| 引用本文: | 岑翼刚,尉 宇,孙德宝.小波阈值神经网络在信号去噪及预测中的应用[J].控制理论与应用,2008,25(3):485~491.[点击复制] |
| CEN Yi-gang,WEI Yu,SUN De-bao.The application of wavelet threshold neural network in the de-noising and prediction[J].Control Theory and Technology,2008,25(3):485~491.[点击复制] |
|
| 小波阈值神经网络在信号去噪及预测中的应用 |
| The application of wavelet threshold neural network in the de-noising and prediction |
| 摘要点击 2195 全文点击 1786 投稿时间：2005-07-17 修订日期：2007-11-08 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/j.issn.1000-8152.2008.3.017 |
| 2008,25(3):485-491 |
| 中文关键词 最优阈值 神经网络 预测 信号去噪 |
| 英文关键词 optimal threshold neural network prediction signal de-noising |
| 基金项目 北京交通大学校人才基金资助项目(2008RC004). |
|
| 中文摘要 |
| 提出了一种小波阈值神经网络模型(wavelet threshold neural network, WTNN), 对合作式接收到的雷达信号进行去噪和预测. 这种网络模型把小波最优阈值去噪器加到神经网络中, 对带噪信号具有小波最优阈值去噪和预测的功能. 对小波系数作单层重构, 可简化训练算法, 使编程得到精简. 其次, 通过对训练算法进行分析, 得出了最优阈值及权值的调整公式. 最后通过对线性调频信号去噪及前向一步预测的实验结果可以看出, 当网络输入分别为带有高斯白噪声、高斯带限噪声、瑞利噪声的线性调频信号时WTNN得到的结果均优于利用Donoho阈值进行去噪后再预测的结果. |
| 英文摘要 |
| A wavelet threshold neural network (WTNN) model is proposed for denoising and prediction of cooperatively received radar signals. This WTNN incorporates a wavelet denoising layer with optimal wavelet thresholds into the neural network, for signal denoising and predicting. The training algorithm is simplified by the single-layer reconstruction of wavelet coefficients, leading to a compact programming. By analyzing the training algorithm, we derive the tuning formulas for searching optimal thresholds and network weights. The results of denoising and one-step ahead prediction for a linear frequency modulation signal with white Gauss noise, Gauss band-limited noise or Rayleigh noise show that the WTNN performs much better than the method of Donoho-threshold for denoising and prediction. |
|
|
|
|
|