【作者】 王金国；

【导师】 张敏政；

【作者基本信息】 中国地震局工程力学研究所 ， 防灾减灾工程及防护工程， 2005， 博士

【摘要】 本文围绕长输油气管道悬索跨越工程的抗震能力和健康诊断技术,利用理论分析、模型实验和数值模拟手段进行了较为系统的研究。在总结震害经验和分析受力特点的基础上,概括归纳了该类结构的三种基本损伤模式。提出了基于模式识别的多级整体健康诊断方法,建立并验证了健康诊断的神经网络模型。此外,还建议了基于简单缆索受力监测的局部诊断方法,该方法可与整体诊断方法结合使用,判断结构系统的损伤模式、损伤位置和损伤程度。主要研究内容和成果如下: 1 设计制作了管道悬索跨越工程的1:8 缩尺模型。在总结震害经验和分析受力特点的基础上,概括归纳了该类结构的三种基本损伤模式(塔架损伤模式、吊索损伤模式和斜拉索损伤模式)。就系统的完好状态和三种损伤模式分别进行了实验研究。进行了模态实验、敲击激振实验、白噪声和地震波输入的振动台实验,以及缆索拉力和振动频率的测试。实验结果为评估结构体系的抗震能力和建立系统的健康诊断方法提供了依据。2 利用ANSYS 程序建立了原型结构和模型结构的有限元分析模型,考虑缆索的预应力和大变形、进行了完好状态和不同损伤模式下的静力分析和模态分析,也尝试进行了地震反应分析。就索力和体系自振频率而言,模型结构的实验结果和数值模拟结果基本吻合。数值分析也表明,原型结构和模型结构具有相似性,模型实验结果可作为分析实际工程的依据。3 模型实验和数值模拟表明,管道悬索跨越工程具有较强的抗震承载能力和系统的稳定性。在三向地震动同时作用下,当输入地震动加速度达到0.5g(相当于地震烈度ⅸ度)时,系统构件仍未达到强度极限、未发生破坏;即使在人为造成的三种损伤模式下,体系仍保持稳定,可维持运行功能。4 对实验得到的结构体系自振频率参数进行了统计特征分析、区别不同损伤模式利用符号检验法进行了实测数据的显著性检验,确定前两阶频率参数可作为健康诊断的指标。建立了基于频率参数(频率和频率比)的多级整体健康诊断方法,可逐次识别结构的三种损伤模式。5 运用MATLAB6.5 工具箱,基于整体模式识别方法,建立了用于管道悬索跨越工程健康诊断的BP 人工神经网络。利用模型实验所得到的部分模态参数进行BP 网络训练,再用其余部分数据对网络功能进行验证。结果表明,该网络模型对三种损伤模式具有良好的识别能力。6 讨论了对缆索应力进行局部监测的技术方法。实验和分析表明,简单拉索的索力和振动频率间存在简单的线性关系,可利用敲击激振方法实测估更多还原

【Abstract】 The aseismic capacity and health diagnosis technology for cable-suspended structures of oil and gas pipelines are studied systematically by the theory analyses and the model experiments and numerical simulation analyses. Three kinds of basic damage modes of structures are briefly induced based on the experience from earthquake damage and the characteristics of structure’s mechanics. The multilevel method of whole health diagnosis is put forward according to mode discrimination. A neural net model is set up and tested by experiment datum. Furthermore, the local health diagnosis methods are proposed through monitoring the cable forces. The damage modes and the damage positions and the damage degrees of structure system can be judged by using the whole health diagnosis methods in combination with the local health diagnosis methods. The main content and outcome of this paper are as following: Firstly, the model scaled to be one eighth of the actual cable-suspended structure of pipelines is designed and made. Three kinds of basic damage modes of the structure (the damage of tower frame, the breaking of a sling and the inclined cable loses its function) are briefly induced based on the experience from earthquake damage and the characteristics of structure’s mechanics. The experiment studies on non-damage mode and three kinds of damage modes are completed respectively. The experiments include modal tests, vibration tests by knocking, shaking table tests by inputting white-noises and seismic waves, and tests of cable forces and vibration frequencies. The experiment results supply a demand for evaluating the aseismic capacity of structural system and putting forward health diagnosis methods of the structural system. Secondly, two finite models are built by ANSYS according to the actual structure and the model structure respectively. Under non-damage mode and three kinds of damage modes the static analyses and modal analyses are made taking cable prestress and large deformation into consideration. The earthquake response analyses are tried to be done. The experiment results basically tally with the values of numerical simulation analyses in cable forces and free frequencies of structural system. Numerical analyses also show that the actual structure is in the similitude of the model structure, and the result of model experiments provides the basis for analyzing actual engineering. Thirdly, the model experiments and numerical analyses all show that the cable-suspended structures of pipelines have a good aseismic capability and systematic stability. While inputting three dimension ground motion acceleration reaches 0.5g (is equivalent to China earthquake intensity ⅸ) the element strengths in the structure are still under the limit of strength, and no element damages. Even though the structure is under three man-made damage modes, the structure system keeps stability and maintains normal function. Fourthly, the first and the second frequencies are determined as index of health diagnoses through analyzing statistical characters of structure systematic free-frequencies got by experiments and checking prominence of the experiment datum by symbol test method in different damage modes. A multi-level whole health diagnosis method is set up based on frequency parameters (frequencies and ratio of frequencies), and can be used to recognize three damage modes of the structure. Fifthly, a BP artificial neural net used for health diagnoses of cable-suspended structures of pipelines is established by the tool box in the MATLAB6.5. The net is trained by parts of modal parameters got by the experiments, and is verified by the rest of modal parameters. The results show that this net model has strong ability to distinguish three kinds of damage modes. Sixthly, the local monitoring technology methods of testing cable stresses are discussed. The experiments and analyses show that the cable forces are linear with the frequencies. The cable forces can be estimated by knock-excited methods. The structural health diagnoses can be made according to the cable forces, because the inclined cables have key function in keeping the system stability and the inclined cable forces vary in different damaged modes. Through combining the local monitoring methods with the whole health diagnosis methods, the damaged modes can be distinguished, and the degrees can be estimated, and the location can be determined.更多还原