摘要
针对二范数的自适应滤波算法受脉冲信号中异常值影响较大的问题,提出一种基于最大相关熵的分布式仿射投影算法。算法通过增加输入信号维度,采用数据重用方法来加快算法收敛速度,有效解决了有色输入信号对分布式最小均方算法的收敛速度变慢问题。同时针对实际应用中广泛存在的非高斯噪声严重影响DAPA收敛性能与稳态性能的问题,采用高斯核的DMCC算法,以最大相关熵作为代价函数。算法利用熵对异常值不敏感的特性有效抑制了非高斯噪声干扰。在非高斯环境和有色输入信号条件下,通过实验仿真,表明该算法具有良好的收敛性和稳态性。
Aiming at the problem that the two-norm adaptive filtering algorithm is greatly affected by outliers in pulse signals,a diffusion affine projection algorithm based on maximum correntropy is proposed.The algorithm accelerates the convergence speed by increasing the dimension of input signals and adopting data reuse method,which effectively solves the problem that the convergence speed of diffusion least mean square algorithm is slow due to colored input signals.At the same time,aiming at the problem that non-Gaussian noise seriously affects the convergence and steady-state performance of DAPA,the DMCC algorithm with Gaussian kernel is adopted,and the maximum correntropy is taken as the cost function.The algorithm makes use of the insensitivity of entropy to outliers to effectively suppress non-Gaussian noise interference.Under the condition of non-Gaussian environment and colored input signal,the experimental simulation shows that the algorithm has good convergence and stability.
作者
于和芳
郭莹
YU Hefang;GUO Ying(School of Information Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)
出处
《微处理机》
2020年第5期30-35,共6页
Microprocessors
关键词
分布式自适应滤波
最大相关熵
仿射投影算法
Diffusion adaptive filtering
Maximum correntropy
Affine projection algorithm
