基于脊路跟踪的变分非线性调频模态分解方法  

Multi-component signal separation using variational nonlinear chirp mode decomposition based on ridge tracking

作　　者：赵雅琴[1] 聂雨亭 吴龙文 张宇鹏 何胜阳 ZHAO Ya-qin;NIE Yu-ting;WU Long-wen;ZHANG Yu-peng;HE Sheng-yang(School of Electronics and Information Engineering,Harbin Institute of Technology,Harbin 150001,China;Beijing Research Institute,Huawei Technology Limited Company,Beijing 100095,China)

机构地区：[1]哈尔滨工业大学电子与信息工程学院,黑龙江哈尔滨150001 [2]华为技术有限公司北京研究所,北京100095

出　　处：《浙江大学学报（工学版）》2020年第10期1874-1882,共9页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(61671185)。

摘　　要：针对多个辐射源信号混合构成的多分量信号分离问题,提出基于脊路跟踪的变分非线性调频模态分解算法.该方法使用改进的脊路重组算法对时频分布图中各分量瞬时频率进行提取,将提取出的各分量瞬时频率作为变分非线性调频模态分解的预设频率;利用重构后的多分量信号进行瞬时频率提取,更新预设频率后继续模态分解;重复上述过程,直到迭代前、后频率差值小于预设阈值,输出对应的模态分解结果.实验结果表明,基于脊路跟踪的变分非线性调频模态分解算法比经典变分非线性调频模态分解算法具有更好的多分量信号分离效果.A variational nonlinear chirp mode decomposition algorithm based on ridge tracking was proposed aiming at the problem of multi-component signal separation caused by mixed signals from multiple emitters.The improved ridge path regrouping algorithm was used to extract the instantaneous frequency of each component from the timefrequency distribution,and the extracted instantaneous frequency of each component was used as the preset frequency of the variational nonlinear chirp mode decomposition.A repeated instantaneous frequency extraction was performed signal to update the preset frequency for iteration based on the reconstructed multi-component.The above processes were repeated until the frequency difference between two iterations was less than the preset threshold,while the corresponding mode decomposition results were output.The experimental results show that the variational nonlinear chirp mode decomposition algorithm based on ridge tracking has better performance of multi-component signal separation than the classical variational nonlinear chirp mode decomposition.

关 键 词：多分量信号 脊路重组 瞬时频率估计 变分非线性调频模态分解 

分 类 号：TN971[电子电信—信号与信息处理]