基于BP神经网络的高阶FIR多阻带滤波器优化设计

Study on Optimal Design of High-Order FIR Filter with Multi-Band Stop Based on BP Neural Network

摘要: 本文对传统BP神经网络算法加以改进，以余弦基函数作神经元的输出函数，克服了传统BP神经网络收数速度慢、学习步长低的致命缺陷。通过对线性相位FIR滤波器幅频特性和搭建余弦基神经网络模型对二者进行研究，并通过仿真研究得出该算法在高阶多阻带滤波器设计中的具有很强的优越性。

Abstract: This paper improves the traditional neural network and conquers its disadvantages of the slow convergence speed and the low learning efficiency with cosine basis functions as output functions of neural network unit. It studies on the amplitude-frequency characteristic of the FIR filters with linear phase by putting up models of neural network based on the cosine basis functions, and drawing the conclusion of relationship between them. The simulation results show that the algorithm has advantages over the design field of high-order FIR filter with multi-band stop.