【作者】 严刚；

【导师】 周丽；

【作者基本信息】 南京航空航天大学 ， 工程力学， 2009， 博士

【摘要】 近年来在智能材料与结构领域取得的新进展,使得利用集成在结构中的先进驱动/传感元件网络,在线获取与结构状态相关的信息,识别结构的安全状况成为可能。应用结构健康监测系统及时获取有关结构损伤的性质、程度、分布和演变的信息,对及时做出决策来阻止结构性能的退化和失效,增加飞行安全,降低维护费用具有至关重要的意义。本文对航空结构健康监测中的若干技术方法进行研究,采用被动和主动相结合的方法,对结构进行在线的连续监测,近乎实时地监测到冲击载荷的发生和作用位置并重建冲击载荷,主动定量地识别出结构中的损伤位置和程度,并考虑实际运行环境对损伤监测和识别的影响,确保结构的安全使用。本文的主要研究内容和取得的成果有:(1)对复合材料结构冲击载荷识别进行研究,提出一种基于智能优化算法的冲击载荷识别方法,同时识别冲击位置并重建冲击力时间历程。采用一种简单有效的冲击载荷参数化方法,将空间域的冲击位置识别和时间域的冲击力识别置于同一算法之下。在算法的运行过程中,冲击载荷识别被转换为一个优化问题,结合复合材料结构在冲击载荷作用下的响应模型,通过最小化理论模型计算结果与实际测量信息之间的差别,使用智能优化算法自适应地搜索出描述冲击位置和冲击力时间历程的参数。为提高运算效率,加速算法收敛性,本文采用微种群遗传算法来对复合材料结构进行冲击载荷识别。所提出的方法算法明确,过程简便,各个环节相互独立,通用性强,为复合材料结构的冲击载荷识别提供了一种可行的方法。进行了数值仿真研究来验证所提出方法的有效性。(2)对应用偏移技术进行损伤成像识别进行研究,提出一种频率-波数域偏移方法用于损伤的识别,将损伤的情况可视化并提高损伤识别的效率。首先基于Mindlin板理论,推导了板结构中弥散性弯曲波频率-波数域的偏移方法,分别对散射波场和入射波场在频率-波数域进行快速延拓,并采用时间一致性成像条件对损伤进行主动成像识别。然后从图像处理的角度提出一种新的方法用于识别板结构中同时发生的多部位损伤源的位置以及发生时刻,采用频率-波数域偏移技术对损伤源发出的弹性波信号进行处理,结合爆炸成像条件快速生成包含损伤源位置信息的图像。通过最小熵原理确定所识别损伤源位置和发生时刻最接近真实情况的最优图像,以实现损伤的被动成像识别。进行了数值仿真研究来验证所提出方法的有效性。(3)对环境变化下的复合材料结构损伤监测和识别进行研究,提出一种两步识别方法用于监测并识别环境变化下损伤发生与否以及损伤的位置。首先采用一种基于Lamb波和统计分析的方法,从统计的角度定义损伤指标以考虑环境变化对Lamb波信号的影响。对损伤指标进行统计偏值分析,并采用蒙特卡罗方法计算损伤指标阈值以确定某一激励-接收路径上是否存在损伤。一旦确定发生损伤,采用损伤存在概率成像算法融合多个激励-接收路径上的信息生成损伤图像实现损伤识别,分别使用损伤指标和其马氏平方距离值作为损伤特征用于获得损伤存在概率图像。建立了结构健康监测实验系统,对温度变化下的加筋复合材料壁板试件进行了实验研究以验证所提出方法的有效性。本文研究得到了国家自然科学基金(项目编号:10572058)、高等学校博士点基金(项目编号:20050287016)和南京航空航天大学博士学位论文创新与创优基金(项目编号:BCXJ07-03)的资助。更多还原

【Abstract】 Recent developments in the area of smart materials and structures have made it possible to obtain the state of structure and to identify the structural safety status online with advanced actuator/sensor network integrated in the structure. It is important and crucial for impeding the structural degradation, increasing the flight safety and reducing the maintenance cost by applying structural heath monitoring system to obtain information about the characteristic, extent, distribution and progress of the damages.This dissertation studies several methods and techniques in structural health monitoring of aircraft structures. Both passive and active methods are applied to monitor the structure continuously and online to ensure the structural safety. These techniques can detect the impact events in/near real time, identify the impact location and reconstruct the impact force history, actively and quantitatively identify the damage location and damage size, and consider the effect of actual operational environment on damage detection and identification. After a brief introduction of the background of this study in Chapter 1, the main contents of this dissertation are as follows:(1) In Chapter 2, the impact identification problem for composite structure is studied. A general approach for impact load identification of composite structure, which can identify the impact location and reconstruct the impact force history simultaneously, is proposed. In the study, a simple and effective parameterization method is employed and the impact load is represented by a set of parameters, thus the impact localization in space domain and the force identification in time domain can be solved within the same algorithm. During the operation process, the impact load identification problem is transformed to a parameter optimization problem. Combining with a forward impact response model, intelligent optimization algorithm is employed to adaptively identify the impact location and time history by minimizing the difference between the computed model outputs and measured responses. Particularly, a micro-genetic algorithms (micro-GA) is employed as the intelligent optimization algorithm to perform the identification task with its global search capability and high computational efficacy. This new approach is simple and clear, each of the steps is independent and can be incorporated with different methods, providing a general and flexible approach for complex composite structures. Numerical simulation studies are performed to validate the effectiveness of the proposed method.(2) In Chapter 3, the problem of damage imaging using migration technique is studied. A frequency-wavenumber (f-k) migration technique is proposed for visualizing and identifying damages in plate-like structures, and improving the efficiency of damage identification. Based on Mindlin plate theory, the fast f-k migration method for dispersive flexural waves in isotropic plates is developed. The scattered and incident wavefields are extrapolated in f-k domain using Mindlin plate theory, respectively. Combined with the time-coincidence imaging condition, damages are actively imaged and identified by performing cross-correlation to the extrapolated scattered wavefield and incident wavefield. Then a new method for identifying the locations and initiation time of multiple simultaneous occurred damages in plate-like structure is proposed from the view of image processing. The f-k migration with the exploding imaging condition is employed to produce image displaying the locations of the damages by back-propagating the elastic waves emitted by the damages to their sources. An optimal criterion using minimum Shannon entropy is adopted to determine the image with the locations and initiation time of the damages mostly approximate the actual ones. Numerical simulation studies are performed to validate the effectiveness of the proposed methods.(3) In Chapter 4, the problem of damage detection and identification for composite structure under environmental changes is studied. A two-stage method is proposed to detect the existence of damage and identify the damage location with environmental effects. First a statistical outlier analysis method using Lamb waves is employed to detect the damage existence. To consider the environmental effect on damage detection, damage index is defined as the damage-sensitive feature from a statistical point. Outlier analysis is employed to process the damage index, and the damage threshold value to distinguish whether the change of Lamb wave signal is induced by damage in an actuator-sensor path or is only affected by environmental change is obtained by Monte Carlo approach. After the damage is detected, a damage presence probability imaging algorithm is used to fuse information collected from multiple actuator-sensor paths to form a diagnostic image to identify the damage. Damage index and its Mahalanobis square distance (MSD) are adopted as the damage features to generate the damage presence probability images, respectively. Experimental studies on a stiffened composite panel with temperature change are performed to demonstrate the effectiveness of the proposed method.This study is supported by the National Natural Science Foundation of China (No.10572058), the Research Fund for the Doctoral Program of Higher Education (No. 20050287016) and the Ph.D Innovation Foundation of NUAA (No. BCXJ07-03).更多还原