设计变量，通过桥梁结构温度与模态参数的内在关系，在构建精确桥梁初始模型基础上，提出考虑温度效应的模型修正技术；最后，以实体工程立交桥为工程实例，考虑温度效应、外部服役荷载、有限元模型修正误差、测试噪声等因素影响，提出一种具备良好鲁棒性的损伤识别方法。本研究将为解决考虑温度效应的桥梁结构损伤识别提供一种新的思路。

基于动力特性的损伤识别方法在桥梁健康监测与损伤识别领域得到了广泛应用。然而，环境因素，特别是温度会造成桥梁模态参数的显著改变，使基于动力特性的损伤识别方法失效。因此，分析温度对桥梁模态参数的影响规律，阐明影响机理，形成有效的模态参数温度影响剔除及损伤识别方法，对提高基于动力特性的损伤识别准确性，促进其工程实用化具有重要作用。本项目通过对钢筋混凝土简支梁的长期监测，揭示了季冻区环境影响下简支梁桥模态参数的变异规律与内在影响机理。考虑温度不均匀分布对模态频率的影响，构建了钢筋混凝土简支梁模态频率量化模型，剔除了模态频率的温度影响。针对剔除温度影响后模态频率存在的随机性与不确定性，提出了基于休哈特均值控制图与粒子群优化模糊C均值聚类的损伤预警与评价方法。基于自联想神经网络、多元休哈特控制图与粒子群优化模糊C均值聚类算法，提出了不依赖温度测试数据的损伤诊断与评价方法。针对剔除温度影响后模态参数不确定性对损伤定位与局部损伤程度识别的影响，基于均匀荷载面曲率差与模糊推理理论，提出了一种具有较强鲁棒性的损伤识别方法。采用基于灵敏度与Tikhonov正则化方法相结合的模型修正方法，将柔度残差作为目标函数，通过修正混凝土弹性模量值使得各节点柔度残差值接近于0，从而使得有限元模型和实际结构相吻合。依托安龙泉互通立交桥实体工程，验证了方法的有效性。