Gamma自适应滤波器及其最小均方误差训练方法  

THE GAMMA ADAPTIVE FILTER AND ITS MINIMUM-MEAN-SQUARE-ERROR CRITERION-BASED TRAINING METHODS

作　　者：罗忠[1] 朱重光[1] 谢永斌[2] 

机构地区：[1]中国科学院遥感应用研究所,北京100101 [2]西安工业学院,西安710032

出　　处：《测试技术学报》1997年第4期1-6,共6页Journal of Test and Measurement Technology

摘　　要：本文提出一种新的自适应滤波器——Gamma 滤波器及其基于最小均方误差准则的训练方法。首先建立了最小均方误差准则下 Gamma 滤波器的 Wiener-Hopf 方程,然后分别提出3种训练算法:(1)高斯-牛顿算法;(2)确定性梯度算法;(3)LMS算法。最后给出Gamma 滤波器在随机信号预测中的应用和仿真结果。A new adaptive filter--the Gamma filter and its minimum-mean- suare-error criterion-based training methods are presented. We first derive the Wiener-Hopf equations for the Gamma filter using the minimum-mean-square-er-ror criterion, and then develop three methods for training the Gamma filter: (1) Gauss-Newton method; (2) deterministic gradient descent method; (3) the LMS method. Finally the application of the Gamma filter to random signal prediction and the corresponding simulation results are given.

关 键 词：滤波器 最小均方误差 自适应滤波器 

分 类 号：TN713[电子电信—电路与系统]