摘要
针对循环相关熵谱估计循环周期图检测(CPD)算法存在计算效率低、频谱分辨率低和易产生“频谱泄漏”的问题,该文提出一种循环相关熵谱估计的Correntrogram算法。Correntrogram算法借鉴Wigner-Ville分布(WVD)时频分辨率高的优势,将WVD算法中瞬时自相关函数替换为时变自相关熵函数,即可得到一种循环相关熵谱密度估计算法,称为Correntrogram。首先计算信号的时变自相关熵函数矩阵,再计算时变自相关熵函数矩阵各行的FFT,得到循环自相关熵函数矩阵,最后计算循环自相关熵函数矩阵各列的FFT,得到循环相关熵谱密度。通过仿真调幅信号的处理结果证明:Correntrogram算法有效提高了循环相关熵谱估计效率,避免了“频谱泄漏”,提高了频谱分辨率,该算法程序运行可靠。
A high computationally efficient algorithm for cyclic correntropy spectral analysis is presented which is named as Correntrogram algorithm.Correntrogram algorithm overcomes the problems of Cyclic Periodogram Detection(CPD)method,such as high computational cost,low resolution and spectrum leakage.Correntrogram utilizes the advantages of Wigner-Ville Distribution(WVD)which has high time frequency resolution.By replacing the time-varying autocorrelation function with the time-varying auto-correntropy function in the WVD algorithm,the cyclic auto-correntropy spectral density estimation algorithm can be realized.First,the time-varying auto-correntropy function matrix of the signal is calculated,and then the Fast Fourier Transform(FFT)of each row of the time-varying auto-correntropy function matrix is computed to get the cyclic auto-correntropy function matrix.Finally,the FFT of each column of the cyclic auto-correntropy function matrix is calculated to get the cyclic auto-correntropy spectral density function.The validity of the proposed estimator is demonstrated on a simulative amplitude modulation signal.The simulative result shows that not only the proposed estimator is computationally efficient,but also has high frequency resolution and overcomes the spectrum leakage.The performance of Correntrogram is better than that of CPD method.
作者
李辉
LI Hui(School of Mechanical Engineering,Tianjin University of Technology and Education,Tianjin 300222,China)
出处
《电子与信息学报》
EI
CSCD
北大核心
2021年第2期310-318,共9页
Journal of Electronics & Information Technology
基金
国家自然科学基金(51375319)。
关键词
相关熵
循环相关熵函数
循环相关熵谱密度函数
循环平稳信号
循环谱分析
谱相关密度
Correntropy
Cyclic correntropy function
Cyclic correntropy spectral density function
Cyclostationary signals
Cyclic spectral analysis
Spectral correlation density