摘要
针对噪声背景下连续型、Chirp型、突发型以及随机型4类引力波信号的分类识别问题,本文提出一种基于图域变换及图连通分量个数特征的识别算法。首先,提取待识别信号功率谱生成图的连通分量个数特征,分离出连续型引力波信号;其次,通过计算待识别信号自相关函数生成图的连通分量个数,分离出随机型引力波信号;最后,借助时滞积频谱模值生成图的连通分量个数特征,区分Chirp型与突发型引力波信号。仿真结果表明,该方法在信噪比不低于-4 dB时,平均正确识别概率可达到90%以上。
A recognition algorithm based on graph domain transformation and the number of connected components of the graph is proposed for four types of gravitational wave signals in noisy backgrounds,namely continuous,chirp,burst,and random signals.Firstly,the number of connected components of the graph is extracted from the power spectrum of the signal to identify continuous gravitational waves.Secondly,the number of connected components of the graph is calculated from the autocorrelation function of the remaining signal to separate random gravitational waves.Finally,the number of connected components of the graph is obtained from the time-delayed spectrogram of the signal to differentiate between chirps and burst gravitational waves.Simulation results show that this method achieves an average correct recognition probability of over 90%when the signal-to-noise ratio is not lower than-4dB.
作者
吴珊珊
胡国兵
Wu Shanshan;Hu Guobing(School of Electronic Information,Nanjing Vocational College of Information Technology,Nanjing 210023,China;School of Electronic and Information Engineering,Jingling Institute Technology,Nanjing 211169,China)
出处
《信息化研究》
2023年第4期7-16,共10页
INFORMATIZATION RESEARCH
基金
南京信息职业技术学院自然科研基金项目(No.YK20200101)
关键词
引力波信号
图域变换
图连通分量个数
信号识别
gravitational wave signals
graph domain transformation
number of connected components of the graph
signal recognition