基于Chirplet原子的雷达辐射源信号特征提取

FEATURE EXTRACTION OF RADAR EMITTER SIGNALSBASED ON GAUSSIAN CHIRPLET ATOMS

特征提取是新体制雷达辐射源信号分选识别的关键技术.本文提出一种全新的雷达辐射源信号时频原子特征提取方法.在过完备多尺度Chirplet原子库基础上,采用匹配追踪(MP)方法对信号进行时频原子分解,并通过改进量子遗传算法(IQGA)降低MP搜索过程的时间复杂性,得到表示雷达辐射源信号特征信息的本征Chirplet原子.实验结果表明使用更少量的Chirplet原子可以得到比Gabor原子分解更准确的特征信息,证实了本文方法的可行性和有效性.

Feature extraction is a crucial technology in advanced radar emitter signal deinterleaving and recognition. A time-frequency atom approach to extract the features of radar emitter signals was presented in this study. Based on the overcomplete multiscale dictionary of Gaussian Chirplet atoms, the signals were decomposed into a linear expansion of atoms by the method of marching pursuit （MP）. Then, the improved quantum genetic algorithm was applied to effectively reduce the time-complexity at each search step of MP, and thus some intrinsic Chirplet atoms describing features of signals were obtained. Experiment results show that the Chirplet atom is better than the Gabor atom in extracting the feature parameters, which confirms the validity and feasibility of the approach.