相控阵结构健康监测的数据压缩采样与损伤识别研究

发布时间：2018-04-26 12:09

本文选题：结构健康监测 + 相控阵　；参考：《南京信息工程大学》2015年硕士论文

【摘要】：智能结构健康监测利用传感器技术及先进的信号处理方法,在结构损伤发生初期实时、在线监测结构健康状态,及时准确的识别结构中损伤。该方法对减少灾难性事故,保证工程安全具有重大意义。超声相控阵技术通过调整阵列中各阵元发射和接收信号的时间延迟,实现波束的偏转、聚焦等效果。与传统超声检测技术相比,超声相控阵技术可在不移动探头的情况下,完成全方位、多角度的扫描,提高检测精度。本文论述相控阵结构健康监测的相关理论,说明相控阵检测技术的优越性。具体采用8阵元的超声相控阵阵列检测带损伤的铝板材料,通过对压电阵元发射与接收过程时间延迟的控制,完成多方向的声束扫描,结合多种信号处理方式处理采集到的相控阵信号,获得损伤位置的信息,在图像中予以显示。针对信号采集过程中,不可避免的会掺杂工频干扰信号,结合巴特沃斯滤波方式、切比雪夫滤波方式以及椭圆滤波方式,对信号做滤波预处理,在Matlab信号处理工具箱中调用数字滤波器的选择函数,对滤波效果进行仿真,获得更加准确的信号。在损伤识别过程中,本文设计高效的相控阵检测系统采集信号,根据公式推导及算法分析,获得发射、接收过程中阵元的时间延迟,对合成信号的幅值做归一化处理,将相应角度上损伤信号的幅值用图像的灰度级来表示,图像最亮处即代表损伤。采用超声相控阵技术采集得到的信号也是含有大量噪声的信号,原始数据得到的损伤图像并不完美,结合图像增强方法,对其进行处理,提高图像的可识别度。传统图像增强的过程中存在对比度减小和噪声放大的缺陷,导致图像细节信息丢失,本文将模糊集的方法应用其中可以克服该缺陷,得到效果更好的增强后的图像。传统采样过程一般遵循奈奎斯特采样定理,超声相控阵技术采集的信号量大,给信号的存储和传输带来困难。本文采用压缩感知技术对超声信号进行处理,大大降低存储和传输的信息量。主要工作涵盖：(1)压缩感知是基于稀疏信号的处理方法,利用稀疏变换基,将非稀疏的超声信号转换为稀疏信号；(2)设计观测矩阵,获得稀疏向量基；(3)利用多种压缩感知信号重构算法,重构原始信号,经重构误差对比,选取最佳的重构算法完成实验。最后利用压缩感知处理后的信号,判定损伤位置。
[Abstract]:Smart structure health monitoring uses sensor technology and advanced signal processing methods to monitor structural health in real time at the initial stage of structural damage and identify structural damage in time and accurately. This method is important to reduce disaster accidents and ensure engineering safety. Ultrasonic phased array technology adjusts the array array by adjusting the array array. Compared with the traditional ultrasonic detection technology, the ultrasonic phased array technology can complete the omni-directional, multi angle scan and improve the detection precision. This paper discusses the related theory of phased array structure health monitoring, and explains the phased array detection. The ultrasonic phased array of 8 array elements is used to detect the damaged aluminum plate material. By controlling the time delay of the transmission and receiving process of the piezoelectric element, the multi direction acoustic beam scanning is completed and the signal number of the phased array is processed by a variety of signal processing methods, and the information of the damage position is obtained, and the image is given in the image. In the process of signal acquisition, it is unavoidable to adulterate the power frequency interference signal, combined with Butterworth filtering, Chebyshev filtering and ellipse filtering, the signal is preprocessed by filtering, and the selection function of digital filter is called in the Matlab signal processing toolbox, and the filtering effect is simulated, and more accurate is obtained. In the process of damage identification, this paper designs an efficient phased array detection system to collect signals. According to the formula deduction and algorithm analysis, the time delay of the array element in the process of transmitting and receiving is obtained. The amplitude of the synthetic signal is normalized, and the amplitude of the damaged signal in the corresponding angle is expressed with the gray level of the image, and the image is the most. The signal obtained by the ultrasonic phased array technology is also a signal containing a lot of noise. The damage image obtained by the original data is not perfect, and the image enhancement method is used to process it to improve the recognition of the image. The defects of contrast and noise amplification exist in the traditional image enhancement process. In this paper, the method of fuzzy set can be lost. In this paper, the method of fuzzy set can be applied to overcome the defect and get better image. The traditional sampling process usually follows the Nyquist sampling theorem, the ultrasonic phased array technology has a large signal collection, which brings difficulties to the storage and transmission of the signal. This paper uses compressed sensing technique. The operation deals with ultrasonic signals to greatly reduce the amount of information stored and transmitted. The main work covers: (1) compression perception is based on sparse signal processing, using sparse transform base, converting non sparse ultrasonic signals to sparse signals; (2) design observation matrix, obtain sparse vector base; (3) use a variety of compressed sensing signal weight. The original signal is reconstructed, the reconstruction error is compared, and the best reconstruction algorithm is selected to complete the experiment. Finally, the signal of compressed sensing processing is used to determine the damage position.

【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN911.7

【参考文献】