基于Sagnac传感器的周界安防入侵信号识别

Intrusion signal recognition of perimeter security based on sagnac sensor

针对光纤安防周界系统采集信号的非线性、非平稳性、间歇性等特点,给出了一种入侵振动信号分帧方法。采用嵌入维数计算方法,在保留时间序列信号的内在动力学特性条件下以确定最小分帧长度。利用短时能量和短时平均过零率双条件判断识别振动信号,再进行小波分解,比较各层小波系数能量分布特征,通过二级判定识别降低入侵振动信号误识率。实验结果表明:该方法可行有效,能准确判定识别入侵振动信号。

Aiming at characteristics of nonlinear, nonstationary and intermittent of signal acquired by fiber security perimeter system, give a framing method for invasive vibrant-signal. Embedding dimension number calculation method is used to determine the minimum framing length, in condition of keeping inherent dynamics characteristics of time sequence signal. Use two conditions ,which are short-time energy and short-time average zero-crossing rate to judge and recognize vibration signal. After wavelet decomposition, compare energy distribution character of wavelet coefficients of each layer, reduce rate of error recognition of intrusion vibration signal through two levels of identification. Experimental results show that, the method is feasible and effective, it can accurately identify invasion vibration signal.