信号源识别的相干函数法

Coherence Functions Method for Signal Source Identification

实际工程中所采集的多个信号往往不满足独立性,而且独立信号源的个数也常常是未知的,针对此问题,提出一种基于相干函数分析的振动信号源识别方法。该方法可用于独立、非独立以及未知独立的信号源识别。对于检测到的振动信号,用虚相干函数中的虚输入矩阵确定信号源独立个数,并以此判断信号源是否独立。对于非独立信号源,提出一种优先级排序的滤波器法。在进行优先级排序后,用重相干函数检测是否有重要信号源被遗漏,然后分别用常相干函数和偏相干函数对独立信号源和非独立信号源进行识别。随机信号的仿真试验说明,基于相干函数分析的振动信号源识别方法对信号源的识别具有满意的效果。

In practical engineering, several signals collected do not often suffice to independence, and the number of independent signal sources is also unknown. The paper introduced a method on the basis of coherence function analysis about vibration signal sources. For the detected signals, virtual input matrix in virtual coherence function was used to determine the number of independent signal sources and independence among them. For dependent signal sources, a filter method for priority ordering was developed. After the priority ordering, multiple coherence functions were adopted to detect whether important signal sources were lost, then independent and dependent signal sources were distinguished respectively by ordinary coherence function and partial coherence function. The simulation testing of random signals suggests that the method is effective for signal source identification.