针对低信噪比超宽带信号的消噪问题,提出一种改进的基于经验模式分解(EMD)的消噪算法。该算法首先对含噪信号进行EMD分解,得到多个固有模态函数(IMF)分量,然后选取高阶IMF重构原信号,达到消噪的目的。针对对UWB信号的IMF重构过程中阶数阈值难以确定的问题,通过数值仿真的方法,得到信号分量和噪声分量在不同阶IMF上的能量分布特性;在对所得特性进行分析的基础上,设计了一种数据自适应的阶数阈值选取算法,解决了EMD消噪中的阶数阈值选取问题。仿真结果表明,EMD消噪算法能够在较低信噪比下提供平均10 dB的信噪比增益,可以有效地对超宽带信号进行消噪。
To solve the problem of denoising ultra- wideband(UWB) signals, a novel denosing algorithm based on empirical mode decomposition (EMD) is proposed. The algorithm first uses EMD to decompose the noised signal into intrinsic mode functions(IMF), then uses high order IMFs to reconstruct the original signal. In order to get the proper .threshold in IMF reconstructing, the energy distribution property of each IMF is examined by numeric simulation. According to the distribution property, a data- adaptive algorithm on choosing the threshold is proposed. Simulation results show that the proposed denoising algorithm can improve the signal signal - to - noise ratio(SNR) by 10 dB under low SNR conditions.