全文下载:
202602018.pdf
文章编号:1672-6987(2026)02-0121-10;DOI:10. 16351/j. 1672-6987. 2026. 02. 018
宋 娟,施 威,王景景* (青岛科技大学 信息科学技术学院,山东 青岛 266061)
摘 要:针对欠定情况下传统盲源分离方法难以恢复出源信号的问题,本研究提出基于能量 峰值检测(energy peaks detection,EPD)和密度峰值聚类(density peaks clustering, DPC)的欠 定盲源分离方法:EPD-DPC。首先,针对低信噪比下难以提取能量峰值的问题,提出了基于 改进灰狼优化算法的小波阈值法(optimized gray wolf optimization-wavelet threshold, OGWOWavelet)并用其对观测信号进行预处理,有助于使用 EPD 提取能量峰值处的频率值。然后, 将预处理后的信号转换到时频域计算其能量,使用 EPD 提取能量峰值处的频率值,接着使用 DPC 算法对频率值处的时频点进行聚类得到混合矩阵的估计。最后,利用 L1 范数最小化方法 恢复出源信号。为验证所提算法的有效性,本文实施了不同信噪比下的仿真实验。仿真结果 显示,基于 EPD-DPC 的欠定盲源分离方法能够有效抑制噪声的影响,提高了混合矩阵估计的 精度,所恢复出的源信号更接近真实源信号,分离效果更好。
关键词:欠定盲源分离;混合矩阵估计;能量峰值检测;密度峰值聚类算法
中图分类号:TN 911. 7 文献标志码:A
引用格式:宋娟,施威,王景景 . 基于 EPD-DPC 的欠定盲源分离方法[J]. 青岛科技大学学 报(自然科学版),2026,47(2):121-130.
SONG Juan, SHI Wei, WANG Jingjing. Underdetermined blind source separation method based on EPD-DPC[J]. Journal of Qingdao University of Science and Technology(Natural Science Edition),2026,47(2):121-130.
Underdetermined Blind Source Separation Method Based on EPD-DPC
SONG Juan,SHI Wei,WANG Jingjing (College of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061, China)
Abstract:In the problem of underdetermined blind source separation, the traditional blind source separation methods are difficult to recover the source signals. To solve this problem, we propose an underdetermined blind source separation method based on energy peaks detection (EPD) and density peaks clustering (DPC): EPD-DPC. Firstly, aiming at the problem that it is difficult to extract energy peaks under low signal-to-noise ratio, a signal preprocessing method based on the optimized gray wolf optimization-wavelet threshold (OGWOWavelet) algorithm is proposed and used to preprocess the observed signal, which is helpful for extracting the energy peaks. Then, the pre-processed signal is converted to time-frequency domain to calculate its energy, EPD is used to extract the frequency value at the energy peak, and then DPC algorithm is used to cluster the time-frequency points at the frequency value to obtain the estimation of the mixing matrix. Finally, the L1 norm minimization method is used to recover the source signal. In order to verify the effectiveness of the proposed algorithm, simulation experiments under different SNR are carried out in this paper. The simulation results show that the underdetermined blind source separation method based on EPD-DPC can effectively suppress the influence of noise, improve the accuracy of mixture matrix estimation, and the recovered source signal is closer to the real source signal, so the separation effect is better.
Key words:underdetermined blind source separation; mixed matrix estimation; energy peaks detection; density peaks clustering algorithm
收稿日期:2025-05-16
基金项目:国家自然科学基金项目(62101298,62171246).
作者简介:宋 娟(1995—),女,硕士研究生 . * 通信联系人 .