介绍自适应滤波算法、多尺度小波算法的基本原理和两种算法结合的实现过程。针对最小均方（LMS）自适应滤波算法不能同时提高收敛速度和收敛精度，提出变步长LMS自适应算法，在滤波过程中算法先用大步长跟踪，提高收敛速度，接近稳态时用小步长跟踪，提高收敛精度。为了能有更好的滤波效果，应该在算法的步长因子上有所突破。在抽样函数的基础上改进算法，并结合多尺度小波分解，使得滤波的效果更加理想。通过Matlab仿真实验，验证了改进算法具有更好的稳定性和优越性。%This paper describes the implementation process of adaptive filtering algorithms, the basic principles of multi-scale wavelet algorithm and a combination of two algorithms. As for the least mean square (LMS) adaptive fil-tering algorithm can not improve convergence speed and accuracy, so the proposed variable step size LMS algorithm, in the filtering process, the algorithm first with a stride length of track, improve the convergence speed, close to steady with a small step tracking, improve the convergence precision. In order to have a better filtering effect, it should be a breakthrough in the step factor algorithm. This paper improves the function on the basis of sampling algo-rithms, combined with multi-scale wavelet decomposition, so that the filtering effect is more ideal. By Matlab simula-tion to verify the improved algorithm has better stability and superiority.
展开▼
