【作者】 杨雅勋；

【导师】 李子青；

【作者基本信息】 长安大学 ， 桥梁与隧道工程， 2008， 博士

【摘要】 桥梁是交通运输系统的重要组成部分，桥梁结构的健康状态是交通运输系统正常运转的前提。桥梁结构在其服役期间，由于受到荷载、环境腐蚀及自然灾害等作用，损伤不可避免。一旦损伤达到一定程度，将导致桥梁结构发生破坏，对国民经济和人民的生命财产安全造成巨大的损失。因此，对桥梁结构进行健康监测研究，有重要的理论意义和实用价值。结构损伤识别方法是桥梁结构健康监测的基础和重要组成部分，近年来人们一直试图寻找能够适用于桥梁结构的损伤识别方法。基于动力测试的结构损伤识别方法是目前国内外研究的热点。虽然该方法的研究已经取得了很大的进展，提出了很多的损伤识别理论和方法，但是由于受到人们认识水平、试验测试技术和现场条件的限制，使得基于动力测试的结构损伤识别的研究仍然进展缓慢，存在很多问题，目前也难以有效应用于工程实践。因此，研究能适合工程实际应用，并且损伤识别结果可靠准确的结构损伤识别方法具有十分重要的现实意义。本文在现有结构损伤识别方法的基础上，针对目前研究中存在的问题，对桥梁结构损伤识别方法进行了研究，主要内容如下：1．在对现有传感器优化布置方法分析比较的基础上，针对传统方法的不足，提出了基于能量系数-有效独立的传感器优化布置方法。本文所提出的传感器优化布置方法是以反映节点自由度模态应变能的系数来修正反映最大线性无关的有效独立法，使传感器布置方案进一步综合优化。以简支钢桁架梁桥为数值算例对本文方法进行验证，所得的方案用多种准则进行评价的结果都较好，既能保证测量振型向量的正交性，又能保证扩阶振型的准确性，并且有较强的抗噪声能力，是一种适合桥梁结构的较为理想的传感器优化布置方法。2．在讨论常用于结构损伤识别的几个指标的基础上，提出了基于灰关联曲率绝对差的损伤识别指标。该指标对结构损伤具有很高的灵敏度，而且仅需低阶模态信息即可获得很好的识别效果。同时，在灰关联曲率绝对差指标的基础上，本文采用分步识别法和多样本平均处理法，用来解决测试模态不完整和测试噪声问题。算例分析表明，本文提出的损伤指标能较好的识别出结构损伤，并具有一定的抗噪声能力。3．在介绍反问题和模型修正方法的基础上，利用坐标模态保证准则和频率构建优化目标函数，提出了基于克隆选择算法的损伤识别方法，克服了传统优化算法在计算中碰到的由于梯度矩阵病态或奇异而使计算无法完成的现象，提高了计算效率。对简支钢桁架梁桥中设定的损伤进行了识别研究，结果表明本文提出的方法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，适合于实际工程应用。4．提出了基于集成神经网络的结构多级损伤识别方法，采用分阶段改进参数作为网络的输入参数，分步实现对损伤的报警、定位和程度识别，克服了单个神经网络功能的单一局限性，增强了网络的泛化能力、降低了网络规模、减少了网络的训练样本集和训练时间。数值模拟的结果表明，本文提出的基于集成神经网络的结构多级损伤识别方法是有效的，而且具有一定的抗噪声的能力。5．在金水沟特大桥竣工动载试验的基础上，确立了大桥的健康基本状态。依托铁道部科学研究院金水沟特大桥健康监测课题，结合金水沟特大桥自身结构特点，对其健康监测实施方案进行了设计。采用本文提出的三种损伤识别方法对金水沟特大桥主梁关键位置所设定的损伤进行无噪声情况下的识别研究，比较了三种方法的优缺点。6．在介绍传统桥梁综合评估方法的基础上，将可拓理论应用于实际桥梁工程之中，提出了基于组合权重的区间可拓评价理论，建立了桥梁状态等级评定模型。工程实例表明，本文所提出的评价模型能够对桥梁状态进行客观、可靠的评价。更多还原

【Abstract】 Bridge is one of the important compositions of the transportation system, and the healthy situation of the bridge structure is the basis for the well operating of the transportation system. During the service lives of the bridge structure, affected by the loads, environmental collisions and natural disaster, the damages of the bridge are unavoidable. Once the damages amount to some extent, the structure of the bridge will be destroyed, which will cause damages to the state economics and the people’s lives and properties. Therefore, it is of theoretical significance and practical value to make researches on the health monitoring of the bridge structure.Structural damage identification method is the basis and important part of the health monitoring of the bridge structure. In recent years, researchers tried a lot to find the damage identification method for the bridge structure. Structural damage identification method based on the dynamic test is the hot issue at home and abroad. Although the research of this method has made great progress and many damage identification methods and theories have been proposed, limited by the factors such as our understanding, the testing technique and the field condition, the structural damage identification method based on the dynamic test has not made any substantial progresses, and could not be in widely applied. Thus, to make research on the structural damage identification method which is proper in practice with accurate results is of great practical value. The thesis is based on the past studies and makes researches on the structural damage identification method aimed at the existed problems at present, and the contents are shown below:1 .Based on the comparison analysis of the existing sensor placement optimization, the improved sensor placement optimization method based on the energy coefficient-effective independence method was proposed, which compensated the weaknesses of the traditional methods. Reflecting the coefficient of the node freedom modal strain energy, the proposed method can modify the effective independence reflecting the maximum linear independence, and further optimize the sensor placement designs. Taking the simply supported steel truss beam bridge as a numerical example, the results showed that the evaluations of the designs were good, and the method can ensure the orthogonality of the measurement vibration pattern vector, and can also ensure the accuracy of the expanded order vibration and has better anti-noise ability, which is an ideal method appropriate for the bridge structure.2.Based on the traditional modal parameter structural damage identification, a damage identification index based on the grey relation curvature absolute difference was proposed. This index had high sensitivity to the structural damage and had good effects only in the need of the low-order modal information. Meanwhile, on the basis of the grey relation curvature absolute difference, the thesis suggested hierarchical identification method and multi-sample average dealing method to solve the incomplete of the measurement modal and the data noise. The numerical example illustrated that the proposed damage index could well reflect the structural damage and had better anti-noise ability.3.On the basis of the reverse problem and model modification method, the coordinate modal assurance criterion and the frequency were used to establish the optimization object function, and the damage identification method based on clonal selection algorithm was proposed, which overcomes the difficulty that during the traditional algorithm when the gradient matrix was morbid or singular, and the calculation would not complete. The calculation efficiency was improved by the proposed method. The identification damage in the simply supported steel truss beam bridge was also studied and the result showed that the proposed method could identify the location and the extent of the damage, had better ability of anti-noise, and was especially proper for the engineering practices.4.Multi-grade damage identification method based on the integrated neural networks was proposed, and the alerting, locating and extent identification were realized by step, which strengthened the generalization ability of the networks, made the networks scale smaller and reduced the training sample size and training time of the networks. The numerical simulation demonstrates that the proposed method is effective and has the anti-noise ability.5.On the basis of the completion and dynamic load test of Jin Shui Gou Bridge, the health condition was established. Relying on the subject of Jin Shui Gou Bridge health monitoring of China Academy of Railway Science, and according to the characteristics of the bridge, the health monitoring implementation design of rigid frame-continuous beam structure was made. Three methods which were proposed in this paper were compared through the research on the damage identification of main beam of Jin Shui Gou Bridge.6.0n the basis of the introduction of the traditional bridge comprehensive evaluation methods, the extension theory was applied in the engineering. The interval extension evaluation theory based on the combined weight was proposed and the bridge condition grade evaluation model was also established. The engineering case shows that the proposed model can make objective and reliable evaluations.更多还原