遥测振动信号单通道盲源分离自适应滤波幅度校正方法

Single channel blind source separation adaptive filtering amplitude correction method for telemetry vibration signals

针对基于经验模态分解(empirical mode decomposition,EMD)和独立分量分析(independent component analysis,ICA)的单通道盲源分离幅度不确定性问题,根据最小失真准则提出了一种自适应滤波幅度校正方法。利用EMD将单通道信号分解为一系列本征模态函数(intrinsic mode function,IMF),依据对数坐标下的边际谱分布确定单通道信号包含的独立分量数目。选择对应的IMF组合作为观测信号分量,利用ICA完成分离。根据分离信号数目确定横向滤波器阶数,并将分离信号作为滤波器的输入信号分量。利用滤波器输出和原始单通道信号设计目标函数,自适应调整滤波器系数使算法完成收敛,算法收敛后的滤波器权系数即为对应分离信号的幅度校正系数。仿真及飞行器试验遥测振动信号的处理结果证明在EMD-ICA基础上,该方法可准确得到信号各分量的幅度信息,为遥测振动信号进行时域统计及时频分析中能量检测提供了有效技术途径。

Here,aiming at the amplitude uncertainty of single channel blind source separation based on empirical mode decomposition(EMD)and independent component analysis(ICA),an adaptive filtering amplitude correction method was proposed according to the minimum distortion criterion.A single channel signal was decomposed into a series of intrinsic mode functions(IMFs)with EMD,and the number of independent components contained in the single channel signal was determined according to the marginal spectral distribution under logarithmic coordinates.The corresponding IMF combination was selected as the observation signal component,and ICA was used to complete the separation.The order of the transverse filter was determined according to the number of separated signals,and separated signals were used as the input signal component of filter.The objective function was designed by using filter output and the original single channel signal,and filter coefficients were adaptively adjusted to make the algorithm complete convergence.The filter weight coefficients after algorithm converging were amplitude correction coefficients of the corresponding separated signals.The processing results of simulation and aircraft test telemetry vibration signals showed that based on EMD and ICA,the proposed method can accurately obtain amplitude information of signal’s each component;it can provide an effective technical path for time domain statistics and energy detection in time-frequency analysis of telemetry vibration signals.