In response to the issue of low efficiency in the classification operation of air ducts in classifier machines, this study investigates the flow field characteristics during the operation of the air duct in cleaning machine and optimizes the design of the structure. This article employs FLUENT simulation software to analyze various optimization schemes for the air duct of the machine. Through an analysis of the overall state of the flow field and the flow characteristics at characteristic cross-sections, Scheme 1 is determined. By reducing the rear baffle by 60 mm to reduce the size of the suction port, the flow field characteristics are more refined; Compared to the original scheme, the convergence of velocity vectors within the vertical duct is improved, focusing on the main working area, which is the primary suction area for the grain mixture within the vertical duct; At the suction port, the velocity meets the requirements, with the mean velocity in the main working area reaching 4.63 m/s. This velocity falls between the mean velocities of impurities and suspended grains, facilitating grain stratification. Additionally, the flow field at this position exhibits good stability, experiencing minimal turbulence effects. Therefore, Scheme 1 has been proven effective through analysis.And through the validation test, it is concluded that the impurity rate of rice-hybrid mixture after the whole machine cleaning under the conditions of Scheme 1 is 0.887%, and the test results meet the design requirements, which can ensure that the cleaning quality of vibrating air-selected grain cleaning machine meets and exceeds the agronomic standards.