电路测试响应信号的GP-KSVD稀疏重构算法

发布时间：2018-08-07 15:42

【摘要】：电路系统测试响应信号具有周期性强、分布较稀疏的特点,针对电路系统测试响应信号的压缩重构问题进行了研究,提出了基于梯度方向追踪的K奇异值分解(GPKSVD)稀疏重构算法。结合单一响应信号以及混合信号其自身特点进行字典训练,利用更新后字典对含噪信号进行梯度追踪稀疏表征,通过对含噪信号的重构,实现了去噪的目的,算法计算复杂度低,储存量小,具有较好的重构效果。仿真中将GP-KSVD表征与使用随机字典、离散余弦字典(DCT)的表征进行比较,从信噪比(SNR)以及相对均方误差(RMSE)2项指标中得出使用KSVD字典具有更好的重构去噪效果;此外将GP-KSVD稀疏重构算法与正交匹配追踪正交匹配追踪(OMP)-KSVD、预处理共轭梯度追踪(PCGP)算法进行比较,得出GP-KSVD的计算时间最短、重构精度更高的结论,并且进行了实测验证。算法可用来对测试响应信号进行预处理,为电路系统设备性能的评估分析提供了理论依据。
[Abstract]:The test response signal of circuit system has the characteristics of strong periodicity and sparse distribution. The compression reconstruction problem of the test response signal of circuit system is studied and a K singular value decomposition (GPKSVD) sparse reconstruction algorithm based on gradient direction tracing is proposed. Combining the characteristics of single response signal and mixed signal, the dictionary training is carried out, and the denoising signal is represented by gradient tracing sparse representation using the updated dictionary. The purpose of de-noising is realized by reconstructing the noisy signal. The algorithm has the advantages of low computational complexity and small amount of storage. The GP-KSVD representation is compared with the (DCT) representation using random dictionaries and discrete cosine dictionaries. From the SNR (SNR) and the relative mean square error (RMSE), it is concluded that using KSVD dictionary has better denoising effect. In addition, the GP-KSVD sparse reconstruction algorithm is compared with the orthogonal matching tracing (OMP) -KSVD algorithm, and the pre-processing conjugate gradient tracking (PCGP) algorithm is compared. The conclusion that the GP-KSVD has the shortest computing time and higher reconstruction accuracy is obtained, and the experimental results are verified. The algorithm can be used to preprocess the test response signal and provide a theoretical basis for evaluating and analyzing the performance of circuit system equipment.
【作者单位】：北京航空航天大学电子信息与工程学院;北京航空航天大学数学与系统科学学院;
【基金】：国家自然科学基金(61379001)~~
【分类号】：TN911.7

【参考文献】