【作者】 尹涛；

【导师】 朱宏平； 余岭；

【作者基本信息】 华中科技大学 ， 桥梁与隧道工程， 2007， 博士

【摘要】 近20年来,基于动力测量数据的结构损伤识别和健康诊断一直是国内外学者的研究热点,在土木工程领域中的研究和应用较多,但对于水工结构系统(特别是水工钢结构)损伤识别和健康诊断尚缺少研究。本文在水利部科技创新项目(No. SCX2003-18)与国家自然科学基金(No. 50608036)的资助下,对于水工钢结构的损伤识别方法进行了初步地探讨,为解决水工钢结构(如水工弧形闸门)这样一个大型复杂空间结构的运营安全评估及损伤识别问题提供了新的思路。具体的研究内容如下:首先回顾和总结了国内外损伤识别领域的研究进展,然后展开了关于损伤识别方法的理论及应用研究。针对基于传统遗传算法的损伤识别方法的不足,改进了传统遗传算法的变异算子,提出了一种基于改进遗传算法的损伤识别方法,一定程度上减少了损伤识别搜索过程中的盲目性并显著提高了识别结果的准确度。某两层门式框架模型损伤识别的仿真算例和一个带栓接接头矩形框架损伤识别的试验研究,证明了该方法的正确性和有效性。针对遗传算法收敛缓慢的固有缺点,提出了一种基于信赖域算法的损伤识别方法,通过求解带边界约束的非线性最小二乘目标函数来实现损伤识别,使得识别过程既具有更强的鲁棒性又在很大程度上节省了计算时间;同时研究了目标函数向量的扩充,使该方法能适用于更加实际的情况。两层框架结构的数值仿真研究与两个实验室框架模型的试验研究结果表明,在较少的迭代次数下,该方法就能成功地进行损伤识别,证明了该方法应用于实际结构的可行性和有效性。接下来,更进一步地将该方法应用于典型水工钢结构——水工弧形闸门模型的损伤识别中。先利用健康状态下的实测模态数据,通过修正初始有限元模型来获得用作损伤识别的基准模型,再运用该方法进行损伤识别的数值仿真和实测数据研究。结果表明该方法能够较为准确地识别出弧形闸门模型的损伤位置及其程度,这说明了将该方法应用于水工闸门结构的损伤识别中是可行而有效的,具有较好地应用前景,也为解决水工钢结构的运营安全评估及损伤识别问题提供了新思路。最后,为研究测量噪声对损伤识别结果的影响,具体针对基于敏感性的损伤识别方法,提出了一种基于统计分析原理的噪声分析方法,能同时对频率噪声和振型噪声进行分析,这就使得随机噪声分析过程更加简洁而又有效;又基于敏感性分析原理提出了一种结构损伤识别的改进算法。某单层门式框架损伤识别的数值仿真研究结果不仅表明了新的噪声分析方法与改进的灵敏度方法的正确性和有效性,同时也证明了改进方法确实比传统方法具有更强的噪声鲁棒性。更多还原

【Abstract】 During the last two decades, there have been great interests in the structural damage identification and health diagnosis using vibration measurements, and the majority of studies can been found in civil engineering. However, studies about the damage identification and health diagnosis of the hydraulic structures are still few, especially for the hydraulic steel structures. Supported jointly by the Innovative Funding of the Ministry of Water Resources of China (No.SCX2003-18) and the National Natural Science Foundation of China (No. 50608036), in this dissertation, the studies on damage identification of hydraulic steel structures are performed, in order to provide a feasible idea for the healthy state assessment and damage identification of the hydraulic steel structures (such as the hydraulic radial gate), a large and complex spatial structures. The content is as follows:Firstly, the methods of structural damage identification developed in the recent two decades are reviewed and summarized. Secondly, theoretical developments and application studies about the damage identification methods are performed in the following parts of this dissertation. As for the shortcomings existing in the traditional genetic algorithm based damage identification methods, a modification has been made to the mutation operator in the traditional genetic algorithm, and an improved genetic algorithm based damage identification method is proposed, which can minimize the blindness of solution to some extent and improve the accuracy of identification results obviously. The applicability of a two-storeyed frame structure and a laboratory frame structure with bolted joints all demonstrates the feasibility of this method in structural damage identification. However, it’s the main disadvantage to hamper its real application that the convergence speed of the genetic algorithm is inherently slow. Thus a new structural damage identification approach is proposed based on the trust-region method, which obtains the identification results by minimizing a bound constrained nonlinear least-squared problem, and can make the identification process more robust and timesaving. A followed expansion process of the original objective function is also investigated, which is very important for the real application of the proposed damage identification method. The results of numerical studies of the two-storeyed frame structure and experimental studies of the two laboratory-tested frame structures all demonstrate the feasibility and efficiency of the proposed method. A further application study of this method is on a typical hydraulic steel structure, i.e., a hydraulic radial gate. After model updating to the initial finite element model (FEM) with the experimental vibration data under the healthy state, a refined FEM is obtained and used to identify structural damages using the tested modal data under damaged states. The assessment results show that damages can be identified using both the numerical and experimental modal data, which indicate that application of the method in the damage identification of the hydraulic steel structures is feasible and efficiency, and it also provided a new idea of the healthy state assessment and damage identification to the hydraulic steel structures. Finally, the influence of experimental noise on the identification results is also investigated for the sensitivity-based damage identification methods. A novel statistic-based random noise analysis method is proposed, which does not only make the analysis process more efficient but also can analyze the influence of noise on both frequencies and mode shapes for three commonly used sensitivity-based damage detection methods in a similar way. In addition, an improved sensitivity-based damage identification method is also proposed. A one-story portal frame is adopted to evaluate the validity and efficiency of both the proposed noise analysis technique and the improved damage identification method, and the assessment results also show that the improved method is more robust to experimental noise than its normal one.更多还原