桥梁结构损伤识别指标比选及损伤程度识别方法研究

发布时间：2018-03-14 23:23

本文选题：桥梁结构　切入点：损伤识别　出处：《吉林大学》2014年硕士论文　论文类型：学位论文

【摘要】：桥梁结构最重要的交通基础设施结构形式，在道路建设中广泛应用。桥梁灾害事故的发生导致巨大的人员伤亡及经济损失，造成了恶劣的社会影响。灾害事故的发生表明桥梁结构在不可见的内在损伤作用下，如果不及时进行检测、维修或加固，，容易导致垮塌事故的发生。因此，对桥梁结构进行及时科学的损伤检测和评估，及时了解结构损伤状况，对于保证结构安全运营，避免灾害事故发生意义深远。本文以广泛使用的简支梁桥和连续梁桥为研究对象，开展了基于结构动力特性损伤识别指标的优选工作；在此基础上，提出了相应的损伤程度识别方法。主要工作如下：（1）基于动力特性损伤识别理论基础，对比分析了频率、振型、模态曲率、模态柔度、均布荷载面曲率等指标在损伤位置识别方面的敏感性。识别结果表明，模态曲率差、模态柔度差曲率以及均匀荷载面曲率差等指标能够实现有效的损伤定位。（2）对模态柔度差曲率、模态曲率差、均匀荷载面曲率差的抗噪声能力进行分析，结果表明模态柔度差曲率具有较强的抗噪声干扰能力，可以作为结构损伤识别的优选指标，并将该指标应用于连续梁桥的损伤识别中。（3）以简支梁桥和连续梁桥为研究对象，选取模态柔度差指标作为遗传优化神经网络的输入参数，验证了方法在单位置及多位置损伤识别方面的有效性。结果表明：对于简支梁桥，遗传算法优化神经网络对于单位置损伤识别的最大相对误差为2.67%，对于多位置损伤识别的最大相对误差为6.83%；对于连续梁桥，遗传算法优化神经网络对于单位置损伤识别的最大相对误差为3.83%，对于多位置损伤识别的最大相对误差为8.14%。（4）以BP神经网络的对比分析表明，遗传优化神经网络对于简支梁桥单位置损伤识别的最大误差为2.67%，而BP神经网络的最大识别误差为4%，遗传优化神经网络的计算精度要优于BP网络。
[Abstract]:Bridge structure, the most important form of traffic infrastructure, is widely used in road construction. The occurrence of bridge disasters and accidents results in huge casualties and economic losses. The occurrence of the disaster accident indicates that the bridge structure is easy to collapse if it is not detected, repaired or strengthened in time under the action of invisible internal damage. Timely and scientific damage detection and evaluation of bridge structures and timely understanding of structural damage conditions are of far-reaching significance to ensure the safe operation of structures and to avoid disasters and accidents. In this paper, the widely used simply supported beam bridge and continuous beam bridge are taken as the research objects, and the optimization work based on the damage identification index of the dynamic characteristics of the structure is carried out, and on this basis, the corresponding damage degree identification method is proposed. The main work is as follows:. 1) based on the damage identification theory of dynamic characteristics, the sensitivity of frequency, mode shape, modal curvature, modal flexibility and uniform load surface curvature to damage location identification is compared and analyzed. The results show that the modal curvature is poor. The index of modal flexibility difference curvature and uniform load surface curvature difference can realize effective damage location. (2) the anti-noise ability of modal flexibility difference curvature, modal curvature difference and uniform load surface curvature difference is analyzed. The results show that modal flexibility difference curvature has strong anti-noise acoustical interference ability and can be used as an optimal index for structural damage identification. The index is applied to the damage identification of continuous girder bridge. Taking simply supported beam bridge and continuous beam bridge as the research object, the modal flexibility index is selected as the input parameter of genetic optimization neural network. The effectiveness of the method in single position and multi-position damage identification is verified. The results show that: for simply supported beam bridges, The maximum relative error of genetic algorithm optimization neural network is 2.67 for single position damage identification, 6.83 for multi-position damage recognition, and 6.83 for continuous beam bridge. The maximum relative error of genetic algorithm for single position damage identification is 3.83 and 8.14 for multi-position damage identification. 4) the comparative analysis of BP neural network shows that, The maximum error of genetic optimization neural network for single position damage identification of simply supported beam bridge is 2.67 and the maximum recognition error of BP neural network is 4. The accuracy of genetic optimization neural network is better than that of BP neural network.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U445.7

【参考文献】