论文部分内容阅读
实际生活中产生的多模噪声会对信号产生严重的破坏甚至损失.传统LMS算法不仅收敛速度慢而且存在较大的稳态误差和失调量,故已不能很好地控制噪声.通过分析NLMS算法和Newton算法的优点,提出一种新的NLMS-Newton算法.该算法的收敛速度高,计算复杂度减小,且应用了随信噪比变化的变步长并修正了自相关矩阵,降低了算法收敛后的稳态误差和失调量.通过仿真表明,NLMS-Newton算法在收敛速度及稳态误差等方面都有了较大改善,并且能在很好地抑制多模噪声的同时提取出有用信号.“,”Signal can be damaged and even lost because of the multi-mode noise generated in actual life. Since the tradi-tional LMS algorithm has slow convergence speed,big steady-state error and high maladjustment,it can′t control the noise per-fectly. By analyzing the advantages of NLMS algorithm and Newton algorithm,a new NLMS-Newton algorithm is proposed. The algorithm has fast convergence speed and low calculating complexity,in which the step length changed with signal-to-noise ratio is applied and the autocorrelation matrix is corrected to reduce the steady-state error and maladjustment after the algorithm con-vergence. The simulation results show that the NLMS-Newton algorithm has improved greatly in the aspects of convergence speed and steady-state error,and can extract the useful signal while suppressing the multi-mode noise.