基于BP神经网络改进的FIR滤波器研究

Study on Improved FIR Filters Using BP Neural Networks

详细研究了BP神经网络算法,分析了FIR线性相位滤波器幅频响应与BP神经网络算法的关系,对滤波器设计方法进行了改进,给出了设计低通滤波器实例。它克服了传统方法的主要缺陷,不涉及逆矩阵的复杂计算。仿真结果表明,该算法在低通、高通和带通滤波器设计中各项性能接近理想状态。

This paper studies BP neural networks algorithm, discusses the relations between the amplitude-frequency response of the FIR filters with linear phase and the algorithm of the BP neural networks,improves the methods of designing the filters and gives the examples of designing low-pass filters.It conquers the primary disadvantages of the conventional methods,and need not to deal with the complex computing of the contrary matrix . The simulation results show that the varieties of performance indexes approch those under ideal conditions by using this algorithm in the designs of the low-pass,high-pass and band-pass filters.