A multi-strategy improved aquila optimizer (MIAO) was proposed aiming at the disadvantages of low search efficiency and easy to fall into local optimal value in aquila optimization algorithm. Generalized normal distribution optimizer (GNDO) was added to AO, and the result obtained by GNDO was compared with the result of AO in the first stage. Then the best value under the two algorithms was selected. The search space was expanded and the quality of solution was improved. Phasor operators were used to transform the second phase into an adaptive non-parametric optimization in order to improve the high-dimensional optimization ability of the AO. The flow operator was used in the third stage of AO aiming at the problems of reduced population diversity and insufficient local exploitation at late iterations of the AO. Then the information can be transferred between each individual. The utilization rate of population information was improved, and the local exploitation capability of the AO was enhanced. Comparative analysis and optimization results of 16 test functions and Wilcoxon rank sum test showed that MIAO optimization ability and convergence speed were greatly improved. The MIAO algorithm was used to solve reducer design problem in order to verify the practicality and feasibility of MIAO algorithm. Comparative analysis of practical engineering optimization problems shows that MIAO algorithm has certain advantages in processing realistic optimization problems.