包络均值滤波算法实时检测微弱信号

在强噪声环境下提取微弱信号一直是信号处理方法研究的难点,提出一种基于包络均值滤波(EMF,Envelope Mean Filter)的微弱信号提取算法,可以实时恢复出深埋在均匀分布噪声中的微弱缓变非周期信号的波形.EMF算法通过对混有均匀分布噪声的信号进行过采样,对信号的包络线进行实时分析,并用上、下包络的均值作为真实信号的估计值.为了使输出信号平滑、无跳变,采用实时滑模平均滤波器对估计值进行滤波,最终恢复出的原始信号.通过仿真研究对包络的分析算法、最优衰减系数的选取、过采样系数、平滑序列长度和不同信噪比的微弱信号对还原精度的影响进行了研究.EMF算法简单,可使信噪比提高60 dB以上.已经成功用于IEEE 1588精密时钟同步的时钟伺服算法中.

Real-time weak signal detection with the envelope mean filter algorithm

The weak signal detection in strong noise environment had been a big difficulty in signal processing domain. A new algorithm that was the envelope mean filter(EMF) was discussed. The waveform of the weak signals that deeply buried in uniformly distributed noise can be recovered by EMF algorithm in real time. By using the mean of upper border and lower border of the envelope as the estimated value of the weak signal, the output waveform was smoothed by a real-time sliding mode mean filter. The basic principles of EMF algorithm were described. The simulation on the envelope of the analysis algorithm, the optimal attenuation coefficient a, oversampling factor k, smooth sequence length M and different SNR of the weak signal to the accuracy of the impact of the recovered signal were studied. The simulation results show that EMF algorithm can improve SNR over 60dB. The EMF algorithm has been successfully used for IEEE 1588 clock synchronization clock precision servo algorithm.