递归式最小均方（RLS）算法的基本思想是力图使在每个时刻对所有已输入信号而言重估的平方误差的加权和最小，这使得RLS算法对非平稳信号的适应性要好。与LMS算法相比，RLS算法采用时间平均，因此，所得出的最优滤波器依赖于用于计算平均值的样本数，而LMS（NLMS）算法是基于集平均而设计的，因此稳定环境下LMS（NLMS）算法在不同计算条件下的结果是一致的-Recursive least-mean-square (RLS) algorithm for the basic idea is to try to make in every moment of all the input signal in terms of re-evaluation of the weighted squared error and the smallest, which allows non-stationary RLS algorithm for adaptive signal better. Compared with the LMS algorithm, RLS algorithm uses the average time, therefore, the resulting optimal filter depends on the used to calculate the average number of samples, and the LMS (NLMS) algorithm is designed based on set average, and therefore a stable environment LMS (NLMS) algorithm in different conditions, the results of the calculation is consistent