QQ客服 官方微博 反馈留言

题名	自适应信号处理系统及其实现
作者	王林章
学位类别	博士
答辩日期	2004
授予单位	中国科学院声学研究所
授予地点	中国科学院声学研究所
关键词	自适应滤波 自适应快速横向算法
其他题名	Adaptive Filter experimental system and its implementation
中文摘要	在自适应信号算法的推导过程中，不同的目标函数定义方法可以导出不同特点的算法。本文从不同的目标函数定义方法入手，研究了目标函数与自适应算法的收敛速度、算法稳定性和运算复杂度之间的关系。本文系统地讨论了基于最小均方误差（LMS）和最小二乘方的自适应算法（LS）两种目标函数的自适应算法。在LMS的自适应算法方面，我们采用与已有算法不同的几何分析方法，深入分析了NLMS算法的收敛特性。研究分析发现，传统的NLMS算法的权系数更新是沿着输入信号矢量方向上进行的，当输入矢量之间是平行的或者夹角很小时，更新效率就会变低，所以我们介绍先正交化输入矢量，然后进行系数更新的算法。在LS方面，本文主要讨论了快速横向算法（FTF）。FTF算法由于其具有较低的运算复杂度而倍受关注，但是这种算法在实际的应用之中由于运算的有限精度而存在稳定性问题。因此我们对快速横向算法的稳定性进行了分析，并且发现更新过程中角参量对算法的稳定性有很大的影响。本文定义了一个控制系数更新方法的补救变量，并根据输出误差和DT检测结果对遗忘因子分段式取值，提高了算法的收敛速度。通过实验，我们比较了基于LMS和LS的两种算法的特点，并证明我们修改后的FTF算法对算法的收敛速度和稳定性都得到明显的提高，但是相对于高阶系统的实际应用，算法稳定性离实际要求还具有一定距离，在论文最后，我们通过总结给出了新的研究方向。
英文摘要	During the derivation of adaptive filter algorithm there are many ways to define the objective function that satisfies the system optimality, which will surely lead to different kind of algorithms with different characteristics. As a result, this dissertation studies the stability, convergence speed and computation complexity of different adaptive algorithms. This dissertation systematically studies different kind of adaptive filtering algorithms based on Lease Mean Square (LMS) and Lease Square (LS) objective function. In the respect of LMS-based adaptive algorithms, we analyze the convergence characteristic of the NLMS algorithm in some depth by using different approach from that used to derive the existing algorithms, we find that the weight coefficient vector in the typical NLMS algorithm is updated along the input signal vector, the efficiency of which become quite low when the input signal vectors are parallel or the inclinations between them are small, so we suggest an algorithm which firstly orthogonalizes the input signal vectors, then updates the weight coefficient vector. On the hand of LS based algorithm, the thesis mainly deals with the Fast Transversal Filter (FTF) algorithms, which are very attractive due to their low computational complexity. However, these algorithms are known to face stability problems in practical implementation as a result of limited precision. So we deeply analyze the stability of the FTF algorithm and from the analysis we find that the angle variable has great effect on the stability of the algorithm, so a new defined rescue variable is proposed to control the update of coefficient vectors. At the same time the forgetting factor is varied piecewise according to the detection of DT and the output of adaptive filter, as a result, the convergence speed of the algorithm is improved. Two different kinds of adaptive algorithms based on LMS and LS are compared through experiments. Experimental results show the obvious enhancement of the modified adaptive algorithm. But the improvement is still limited for the practical implementation of high order system, and new research directions are introduced at the last sect of the thesis.
语种	中文
公开日期	2011-05-07
页码	75
内容类型	学位论文
源URL	[http://159.226.59.140/handle/311008/874]
专题	声学研究所_声学所博硕士学位论文_1981-2009博硕士学位论文
推荐引用方式 GB/T 7714	王林章. 自适应信号处理系统及其实现[D]. 中国科学院声学研究所. 中国科学院声学研究所. 2004.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。