APP下载

基于奇异值分解改进观测矩阵的FBG传感信号处理

2020-09-02刘航王波郎代志

软件导刊 2020年8期

刘航 王波 郎代志

摘 要:针对地下管廊环境恶劣复杂、噪声干扰较大的问题,提出一种基于奇异值分解的改进观测矩阵方法,通过提高压缩感知处理光纤布拉格光栅传感信号精度,完成噪声预处理。对观测矩阵进行分解重构,提高信号重构保真度。首先对随机观测矩阵进行正交化处理,再对其进行奇异值分解,将特征矩阵特征值改为恒定值,带入新的特征值后将产生新的观测矩阵。对信号进行观测,并采用信噪比、重构误差等指标确定稀疏度K。仿真结果表明,该方法重构精度提高约72%。

关键词:压缩感知;奇异值分解;观测矩阵;传感信号

DOI:10. 11907/rjdk. 192419 开放科学(资源服务)标识码(OSID):

中图分类号:TP393文献标识码:A 文章编号:1672-7800(2020)008-0202-05

Abstract: Aiming at the problem of large noise interference in the harsh complex environment of underground pipe gallery, we propose an improved observation matrix method based on singular value decomposition(SVD) to improve the accuracy of the compression sensing (CS) processing fiber Bragg grating sensing signal and achieve the purpose of noise preprocessing. The signal reconstruction fidelity is improved by decomposing and reconstructing the observation matrix. Firstly, the orthogonal observation matrix is orthogonalized, and then the singular value decomposition is performed. The eigenvalue of the feature matrix is changed to a constant value, and a new eigenvalue is generated to generate a new observation matrix to observe the signal. The SNR and reconstruction error are used to simulate and evaluate. The results show that the reconstruction accuracy of the proposed algorithm is improved by about 72%.

Key Words: compressed sensing; singular value decomposition; observation matrix; sensing signals

0 引言

随着我国城市化进程的加快,地下管廊设施规模建设逐渐庞大、密集,结构布局复杂,安全监管诉求十分迫切[1-3]。地下管廊因处于封闭的高湿恶劣环境,传感器稳定性差、寿命短、运维频繁。光纤传感器因其耐腐蚀、抗电磁干扰等优点被广泛应用于煤矿矿井安全检测[4-5]、地面沉降监测[6]、地下管廊等方面。由于应用环境恶劣,FBG传感信号在采集和传输过程中容易引入噪声,从而使信号后期分析更困难。因此,对被噪声污染的信号进行去噪预处理显得尤为重要。

小波去噪因其多尺度、多分辨率特性在信号处理领域受到广泛关注,但小波去噪效果受小波基、分解尺度、阈值和阈值函数影响,处理效率有限,若影响参数选择不合理将导致较大的重构偏差[7]。……

登录APP查看全文