将多尺度小波变换的理论引入到LMS自适应滤波器的设计中,分析了基于多尺度正交小波变换的自适应滤波算法的原理;将变步长LMS算法与多尺度小波变换的思想结合,提出了一种新的小波自适应滤波算法(MSWT-MVSS-LMS),新算法既减少了输入向量自相关矩阵条件数,又克服了固定步长LMS算法在收敛速度与收敛精度方面与步长因子μ的矛盾,获得了更好的收敛速度和稳定性。仿真结果表明新算法是有效的和优越的。 The theory of multi-scale wavelet transform is introduced into the design of LMS adaptive filter, and the principle of adaptive filter algorithm based on multi-scale orthogonal wavelet transform is analyzed. Combining the idea of ​​variable-step LMS algorithm and multi-scale wavelet transform, A new wavelet adaptive filtering algorithm (MSWT-MVSS-LMS) is proposed. The new algorithm not only reduces the condition number of the autocorrelation matrix of the input vector, but also overcomes the disadvantages of the fixed-step LMS algorithm in terms of convergence speed and convergence accuracy, The contradiction between the factor μ, get better convergence speed and stability. Simulation results show that the new algorithm is effective and superior.