改进Chirplet时频原子的非线性调频信号分解

Non-linear frequency modulated signal decomposition based on the improved chirplet time-frequency atom

针对Chirplet时频原子对正弦类型的非线性调频信号的分解性能较弱的问题,提出一种改进的Chirplet时频原子.首先从理论上分析Chirplet时频原子对正弦类型的时频分布的失配问题,其次在Chirplet时频原子中加入正弦调频因子,使原子的时频曲线产生类正弦形状的弯曲性能,最后使用遗传算法代替匹配追踪算法,提高原子搜索效率.仿真实验结果表明,改进的Chirplet原子对于非线性调频信号的分解性能与Gabor、Chirplet和FMmlet原子相比,有了较大的提升.

Regarding the deficiencies that the Chirplet time-frequency atom has weak decomposition performance to non-linear frequency modulation signal of a sine type, an improved Chirplet atom （IChirplet） is proposed. First, it is analyzed that the Chirplet atom has a mismatch with time-frequency distribution of＂the non-linear signal. Then, a sine frequency modulation factor is introduced into the Chirplet atom to make the atom time-freqency curve have sine similarity bending performance. Finally, the matching pursuit algorithm is replaced by the genetic algorithm to improve the efficiency of atom searching. Simulation results show that the improved Chirplet atom has better decomposition performance to the non- linear frequency modulated signal compared with the Gabor, Chirplet, and FM^mlet atom.