基于时域响应统计特征的桥梁结构损伤识别方法研究

发布时间：2018-04-26 21:32

本文选题：桥梁结构 + 损伤识别　；参考：《吉林大学》2014年硕士论文

【摘要】：随着现代经济的高速发展，桥梁所承受的交通荷载日益增加，外部恶劣环境导致材料不断老化，桥梁结构容易出现累积损伤，进而导致承载能力不足。综合考虑安全和经济因素，十分有必要开展对现役桥梁结构的损伤识别和健康监测，，了解桥梁的健康状况，保障结构的安全运营，减少垮塌事故的发生。基于动力响应的损伤识别技术因为其不影响结构正常工作，可实现长期或在线监测，操作简单方便等特点，而得到了广泛的关注和研究。但大多是借助模态分析理论，得到系统的模态参数，然后进行损伤识别。模态分析的方法具有良好的理论基础，并且在相关的研究中取得了不错的损伤识别效果，但仍需解决诸如功率泄露等一系列的问题。本文结合时间序列分析等统计学的办法，直接利用结构时域响应（加速度信号）的统计特征进行损伤识别。相比于模态分析，此类方法更加便捷，同时避免了有效信息的丢失。本文开展的研究工作如下： 1.时间序列模型的自回归系数描述的是结构某时刻的响应受之前时刻响应的影响程度，反映的是结构的固有特性。进一步的，我们可建立起自回归系数与结构物理特征量的关系式。因此，结构的损伤可以通过自回归系数改变体现出来。基于时间序列模型的自回归系数，本文建立了加权距离指标（），数值算例表明该指标可以识别损伤的出现，但当损伤程度较小时识别效果不够理想，并且无法损伤定位。 2.随机信号的自相关系数反映的是信号前后时刻之间的相关程度，它对系统的描述性跟时间序列模型的自回归系数是类似的。本文利用响应信号的一阶自相关系数（）进行了损伤识别，相比于基于自回归系数的加权距离，该指标对小损伤的识别更加敏感，但无法进行损伤定位。 3.我们将结构上某点在一段时间内测得的信号值的平方和记作，损伤位置的在损伤前后的变化程度（）要大于其他位置。基于此原理，我们对前文的数值算例进行损伤定位，结果表明，该指标对损伤位置具有良好的识别效果，并且能够定性反映损伤程度，但是该指标不能用来在线监测损伤的产生。考虑到以上三种指标损伤的识别效果，本文采用两步法进行损伤识别：先利用或者识别损伤的出现，在此基础上利用C指标进行损伤定位。
[Abstract]:With the rapid development of modern economy, the traffic load on the bridge is increasing day by day, and the external environment leads to the aging of materials, and the bridge structure is prone to cumulative damage, which leads to the insufficient bearing capacity. Considering the safety and economic factors, it is necessary to carry out damage identification and health monitoring of existing bridge structures, to understand the health status of bridges, to ensure the safe operation of the structures and to reduce the occurrence of collapsing accidents. The damage identification technology based on dynamic response has been widely studied because it can realize long-term or on-line monitoring and easy operation because it does not affect the normal work of the structure. However, the modal parameters of the system are obtained by modal analysis theory, and then damage identification is carried out. The modal analysis method has a good theoretical foundation, and has achieved good damage identification effect in related research, but a series of problems such as power leakage still need to be solved. Combined with time series analysis and other statistical methods, this paper directly utilizes the statistical characteristics of the time-domain response (acceleration signal) of the structure to identify the damage. Compared with modal analysis, this method is more convenient and avoids the loss of effective information. The research work carried out in this paper is as follows: 1. The autoregressive coefficients of the time series model describe the degree to which the response of the structure at a certain time is affected by the response of the previous moment, and reflect the inherent characteristics of the structure. Furthermore, we can establish the relation between the autoregressive coefficient and the structural physical characteristic quantity. Therefore, the damage of the structure can be reflected by the change of autoregressive coefficient. Based on the autoregressive coefficient of time series model, a weighted distance index is established in this paper. The numerical example shows that the index can identify the occurrence of damage, but when the degree of damage is small, the identification effect is not satisfactory, and the damage location can not be achieved. 2. The autocorrelation coefficient of the random signal reflects the correlation between the time before and after the signal, and its descriptive property of the system is similar to that of the autoregressive coefficient of the time series model. In this paper, the first order autocorrelation coefficient of the response signal is used to identify the damage. Compared with the weighted distance based on the autoregressive coefficient, the index is more sensitive to the identification of small damage, but it can not locate the damage. 3. We take the square sum of the signal values measured internally at a point in the structure over a period of time as the damage location. The degree of change before and after injury was greater than that in other locations. Based on this principle, the damage location of the numerical examples in the previous paper is carried out. The results show that the index has a good effect on identifying the damage location, and it can reflect the damage degree qualitatively. But this indicator cannot be used to monitor damage online. Considering the damage recognition effect of the above three indexes, this paper adopts the two-step method to identify the damage. Or identify the appearance of damage, on the basis of C index damage location.
【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】：U446;U441

【参考文献】