摘要
详细讨论了4型线性相位滤波器的幅频特性与正弦基函数神经网络算法的关系,分析了神经网络系统的稳定条件,给出了FIR滤波器优化设计实例。根据4型FIR滤波器的幅频响应特性,构造出一个相应的神经网络模型,并建立了FIR线性相位数字滤波器的神经网络算法。该算法通过训练神经网络权值,使设计的数字滤波器与希望得到的FIR线性相位滤波器的幅频响应之间的误差平方和最小化,从而获得FIR线性相位数字滤波器的脉冲响应。计算机仿真表明了该算法的有效性和优异性能。
The relationship between the amplitude -frequency characteristic of linear phase type -four FIR filter and the algorithm of neural network based on sine basis function is discussed in detail. Related convergence theorem is deduced. Examples of optimized design of type - four FIR filter are also introduced. According to the amplitude - frequency response characteristic of type- 4 FIR filter, the paper presents a corresponding neural -network model,and establishes an algorithm for linear - phrase FIR digital filter. By optimizing neural network weight,this algorithm minimize the squared -error function in the frequency - domain between the digital filter designed and linear phase filter expected. As a result,achieving the impulse response of FIR linear phase digital filter. Simulation results show that the proposed neural network is effective for the design of FIR high - pass.
出处
《现代电子技术》
2007年第23期80-81,84,共3页
Modern Electronics Technique
关键词
神经网络
正弦基
高通滤波器
优化设计
线性相位
neural network
sine basis function
high - pass filter
optimized design
linear phase