基于串联动量滤波器的自适应谱线增强器

针对水下环境噪声的非高斯性特点,利用最小均方算法、动量最小均方算法、变步长算法和极性算法的性能,将基于短时相关的自适应谱线增强器修改为基于短时相关动量滤波算法的自适应谱线增强器(SCMBALE),提出了四阶累积量变步长极性动量滤波算法,将基于该算法的自适应谱线增强器(FVSMBALE)、基于动量最小均方算法的自适应谱线增强器(MLMSBALE)及基于短时相关动量滤波算法的自适应谱线增强器(SCMBALE)依次串联起来,构造出基于串联动量滤波器的自适应谱线增强器(SMFBALE)。通过理论分析与仿真实验对该增强器的性能进行了研究。结果表明:该增强器在抑制非高斯噪声、增强线谱信号、跟踪时变信号等方面的性能优于单个的FVSMBALE、MLMSBALE和SCMBALE。

Series Momentum Filter Based Adaptive Line Enhancer

Fourth order cumulant Variable step-size Sign Momentum filtering algorithm(FVSM)was proposed based on analyzing the non-Gaussian features of underwater acoustic environment and utilizing the advantages of Momentum Least Mean Square(MLMS)algorithm,LMS algorithm,variable step-size algorithm and sign one.Series Momentum Filter Based Adaptive Line Enhancer(SMFBALE),whose first filter is a FVSM Based Adaptive Line Enhancer(FVSMBALE),whose second filter is a MLMS Based Adaptive Line Enhancer(MLMSBALE),and whose third filter is a Short-term Correlation Momentum filtering Based Adaptive Line Enhancer(SCMBALE),was established.Filtering coefficients of the FVSMBALE were defined using fourth order cumulant diagonal slice of input signals and indirectly updated using input signals,sign of function of input signals,variable step size,and momentum filtering algorithm,filtering coefficients of the MLMSBALE were updated using second order cumulants and momentum filtering algorithm,and filtering coefficients of SCMBALE were updated using instantaneous autocorrelation function and momentum filtering algorithm.Theoretical analysis and simulation tests with real data of the underwater moving target-radiated noise show that the SMFBALE has immunity to Gaussian noise and that better ability to suppress mixed distributed noise or non-Gaussian noise and to trace time-varying signals is much stronger than that of the single MLMSBALE,the FVSMBALE,and the SCMBALE.