摘要
本文介绍了基于神经网络的FIR滤波器的设计方法 ,给出了神经网络的训练算法 ,并利用该设计方法构造了一个 5 0Hz带阻滤波器 ,与常规的用窗函数法设计的滤波器进行了对比 ,发现基于神经网络的方法所设计出的滤波器通带阻带无过冲无波动 ,具有更好的性能 .最后将这两种滤波器作用于受噪声污染的脑电信号 。
The design of a FIR(Finite impulse response) filter based on neural networks is introduced, A 50 Hz band stop filter has been designed by this method. Comparing with the window function, this algorithm can more precisely control the stop band frequency without causing wavelet and impulse. As a conclusion,a EEG signal polluted by 50Hz noise was filtered by these two filters with experimental results given separately.
出处
《山东大学学报(工学版)》
CAS
2003年第1期50-54,共5页
Journal of Shandong University(Engineering Science)
基金
山东省自然科学基金
项目编号 :Y2 0 0 0C2 5