基于峭度激励的变步长自适应谱线增强算法

提出了一种基于峭度激励的变步长自适应谱线增强新算法，该算法以输入信号峭度与误差信号峭度为联合激励因子，构造出一种指数型变步长的算法模型。该步长对高斯色噪声或非高斯色噪声均有一定的免疫性，跟踪时变信号的能力强，且收敛速度快。用实测的某水下目标辐射噪声数据进行了仿真实验。与传统的LMS算法相比，仿真结果表明，该算法具有较强的谱线增强能力、快速收敛性和良好的跟踪性能，能有效地实现目标线谱信号与混合色噪声环境的分离。

Kurtosis driven variable step size LMS adaptive line enhancer

A novel algorithm of kurtosis driven variable step size LMS based adaptive line enhancer (KDLMSBALE) was proposed. This algorithm is to use input signal kurtosis and error signal kurtosis as the independent variable of step size function. This step size has definite immunity to Gaussian noise or non_Gaussian noise, fast convergent speed, and a strong ability to track time_varying signals. Simulation tests were carried out using the real data of an underwater moving target and mixed colored noises composed of Gaussian distribution noise, uniformity distribution noise, and Raleigh distribution noise. The results show that the ability of the KDLMSBALE to suppress mixed colored noise is much stronger than LMSBALE (least mean square based adaptive line enhancement) and that KDLMSBALE outperformed LMSBALE in convergent rate, tracking performance, and extracting useful signals from noises.