Addressing the issues of poor performance and low single-seed rate during seed filling and cleaning for corn, an air-suction roller precision dibbler was designed, and its main structure and working principle were elaborated. Based on the structure of the seed metering disk and the tri-axial dimensions of corn seeds, an auxiliary filling type hole boss was designed with a radius of 4 mm and a thickness of 1.5 mm. A sawtooth-type seed cleaning device was also designed, with an installation angle greater than 3° and a serrated chamfer less than 62.1°. By coupling the Discrete Element Method (EDEM) with Computational Fluid Dynamics (CFD), the flow field structure and seed metering performance were simulated and analyzed, examining the movement states of the seed population and individual seeds during the processes of filling, cleaning, carrying, and initial seed dropping. The variation patterns of the internal flow field pressure and fluid velocity of the seed metering device under different negative pressures and operating speeds were studied. Through the construction of a test bench and single-factor and three-factor orthogonal experiments, along with multi-objective optimization methods, the primary and secondary factors affecting the qualification index were determined. The optimization results showed that when the flow field negative pressure was 3.52 kPa, the operating speed was 5.81 km/h, and the suction hole diameter was 4.15 mm, the qualification index of the seed metering device was 93.2%, the missed seeding index was 3.3%, and the multiple seeding index was 3.5%. An adaptability test was conducted to verify these results, revealing that the qualification index for all three varieties of corn seeds exceeded 92.4%, the missed seeding index was less than 3.7%, and the multiple seeding index was less than 3.9%, satisfying the precision seeding requirements for corn seeds.

As important factors in discrete elements, the physical parameters of watermelon seeds play a pivotal role in discrete element method. To obtain the discrete element parameters of watermelon seeds and improve the accuracy of the discrete element model, through a combination of actual and simulation tests, this research has calibrated the seeds’ physical and contact parameters with the seed metering device. Employing the Plackett-Burman experiment, this study has identified three critical factors affecting the stacking angle: the static and rolling friction coefficients between seeds, and the collision recovery coefficient between seeds and plexiglass. Using the steepest-climbing design and Box-Behnken response surface analysis, this research has optimized these factors to values of 0.716, 0.051, and 0.787, achieving a calibration error of just 2.60%. Verification with an air suction precision seed metering device confirmed the parameters’ accuracy, with relative errors below 7.65%. The discrepancy between the simulation and actual test results, as measured by the qualified index error, is successfully reduced to below 4.38%. This study thus establishes a solid foundation for the structural optimization of air suction precision watermelon seed metering devices.