中文摘要：

为构建合适的时频原子库和信号分解算法,提出一种基于改进遗传算法和Sin-Chirplet原子的调频雷达信号稀疏分解算法.这种新的Sin-Chirplet原子在Chirplet原子的基础上增加正弦调频因子,改善原子时频曲线的弯曲性能,使原子对非线性调频信号具有较强的匹配性能.然后基于原子的匹配特性,改进遗传算法中初始原子种群产生机制,提高了最佳原子搜索速度.理论分析和仿真结果表明,基于改进遗传算法的信号稀疏分解效率高于传统遗传算法和匹配追踪算法.相比现有的3种典型时频原子,Sin-Chirplet原子的匹配性能良好,可以更有效地分解调频雷达信号及其混合信号.

英文摘要：

In order to improve the performance of sparse decomposition for radar signals,a signal sparse decomposition algorithm（SDA）based on a novel Sin-Chirplet atomic dictionary and an advanced genetic algorithm（GA）is proposed.Firstly,a sinusoidal frequency modulating factor is added into Chirplet atom so that the curvature performance of the atom time-frequency curve can be improved.Therefore Sin-Chirplet atom has good matching performance for non-linear frequency modulation signals.Secondly,based on the matching characteristic of Sin-Chirplet atom,the generation of atom initialization population in GA is improved so that the best-atom searching speed can be raised.The theoretical analysis and the simulation results show that the decomposition efficiency of advanced GA is higher than that of GA and matching pursuit（MP）algorithm.Meanwhile,the matching performance and the decomposition efficiency of Sin-Chirplet atom are better than those of other three traditional time-frequency atoms.