摘要
提出一种改进的自适应变步长最小均方(LMS)算法,该算法利用e4(n)和遗忘因子λ(n)共同调整步长,同时具有在初始阶段和未知系统时变阶段自适应步长增大而稳态时步长变小的特点,更好地解决了稳态误差与收敛时间之间的矛盾。将该算法应用到系统辨识中,与一般的变步长算法相比,改进算法具有更快的参数辨识速度和更小的稳态误差,同时还具有很好地跟踪多时变系统的能力。
An improved Least Mean Square（LMS） algorithm with adaptive step size is proposed. Utilizing the fourth power of instantaneous error and forgetting factor to adjust the step size, the improved algorithm increases adaptively at the beginning of the algorithm or unknown system changing with time, and it is smaller during the steady state. Meanwhile the algorithm ef- ficiently overcome the discrepancy between the convergence rate and the steady error. When this algorithm is applied to system identification,a significant improvement can be achieved in the identifying speed, smaller steady error and better tracking capability,as compared with the traditional algorithms with adaptive step size.
出处
《现代电子技术》
2010年第6期145-148,共4页
Modern Electronics Technique
关键词
最小均方算法
变步长
自适应滤波
系统辨识
LMS algorithm
variable step size
adaptive filter
system identification
