1、 计算机与通信学院计算机与通信学院 本科生毕业论文 LMSLMS 及其改进算法研究及其改进算法研究 2 LMS 及其改进算法研究 The study of LMS algorithm and its improve algorithms 摘摘 要要 因 LMS 算法具有低计算复杂度、在平稳环境中的收敛性好、其均值无偏地收敛到 wiener 解和利用有限精度实现算法时的稳定性等特性,使 LMS 算法成为自适应算法中应用最广泛的 算法。对 LMS 算法及其改进算法进行了研究,探讨了步长因子 n对各种算法收敛性、稳定 性的影响。 并用 MATLAB 对其学习曲线、 收敛速度等进行了仿真分析。 结果表
2、明, 变步长 n 的取值尤为重要,如果 (n)取较大值则具有较快的收敛速度,如果 (n)取值很小,则 MLMS 算法近似等效于 LMS 算法。它们的自适应过程较快,性能有了很大改进。 Abstract Because of low computational complexity, stable environment in the convergence of good, unbiased and its mean converges to the wiener solution and implementation algorithms using finite precision sta
3、bility and other characteristics, LMS algorithm as adaptive algorithm in the application of the most a wide range of algorithms.We have a detailed study on LMS algotithm and its complementary algotithm,disscused the step-sizes influent for the algorithms convergence speed and stability. And using MA
4、TLAB simulated the learning curve, convergence speed of LMS algotithm.The result observed that the value of variable step-size (n)is very important,if it is a bigger may have a fast convergence speed,but if not ,the NLMS algotithm can instead the LMS algotithm in the characteristics. In addition , they have a fast adaptive course and greatly progress in performance. Keywords:LMS algorithm,Adaptive,NLMS algorithm,Variable step,MATLAB simulation. 1.1 自适应滤波理论的发展 早在 20 世纪 40 年代,就对平稳随即信号