基于阈值自学习小波算法的压力信号去噪方法  

Pressure Signal Denoising by Threshold Self-Learning Wavelet Algorithm

作　　者：郭新蕾[1] 杨开林[1] 郭永鑫[1] 

机构地区：[1]中国水利水电科学研究院水力学研究所,北京100038

出　　处：《数据采集与处理》2008年第3期322-326,共5页Journal of Data Acquisition and Processing

基　　金：国家自然科学基金(50679085)资助项目

摘　　要：提出了信号预滤波结合阈值自学习小波去噪的综合滤波方法。该法通过对恒定状态下带噪压力信号阈值自学习使得重构信号与期望输出均方误差最小来获得单一工况下的最佳去噪阈值,再将此阈值用于同一工况下整个时间段信号的去噪,不同工况下得到不同的最佳阈值进而获得最优输出。数值计算对比证明,该方法对噪声抑制作用明显,比传统小波去噪、改进神经网络去噪等方法效果更好。Pre-filter combined with threshold self-learning wavelet algorithm is proposed for pressure signal denoising. Firstly, the denoising threshold is self-learnt in the steady flow state, and then it is modified on a given limit to make the mean square errors between recon- struction signals and desirable minimal outputs, thus the corresponsive optimal denoising threshold in a single operating case can be obtained. These optimal thresholds are used for the whole signal denoising and different optimal thresholds are obtained in various cases. Simulation results and comparative studies show that the approach has an obvious effect of noise suppression. The method is superior to traditional wavelet algorithms and artificial neural networks.

关 键 词：压力信号 小波 阈值 去噪 人工神经网络 

分 类 号：TN911.72[电子电信—通信与信息系统]