论文部分内容阅读
本文研究具有最优下降步长序列的自适应滤波。最优一词此处系指：①对于给定的均方误差的初值，使均方误差经给走次数学习后达到最小；②对于均方误差的给定初值和终值，经过最少次数的学习，均方误差由初值下降至终值。对于①②两种不同意义的最优，我们采用动态规划和对数比例分割方法来求各自的最优步长序列。同时提供一组仿真结果。
This paper studies adaptive filtering with the optimal sequence of descending steps. The optimal term here refers to: ① for a given mean square error of the initial value, so that the mean square error to take to go after the number of learning to a minimum; ② For the mean square error of a given initial value and final value, after a minimum The number of learning, mean square error decreased from the initial value to the final value. For ① ② the best of two different meanings, we use the dynamic programming and logarithmic proportional division method to find the optimal sequence of steps. At the same time provide a set of simulation results.