Abstract: Sparse decomposition can represent signal with small number of atoms. But, the high computational complexity hinders its practical application. Although Fruit Fly Optimization Algorithm （FOA） improves the efficiency of atoms searching, the solution may not be global optimum sometimes. Orthogonal Matching Pursuit （OMP） increases the convergence speed of sparse decomposition, but its computational complexity is also increased. In order to decrease the computational complexity of OMP and promote the searching capability of optimal atom, Adaptive Chaos Fruit Fly Optimization Algorithm （ACFOA） is employed in OMP. And the flavor concentration decision value and chaotic mapping function of ACFOA are also improved. The experimental results show that, compared with several other algorithms, the improved ACFOA-OMP algorithm performs best in terms of the Mean Square Error （MSE） of the reconstructed signal.

Key words: sparse decomposition, Orthogonal Matching Pursuit（OMP）, Adaptive Chaos Fruit Fly Optimization Algorithm（ACFOA）, computational complexity, global optimum

摘要： 稀疏分解能用少数原子表示原始信号，但运算复杂是阻碍其实际应用的一个重要原因。果蝇优化算法（FOA）能有效地提高稀疏分解中原子的搜索效率，但其易于陷入局部最优。自适应混沌果蝇优化算法（ACFOA）能够针对局部最优进行混沌操作，提高全局寻优性能。正交匹配追踪（OMP）通过对已选原子的正交化，能够增加稀疏分解的收敛速度，但计算复杂度却有所增加。因此，利用智能算法的并行性，将ACFOA应用于OMP，并对其味道浓度判定值和混沌映射函数进行改进，以降低整个算法的复杂度，提升最优原子的搜索性能。实验结果表明，相比于其他几种算法，改进的ACFOA-OMP算法重建信号均方误差是最佳的。

关键词: 稀疏分解, 正交匹配追踪（OMP）, 自适应混沌果蝇优化算法（ACFOA）, 计算复杂度, 全局最优