基于SVM模式识别方法的桥梁频域损伤识别
发布时间：2018-08-13 18:18
【摘要】：桥梁在营运过程中，由于外部环境和使用条件中的不可预知性，结构的可靠性并不能完全保证。材料老化、自然条件恶劣及超载严重，已经严重影响到桥梁的安全性和使用寿命。桥梁在安装健康监测系统后，可以评估结构的完整性、耐久性和可靠性在桥梁的正常使用期内，方便制定最优的维修养护计划，确保结构运行安全。作为健康监测系统的核心，桥梁损伤识别已经成为近年来桥梁工程界的研究热点。 本文简述了桥梁健康监测和损伤识别等基本概念，归纳了目前用于桥梁损伤识别的各种方法，简单分析了损伤识别研究的难点。首先将支持向量机方法应用于较简单的桁架结构损伤识别中，，最后将支持向量机方法应用于大跨度斜拉桥的损伤识别中，取得了较好的识别效果。具体工作有以下几个方面： 1.介绍了结构频域损伤识别的基本特点，以桁架结构为例选择了较为合理的损伤指标。 2.以某大跨度斜拉桥为研究对象，通过传感器优化布置，对斜拉桥损伤前后进行数值模拟和计算分析，构建样本集。应用支持向量机的分类和回归原理对损伤位置和损伤程度进行了识别，对不同情况下的损伤识别效果进行了对比，考虑了不同噪声水平对于损伤识别效果的影响。 3.最后，针对本文的成果进行了总结，提出了在以后损伤识别中应该改进的地方。
[Abstract]:During the operation of the bridge, the reliability of the structure can not be guaranteed completely due to the unpredictability of the external environment and the operating conditions. The aging of materials, bad natural conditions and serious overload have seriously affected the safety and service life of bridges. After the health monitoring system is installed, the integrity, durability and reliability of the bridge can be evaluated during the normal service life of the bridge. It is convenient to make the best maintenance and maintenance plan and to ensure the safety of the structure. As the core of health monitoring system, bridge damage identification has become a research hotspot in the field of bridge engineering in recent years. In this paper, the basic concepts of bridge health monitoring and damage identification are briefly introduced, and various methods used in bridge damage identification are summarized, and the difficulties in the research of damage identification are simply analyzed. The support vector machine (SVM) method is applied to the damage identification of truss structures. Finally, the SVM method is applied to the damage identification of long-span cable-stayed bridges. The concrete work has the following several aspects: 1. This paper introduces the basic characteristics of structural damage identification in frequency domain, and takes truss structure as an example to select a more reasonable damage index. 2. Taking a long-span cable-stayed bridge as an object of study, a sample set was constructed by numerical simulation and calculation before and after the damage of the cable-stayed bridge. The classification and regression principle of support vector machine are applied to identify the location and degree of damage. The effects of different noise levels on damage identification are compared. 3. Finally, the results of this paper are summarized, and some improvements in damage identification are put forward.
【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U446
本文编号：2181797
[Abstract]:During the operation of the bridge, the reliability of the structure can not be guaranteed completely due to the unpredictability of the external environment and the operating conditions. The aging of materials, bad natural conditions and serious overload have seriously affected the safety and service life of bridges. After the health monitoring system is installed, the integrity, durability and reliability of the bridge can be evaluated during the normal service life of the bridge. It is convenient to make the best maintenance and maintenance plan and to ensure the safety of the structure. As the core of health monitoring system, bridge damage identification has become a research hotspot in the field of bridge engineering in recent years. In this paper, the basic concepts of bridge health monitoring and damage identification are briefly introduced, and various methods used in bridge damage identification are summarized, and the difficulties in the research of damage identification are simply analyzed. The support vector machine (SVM) method is applied to the damage identification of truss structures. Finally, the SVM method is applied to the damage identification of long-span cable-stayed bridges. The concrete work has the following several aspects: 1. This paper introduces the basic characteristics of structural damage identification in frequency domain, and takes truss structure as an example to select a more reasonable damage index. 2. Taking a long-span cable-stayed bridge as an object of study, a sample set was constructed by numerical simulation and calculation before and after the damage of the cable-stayed bridge. The classification and regression principle of support vector machine are applied to identify the location and degree of damage. The effects of different noise levels on damage identification are compared. 3. Finally, the results of this paper are summarized, and some improvements in damage identification are put forward.
【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U446
【参考文献】
相关期刊论文 前10条
1 王文静,刘志明,缪龙秀;基于实验模态的结构应变模态分析[J];北方交通大学学报;2000年04期
2 陈理渊;黄进;;基于支持向量机的电机转子断条故障诊断[J];电工技术学报;2006年08期
3 邱洪兴,蒋永生;古塔结构损伤的系统识别Ⅰ:理论[J];东南大学学报(自然科学版);2001年02期
4 宗周红,WANG T L,HUANG D Z,郑振飞;桥梁健康监测应用与研究现状(英文)[J];福州大学学报(自然科学版);2002年02期
5 谢峻,韩大建;一种改进的基于频率测量的结构损伤识别方法[J];工程力学;2004年01期
6 孙宗光,高赞明,倪一清;基于神经网络的桥梁损伤位置识别[J];工程力学;2004年01期
7 刘龙;孟光;;基于曲率模态和支持向量机的结构损伤位置两步识别方法[J];工程力学;2006年S1期
8 林立香,滕海文,霍达,李秀芬;基于曲率模态比值的城市公路桥梁损伤识别[J];河南科学;2005年06期
9 朱宏平,千力;利用振动模态测量值和神经网络方法的结构损伤识别研究[J];计算力学学报;2005年02期
10 李杰,陈隽;未知输入条件下的结构物理参数识别研究[J];计算力学学报;1999年01期
本文编号：2181797
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2181797.html
