摘要
实际工程中所采集的多个激励源信号之间往往存在着复杂的相关性,针对此问题,提出了一种时延平均白化的去相关方法。该方法借鉴了联合对角化思想,对白化信号的多个非零延时相关矩阵进行了平均,提取了正交解相关分离矩阵,完成了信号的去相关分离。在阐述理论依据的基础上,给出了时延平均白化去相关方法的详细分析流程。仿真信号以及实车振动信号的分析,说明了该方法对于具有复杂相关性的振动源信号具有满意的去相关分离效果。
In the real engineering, complex correlation usually existed among multiple acquired excitation source sig- nals. In order to solve this problem, a new de-correlation method based on the average delay whiting is proposed to de-correlate these signals. In this method, several correlation matrixes with different non-zero delays are averaged based on the theory of joint diagonalization, and then the orthogonal matrix to de-correlate the signals is extracted to achieve the separation of correlated signals. A detailed analysis process is presented by elaborating the basic theory of this method. The analysis of simulated signals and vibration signals of real vehicle shows that this method has satisfied separation result for complex correlated vibration source signals.
出处
《机械科学与技术》
CSCD
北大核心
2014年第4期546-550,共5页
Mechanical Science and Technology for Aerospace Engineering
关键词
相关
时延
白化
振动信号
computer simulation
correlation
delay
experiments
matrix algebra
sensors
singular value decom- position
time delay
vibration analysis
vibration signals
vibrations （mechanical）
whiting
