一种惯性传感器混叠干扰信号分离方法

Inertial sensor aliasing interference signal separation method

光电跟踪系统中,在惯性传感器混叠信号中分离出各种干扰进行单独抑制具有重要意义,针对目前不同干扰信号单独分离所存在的困难,提出对混叠干扰信号进行小波滤波,得到低频干扰信号,再对滤波后的信号进行集合经验模态分解,得到多个信号分量并利用主成分分析进行降维,最后通过盲源分离算法得到高频混叠信号中的不同频率段干扰信号。仿真结果表明该算法分离出的信号频谱不混叠并且与相对应源信号的相关系数高,分别为0.998 2、0.944 2、0.913 7。真实传感器数据实验证明,算法得到的频率为0.1Hz、8Hz、16Hz、33Hz的信号频谱不混叠,实现了混叠信号不同频段干扰信号的单独分离。

In electro-optical tracking systems,it is of great significance to separate out various interference in the aliasing signal of inertial sensor to suppress them individually.Aiming at the difficulty of separate separation of different interference signals,wavelet filtering is applied to the aliased interference signals to obtain low-frequency interference signals,and then the filtered signal is subjected to ensemble empirical mode decomposition to obtain multiple signal components and the principal component analysis is used to reduce the dimension.Finally,the blind source separation algorithm is used to process the signals to obtain different frequency segment interference signals in the high-frequency alias signals.Simulation results show that the signals separated by the algorithm are not aliased and their correlation coefficients with the corresponding source signals are high,which are 0.998 2,0.944 2,and 0.913 7,respectively.Experiments on real sensor data show that the frequency spectrum of the signal obtained by the algorithm is 0.1 Hz,8 Hz,16 Hz and 33 Hz without aliasing,which realizes separate separation of the interference signals of different frequency bands of the aliased signal.