3型FIR带通滤波器优化设计

Optimal Design Study of High-Order Type-Three FIR Band-Pass Filter

研究了3型FIR线性相位滤波器的幅频特性与神经网络算法之间的关系,提出了该神经网络算法的收敛性定理,为神经网络训练中学习率大小的选择提供了理论依据.该算法有效避免了FIR高阶数字滤波器优化设计中求高阶矩阵逆的困难,为高阶FIR数字滤波器的优化设计提供了一种快速有效的设计方法.最后给出了FIR高阶数字滤波器的优化设计实例.计算机仿真结果表明了该神经网络并行算法在3型FIR高阶数字滤波器优化设计领域的有效性.

This paper traverses the relations between the amplitude-frequency response of FIR filter on type three with the linear phase and the algorithm of neural network, presents the convergence theorem of the neural-network algorithm, and the theorem gives the magnitude criterion of the learning rate training neural network. The algorithm in this paper effectively avoids the difficulty solving high-order inverse matrix, and is a fast and effective method in the field of the optimal design of high-order FIR digital filters. Finally, this paper gives the optimal design examples of FIR digital filters. The results show that the algorithm is efficient in the field of optimal design of the high order FIR digital filters on type three.