摘要
基于经验模式分解(EMD)时间尺度滤波特性,在引入相关度分析的基础上提出了EMD相关度去噪方法。首先对含噪信号进行EMD分解得到信号各个本征模函数(IMF)分量,再根据所建立的相关度阈值函数计算各个分量的相关度值,在与预定阈值比较得到满足阈值要求的IMF分量,然后对这些分量进行信号重构得到去噪信号。该方法消除了EMD时间尺度滤波不适用于噪声和信号在IMF成分混叠情况下的限制。通过对平稳含噪信号和非平稳含噪信号进行的去噪仿真研究,表明了该方法的有效性。通过轧机在轧钢时实测信号分析验证了该方法的可靠性。
Based on the idea of the correlation degree and the EMD time scale filtering, the EMD de-noising method based on the correlation degree is proposed. Firstly the EMD decomposition is carried on to the signal which has noise for obtaining each IMF component of the signal. Then according to the established threshold value function of the correlation degree, the values of correlation degree for every component are calculated. After comparing to the predetermined threshold value, the IMF components which are satisfied the request of threshold value are obtained. Then the denoised signal is obtained by signal reeonfiguration. This method eliminates the limit that the EMD time scale filtering is not suitable in the condition of the noise and the signal which mixed in the IMF ingredient. Through the denoising simulation research to the stationary and nonstationary signals which have noise, the validity of this method is proved; the example analysis of steel rolling signal show the reliability of the method.
出处
《计量学报》
CSCD
北大核心
2009年第1期73-77,共5页
Acta Metrologica Sinica
关键词
计量学
经验模式分解
时间尺度
相关度
信号去噪
信噪比
Metrology
Empirical mode decomposition
Time scale
Correlation degree
Signal de-noising
Signal-tonoise ratio