Abstract: The optimization mechanism and deficiency of AFSA were analyzed, and an improved AFSA using Opposition-based Learning(OBL) and Tent chaotic map was proposed. Choosing some elited individuals exeute with OBL to guid the search space to approach the space in which the global optimum is included.This mechanism is helpful to get tradeoff between exploration and exploitation ability of AFSA.When the diversity of population descend to a limen, preserving some elites others excute Tent map variation. Simulation results for 4 benchmark functions show that the proposed algorithm has higher precision and global optimization ability than AFSA.