结构健康监测的稀疏恢复算法与压缩采样实现

This project is to solve several key problems and technologies of Structural Health Monitoring (SHM) by introducing newly developed Compressive Sensing (CS) and Sparsity Recovery (SR) theories, and hence to develop new SHM method and technology based on CS.Majors researches planned in this project are in the following: 1.It is to develop distributed compressive sensing theory and technology for wireless sensor networks of SHM. Signal models for distributed compressive sensing will be established, and corresponding sparse signal recovery algorithms will be developed.Structural damage detection using compressed samples will be explored.2. Compressed sampling technology for Lamb waves is to be investigated. Efficient recovery algorithms for compressed high frequency signal samples are to be developed. New damage detection method based compressed sample of Lamb waves is to be researched. 3. New parameter (damage) identification and model updating methods based SR theory are to be investigated, and correspondingly, general theory for modal parameter identification is to be established. 4. Compressive sampling technology for dynamic signal and Lamb waves is to be investigated. A multi-channel compressive sampling hardware is to be developed, into which the developed algorithms are to be integrated and tested. The technologies developed in this project are expected to significantly reduce the energy consumption of wireless sensor network, and to overcome the constraints posed by limited current data acquisition technology on high-frequency signal sampling, and to establish a new category of methods for structural parameter (damage) identification and model updating with higher robustness and more efficiency. Researches of this project will greatly enrich and advance SHM technology, and hence are of significant scientific values and with great potential for applications.

结合结构健康监测（SHM）若干关键问题和技术，引入压缩感知（CS）和稀疏恢复（SR）理论，发展基于CS的SHM新方法和新技术。研究内容包括：1.发展SHM无线传感器网络的分布压缩感知理论和技术，建立分布压缩感知信号模型，开发相应的稀疏信号恢复算法，并建立基于压缩采样的结构损伤识别方法；2.研究导波信号压缩采样技术，发展高频信号压缩采样的信号恢复高效算法，发展基于导波压缩采样的损伤探测方法；3.研究基于SR理论的结构参数（损伤）识别和模型修正新方法，建立相应的模态参数识别一般理论；4.研究结构动态信号和导波信号压缩采样技术，开发多通道压缩采样硬件系统，集成并检验信号恢复算法。研究成果将大幅度降低无线传感器网络的能耗；突破现有数据采集技术对高频信号数据采集的制约；建立一类鲁棒性更好、更有效的结构参数(损伤)识别和模型修正新方法。本项目将极大地丰富和发展SHM技术，具有重要的科学价值和应用前景。

土木工程结构健康监测技术经过最近二十多年的发展，虽然在诸多长大桥梁结构上得到了应用，但期待该技术在桥梁结构的管养维护中充分发挥作用仍面临着诸多困难，其中包括海量数据的处理和挖掘，强噪声环境下的系统参数识别、损伤诊断和评定等。本项目将压缩感知理论及相关的稀疏恢复理论引入结构健康监测，发展监测压缩采样技术、数据分析和处理算法，试图解决结构健康监测所面临的问题。本项目在以下几个方面开展了深入研究并取得显著成果。.1.以无线传感网络为应用背景，开发了基于数字信号的结构振动压缩采样系统，并进行了实验室测试和试验验证。与传统数据压缩技术硬件相比，本项目开发的压缩采样硬件更加简洁和高效；2.研究了振动信号压缩采样的信号恢复算法，提出了一种基于Polar插值连续原子库的信号重构算法，对振动信号重构精度比传统的BP和OMP算法更高；研究了基于联合稀疏模型的结构振动信号恢复算法；3.研究了基于压缩振动信号的模态参数识别方法，提出了两种直接基于压缩采样的模态参数识别方法，该算法不需要重构信号，模态参数识别结果显著优于基于信号重构的识别结果，特别适合压缩感知无线传感网络；4.研究提出了基于稀疏信号恢复理论的结构损伤识别和模型修正算法，提出了基于稀疏恢复理论的传感器优化布置策略和模态定阶方法,求解结果具有更好的精确性和鲁棒性；5.在实验室条件下，对开发的振动信号压缩采样系统进行了测试，对一框架模型振动信号进行压缩采样，并应用本项目发展的数据分析和处理算法，成功且高效地提取了结构的振动模态参数。.本项目针对结构健康监测技术目前面临的关键科学问题和核心技术，在振动信号采集和信息处理等方面取得一定的突破，开发了高效的振动信号压缩采集技术，建立一类鲁棒性更好、更有效的结构模态参数识别和损伤识别新方法，为将压缩感知技术应用于结构健康监测的无线传感网络奠定了基础，有望显著降低无线传感器网络的能耗。这些成果丰富和发展结构健康监测技术，具有重要的科学价值和应用前景。