【作者】 荆龙江；

【导师】 项贻强；

【作者基本信息】 浙江大学 ， 桥梁与隧道工程， 2007， 博士

【摘要】 结构损伤识别方法是桥梁结构健康监测系统的重要组成部分，而基于结构动力响应的损伤识别方法更是目前的研究热点。本文在现有结构损伤识别研究的基础上，以一座大跨度预应力混凝土斜拉桥为主要背景，对桥梁结构损伤识别所面临的难点问题进行了深入研究。本文首先建立了斜拉桥空间精细有限元模型，通过雅可比参数型优化法对建立的有限元模型进行修正研究，获得了与实际桥梁结构状态更加符合的基准有限元模型，并对该斜拉桥进行了模态分析，结果表明该斜拉桥结构固有频率很低，模态密集，存在很多耦合振动的情况。根据多自由度体系的振动特征方程，本文从理论上推导了模态参数中特征频率和模态振型对结构不同位置的灵敏度公式，并以一三跨混凝土连续梁桥模型，对推导的灵敏度公式进行了验证，结果表明模态参数对连续梁桥不同位置的敏感程度是各不相同的，所推导的理论计算方法适用于类似的梁系结构。基于模态曲率和柔度矩阵的概念，本文构造出了两个损伤指标，提出了利用局部最优原则并根据损伤指标曲线的峰值点来对损伤进行判定的方法。根据模态分析结果，对斜拉桥主梁结构中设定的损伤进行了识别研究，研究表明构造的损伤指标能明确的识别出结构的损伤位置及程度而且具有很好的抗噪声能力。利用遗传算法，本文以结构各测点模态柔度变化率为设计变量来构造遗传算法损伤识别所需的适应度函数，对斜拉桥主梁结构中设定的损伤进行了识别研究，研究表明遗传算法结合柔度变化率能明确的识别出结构的损伤位置及程度而且具有较好的抗噪声能力。小波变换是为了解决传统傅里叶变换不能同时在频域和时域取得最佳分辨率而提出的一种新的变换。本文最后研究了小波及小波包分解在桥梁结构损伤预警中的方法和应用。对于不同损伤情况下的时域信号，经过小波及小波包分解后都可以明确的判定结构损伤发生的时刻，达到损伤预警的目的，而且通过小波包分解也验证了高频信号对结构的损伤更加敏感。更多还原

【Abstract】 Structural damage identification techniques are the important components of the health monitoring system of the bridges. The structural damage identification methods which based on the dynamic response of structures are very popular. The dissertation is devoted to break through the difficulty in the damage identification field on the basis of the existing methods.Firstly, a precise three-dimensional finite element model of a prestressed concrete cable-stayed bridge is established. Then the model is updated by a Jacobi parametric optimization method and the baseline FE model is obtained, which is more precise to describe the behavior of the cable-stayed bridge. Lastly, dynamic properties of the cable-stayed bridge are analyzed and the conclusion indicates that it has lower natural frequencies, dense modals and many coupled vibration.The sensitivity formulas of natural frequencies and mode shapes for different position in structure are derived based on the dynamic equilibrium equation of a free-vibration multiple-degree-of-freedom structural system. A three-dimensional finite element model of three-span continuous bridge is established and the sensitivity formulas are verified. In the end, the dissertation draws a conclusion that sensitivity of modal parameters for structure damage is different and the conclusion is also suitable for other beam-type structures.Based on the concept of modal curvature and flexibility matrix, two damage locating indices are constructed and the method which utilizes the local optimum rule to identify the damage is presented. Various damage cases simulated in the main girder are identified. The conclusions indicated that the damage locating indices can detect definitely the position and level of damage and it also has a better ability to resist the noise.Genetic algorithm (GA) is a novel global optimization algorithm appeared in recent years. When GA is applied to the damage identification field, it is necessary to regard the problem as an optimization one. A fitness function of GA is constructed with the flexibility rate of all measured points as the design variables. Various damage cases simulated in the main girder are identified. The conclusions indicated that GA can detect definitely the position and level of damage and it also has a better ability to resist the noise.Wavelet transform, a new mathematical manipulation, can obtain the best resolution both in frequencies domain and in time domain, which is the deficiency of the Fourier Analysis. The wavelet and its package decomposition in the damage alarm of bridges are researched. The signal in time domain decomposed by wavelet and its package can definitely identify the time of day when the damage occurred. So the purpose of damage alarm is reached, also the conclusion which high frequency signal is more sensitive to damage in the structures is verification through wavelet package decomposition.更多还原