在LMS算法中，由于滤波器抽头权向量修正项与抽头输入向量u(n)成正比，因此当u(n)较大时，LMS算法的梯度噪声将被放大。采用归一化LMS(NLMS)算法可以克服LMS算法这一缺点。-In the LMS algorithm, since the filter tap weight vector correction term and tap input vector u (n) is proportional, so when u (n) is large, the gradient noise LMS algorithm will be magnified. Using the normalized LMS (NLMS) algorithm LMS algorithm can overcome this disadvantage.