结合迭代阈值法的多任务贝叶斯压缩感知结构振动信号重构  

Structural vibration responses reconstruction based on multi-task Bayesian compressed sensing combined with iterative threshold method

作　　者：刘婷婷 张笑华[1] 方圣恩[1] LIU Tingting;ZHANG Xiaohua;FANG Sheng'en(College of Civil Engineering,Fuzhou University,Fuzhou,Fujian 350108,China)

机构地区：[1]福州大学土木工程学院,福建福州350108

出　　处：《福州大学学报（自然科学版）》2023年第2期184-190,共7页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省自然科学基金面上资助项目(2021J01598);福州市科技计划项目(2021-Y-084)。

摘　　要：为缓解结构健康实时监测中因为海量数据导致的数据采集、存储和传输成本高的问题,采用压缩感知理论结合迭代阈值法对数据进行压缩采样;然后用多任务贝叶斯压缩感知重构算法,通过少量采样数据恢复原始信号.利用吉安大桥的现场环境振动试验数据,验证结合迭代阈值法的多任务贝叶斯压缩感知重构算法的有效性及可行性.研究结果表明,相比于传统的正交匹配追踪算法、单任务贝叶斯压缩感知算法和多任务贝叶斯压缩感知算法,利用结合迭代阈值法的多任务贝叶斯压缩感知重构算法计算得到的重构信号与原始信号吻合度更好,性能更优.High economic cost caused by massive data acquisition,storage and transmission is a critical problem in the structural health monitoring.Thus,vibration responses reconstruction based on multi-task Bayesian compressed sensing combined with iterative threshold method is proposed in this paper.The compressed sensing theory combined with iterative threshold method is firstly utilized to compress and collect the data,and then multi-task Bayesian compressed sensing reconstruction algo­rithm is used to reconstruct the responses with limited measurements.The effectiveness and feasibility of the proposed method were verified by field ambient vibration test data of Ji'an Bridge.The results demonstrate that the reconstructed responses used the multi-task Bayesian compressive sensing combined with the iterative threshold algorithm have better accuracy than the results used orthogonal matching pursuit algorithm,the single-task Bayesian compressive sensing algorithm and the multi-task Bayesian compressive sensing algorithm.

关 键 词：结构健康监测 压缩感知 多任务贝叶斯 迭代阈值法 信号重构 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]