论文部分内容阅读
把小波包变换和神经网络相结合,对含多模噪声信号进行消噪研究.首先对含噪信号序列进行小波包分解,得到不同尺度下的小波包分解系数,接着以这些不同频带的分解系数作为三层前向BP神经网络的输入特征向量,通过神经网络不断的修正和优化处理,最后以处理后的分解系数进行小波包重构,从而达到消噪的目的.实际计算及仿真表明,小波包神经网络消除多模噪声是一种非常有效的方法.
The wavelet packet transform and neural network are combined to denoise the multi-mode noise signal.First, the wavelet packet decomposition of the noisy signal sequence is carried out to get the wavelet packet decomposition coefficients at different scales, and then the decomposition coefficients of these different frequency bands As the input eigenvector of the three-layer forward BP neural network, through the continuous correction and optimization processing of the neural network, and finally the wavelet packet reconstruction with the processed decomposition coefficient, so as to achieve the purpose of noise elimination.The actual calculation and simulation show that the wavelet Packet neural network to eliminate multi-mode noise is a very effective method.