Abstract: Aiming at the shortcomings of the marine predator algorithm, such as low accuracy and slow convergence, a multi-subgroup improved marine predator algorithm is proposed. According to different fitness values, the marine predator population is divided into three subgroups: leader, follower and tailer. The leader subgroup keeps its position unchanged, the follower subgroup undergoes Gaussian mutation, and the tail subgroup is generated by the global optimal position and the average position vector. Classical benchmark functions of different dimensions are used to evaluate the efficiency of the improved marine predator algorithm. The results show that the improved marine predator algorithm has better optimization accuracy and stability.