The Research And Application Of Adaptive LMS Algorithm

Modern electronics technique has already been changed from analog technique toward digital technique, and the traditional analog signal processing has just been replaced by the digital signal processing. The digital filter is a part of digital signal processing, besides having the advantages of the stability, repetition and the good adaptability of the digital signal processing, it has still advantages of predictability and no phase shift, so we can use an imitate software to test a design in advance. We have a lot of methods to carry out digital filter, among them, the application of the least mean square( LMS) algorithm in the theories of adaptive filter is the most extensive. The reason is that the LMS algorithm has characteristics that includes low calculation complications, the constringency under steady environment, the mean no shift constringe to the Wiener solution and the stability using the limited accuracy to carry out the algorithm.This paper researches sorts of LMS algorithms in detail, including the basic LMS and it' s changed type, quantization error LMS algorithm, decorrelation LMS algorithm, adaptive lattic LMS algorithm ,parallel delayed LMS algorithm, and LMS - Newton algorithm. Among them, thequantization error LMS algorithm includes the sign error LMS algorithm and the sign data LMS algorithms, the decorrelation LMS algorithm includes time domain decorrelation LMS algorithm and transform domain LMS algorithm. Also, this paper utilizes MATLAB to imitate these sorts of algorithms, through the simulate result diagram, intuitionistic revelation various characteristics of LMS algorithm. In this paper ,we draw the study curve that basic LMS algorithm that ju is not same, ina sketch frame, this makes us keep intuitionistic view to see that along with increment of fi, the study curve of the algorithm also in the gradual increment. However, when // reaches a certainly value, then, if ju is continue increment, the study curve of basic LMS algorithm would fall into divergence. Same as the basic LMS algorithm, this paper also give study curve of the other sorts LMS algorithm, and discussed the performance of the other sorts LMS algorithm in brief.In addition, this paper detailed analysis the selection method that the study velocity parameter of LMS algorithm, analyzed the influence of parameter fi to LMS algorithm' s performance, and how tominish the influence. It also analyzed the statistics and tracking performance, disadjustment and nonsmoothness. The relation between disadjustment and nonsmoothness is discussed.Ultimately, this paper gives the project that application the LMS algorithm into the control system of excitation source with the advantage of DSP, and design thinking of hardware, software elucidation, hardware carries out frame diagram, software flow chart and actual electric circuit and parts of program.