移动荷载作用下梁型结构健康诊断方法研究

发布时间：2018-06-14 19:19

本文选题：梁型结构 + 损伤识别　；参考：《浙江大学》2014年博士论文

【摘要】：梁型结构广泛应用于国民经济的各个方面,尤其是桥梁。桥梁作为交通运输的重要组成部分,是一个国家基础设施建设的重点,同时也是经济发展与技术进步的象征。近些年来,梁型结构的健康诊断技术已经成为工程界的研究热点,健康诊断系统及其理论研究也取得了很大进展。但是由于健康诊断系统本身的多学科交叉性、大型梁型结构及其环境的复杂性和不确定性,使得梁型结构的健康监测系统的许多关键技术从理论到实际应用还存在许多不足。本文采用理论与试验相结合的方法,针对梁型结构健康监测过程中的一些关键技术问题进行了研究。在基于结构振动特性的损伤识别技术的基础上,采用了信息理论、信号处理、时频分析、统计分析等领域内的先进方法,对梁型结构损伤识别的信号处理、特征参数识别等方面进行了探索。论文的主要工作内容体现在以下几个方面： (1)针对交通荷载的移动特性,通过研究和实践提出了用移动荷载作用于梁型结构上,对其产生的振动响应数据进行分析。当移动荷载作用于梁型结构上时,损伤引起的振动响应数据特征参数的变化将被放大,可以提高损伤检测特征参数的提取精度。 (2)桥梁结构健康监测过程中获得的观测数据具有非线性、非平稳等复杂性,样本熵可以有效地表征信号的复杂性,估计信号的非线性程度,本文提出了使用样本熵来提取结构损伤信息,并且使用经验模式分解及神经网络对该方法进行了进一步的改进。 (3)当移动荷载作用于有损伤的结构上时,产生的振动信号是非平稳信号,桥梁损伤结构振动响应的统计量将随着时间和载荷而变化。本文提出了梁型结构损伤分析的时频分析方法,并对时频分析中的交叉项问题进行了讨论,分析了抑制交叉项干扰的方法,并结合信息熵、神经网络进行结构损伤识别。 (4)信号的高阶统计量具有良好的非高斯、非平稳信号处理能力,本文提出了损伤结构振动信号的高阶谱分析方法。由于高阶谱分析结果为二维甚至更高维的,包含的信息量大,因此本文提出了双谱的有效值熵分析方法,结合神经网络的模式识别能力进行结构的损伤识别。 (5)为充分挖掘高阶谱分析结果中所包含的结构损伤信息,需要配合有效的降维分析方法。本文提出了一种改进的有监督保局投影数据降维方法,通过该方法对高阶谱分析结果进行特征向量的提取,并将损伤识别过程分为两个模块进行综合信息融合,运用神经网络进一步进行损伤识别。
[Abstract]:Beam structure is widely used in all aspects of national economy, especially bridges. As an important part of transportation, bridge is the focus of a country's infrastructure construction, and also a symbol of economic development and technological progress. In recent years, the health diagnosis technology of beam structure has become a hot topic in engineering field, and great progress has been made in the research of health diagnosis system and its theory. However, because of the interdisciplinary nature of the health diagnosis system and the complexity and uncertainty of the large beam structure and its environment, many key technologies of the health monitoring system for the beam structure still have many shortcomings from theory to practice. In this paper, some key technical problems in the process of beam structure health monitoring are studied by combining theory with experiment. On the basis of damage identification technology based on structural vibration characteristics, advanced methods in the fields of information theory, signal processing, time-frequency analysis and statistical analysis are adopted to process the damage of beam structure. The characteristic parameter identification and other aspects are explored. The main contents of this paper are as follows: (1) aiming at the moving characteristics of traffic load, the vibration response data generated by moving load acting on beam structure are analyzed through research and practice. When the moving load acts on the beam structure, the change of the characteristic parameters of the vibration response data caused by the damage will be amplified. It can improve the accuracy of extracting characteristic parameters of damage detection. The observed data obtained in the process of bridge structure health monitoring are nonlinear, non-stationary and so on, and the sample entropy can effectively characterize the complexity of the signal. In order to estimate the nonlinearity of the signal, the sample entropy is used to extract the structural damage information. The method is further improved by empirical mode decomposition and neural network. When moving load acts on the damaged structure, the vibration signal is non-stationary. The statistics of vibration response of damaged bridge structure will change with time and load. In this paper, a time-frequency analysis method for damage analysis of beam structures is presented. The crossover terms in time-frequency analysis are discussed, and the methods to suppress cross-term interference are analyzed, and the information entropy is combined. Neural network is used to identify structural damage. The high-order statistic of the signal has good ability of non-Gao Si and non-stationary signal processing. In this paper, a method of high-order spectrum analysis for the vibration signal of damaged structure is proposed. Since the results of high-order spectral analysis are two-dimensional or higher, and contain a large amount of information, a bispectral effective value entropy analysis method is proposed in this paper. In order to fully mine the structural damage information contained in the results of higher-order spectral analysis, it is necessary to cooperate with an effective dimensionality reduction analysis method for structural damage identification based on the pattern recognition ability of neural networks. In this paper, an improved dimensionality reduction method for supervised local projection data is proposed, by which the feature vectors are extracted from the results of high-order spectral analysis, and the damage identification process is divided into two modules for comprehensive information fusion. Further damage identification using neural network.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TU317

【参考文献】