The accuracy of discrete element simulation is determined by simulation parameters. In this research, the parameters of coated maize particles were calibrated and verified using the discrete element method. The optimization object was to take the angle of repose at 27.05°, as determined by the physical test. The optimal contact parameters combination based on the discrete element method was obtained using the Plackett–Burman design, steepest ascent experiment, and Box–Behnken test methods. According to the findings, the static friction coefficient of maize particle–maize particle was 0.237, the rolling friction coefficient of maize particle–maize particle was 0.029, and the rolling friction coefficient of maize particle–aluminum alloy was 0.103. The verification test revealed a relative angle of repose inaccuracy of 0.22% between simulation and physical values. It was possible to get the velocity variation laws, and stability values of maize particles that were consistent with the physical situation. The variation tendency of values was compatible with the physical test when the velocity of the coated maize particles was stable, demonstrating the correctness and dependability of the calibrated parameters.

To determine the desirable bonding parameters of the soybean-bonded particle model for accidentally simulating the working process of a pneumatic soybean seed-metering device. Taking the compressive destructive force (F_c,p) derived from the uniaxial compression test of soybean seeds as the evaluation index for the compression simulation tests. The Plackett–Burman and the steepest ascent tests were executed to identify the centroids of the influential factors that substantially affect the bonding force of the soybean-bonded particle model. The optimal values of the significance influencing variables were determined based on the Box–Behnken response surface test. The results indicated that the effect of bonded disk radius (R_B,p) between fraction particles on the F_c,p was extremely significant, and the effects of the restitution coefficient (e_p-steel) and static friction coefficient (μ_p-steel) of soybean-steel, normal stiffness per unit area (k_n,p) and critical normal stress (σ_max,p) were found to be statistically significant. The preferred values identified by Box–Behnken response surface test were 0.520 for e_p-steel, 0.274 for μ_p-steel, 4.082 × 10⁷ N/m³ for k_n,p, 3.517 × 10⁵ Pa for σ_max,p, and 0.982 mm for R_B,p, respectively. The compressive destructive force of soybean seeds was 211.32 N at this point, which was 0.2% less than the measured value of 211.74 N. The results of comparing the grain morphologies during the actual and simulated compressions indicated that the compression states had a superior consistency. It was determined that the DEM simulation input parameters for the soybean-bonded particle model calibrated were proven to be effective and dependable. The investigation presented in this paper can be utilized to effectively analyze the working process of the pneumatic soybean seed-metering devices through coupled simulation. It can also serve as a reference for other researchers to construct a particle model for DEM simulation using the BPM approach.

Aiming at the problem of lower single granularity and uniformity of the existing air-suction seed-metering device under the condition of densely planted crops according to the agronomic requirements of soybean-maize strip intercropping mode, the disturbing effect of seeds on the seed-filling performance was studied to improve the seeding performance of the seed-metering device under the condition of high speed. Based on the theory of discrete element method (DEM), taking the air-suction seed-metering device designed for both soybean and maize as a model and maize variety of ‘Zhengdan 958’ suitable for densely planted as the research object, the DEM single-factor test was conducted on the influence of the structure, position, and number of bosses on disturbing performance with the average kinetic energy, average velocity, and average normal force of seeds as evaluation indexes. The results showed that the disturbing performance of seeds was better when the boss structure was D-type, the diameter of the base circle where bosses located was 160 mm, and the number of bosses was 13. To further improve the structural parameters of the D-type boss, the primary structural parameters, such as the side length and arc radius, were simulated by a single-factor test. The position, side length, and arc radius of the D-type boss were further optimized and verified using central composite design (CCD) based on the results of the single-factor test. The CCD results showed that the average kinetic energy of seeds was 7.39 × 10⁻⁷ J, the average velocity was 4.53 × 10⁻² m/s, and the average normal force was 6.18 × 10⁻² N when the diameter of the base circle where bosses located at 159.865 mm, the side length and arc radius of the D-type boss of 5.690 mm and 5.476 mm, which significantly enhanced the disturbing performance of the seed-metering device. The optimized boss structures and parameters of the seed-metering device were validated with an all-factor test of working speed and working pressure. When the working speed and working pressure were 4 ~ 6 km/h and 7 ~ 7.5 kPa, the qualified rate was 94.75% ~ 97.49%, the multiple rate was 0.92% ~ 1.43%, and the leakage rate was 1.59% ~ 3.82%, all of which were substantially lower than the original one. Therefore, the seed-filling performance of the seed-metering device will be improved by increasing the disturbance of seeds to a certain extent.

To gain the corn kernel’s bonded particle model and the optimal bonding parameters during harvest for later constructing the discrete meta-model of the integral corn ear, that may be applied for threshing simulation and analysis. Taking the ventral and lateral compressive destructive forces of the large flat kernel as evaluation indexes, the constructed large flat kernel bonded particle model was put through simulated compression tests by introducing the Plackett–Burman and steepest ascent tests to screen out the factors and their centroids with significant effects on the bonding force of corn kernels. The Box-Behnken response surface test was conducted to identify the optimal relevance factor values. The results revealed that the effect of bonded disk radius (R_B) on the large flat kernel’s ventral compressive destructive force was extremely significant, and that of shear stiffness per unit area (k_s) was generally significant. Each saliency variable had an impact on the large flat kernel’s lateral compressive destructive force, in descending order: R_B, k_s, normal stiffness per unit area (k_n), and shear modulus (G_p). The response surface test revealed that the preferred materiality factors were 2.935 × 10⁸ Pa for G_p, 4.069 × 10⁷ N/m³ for k_n, 3.147 × 10⁷ N/m³ for k_s, and 1.036 mm for R_B. On this occasion, the large flat kernel’s simulated ventral and lateral compressive destructive forces were 325.16 N and 114.94 N, with an error of 0.40% and 0.85% from the measured values. A comparison of particle morphologies during simulated and actual compression revealed that the large flat kernel’s ventral and lateral compression states were highly consistent. Simulations of large spherical kernel’s compression with the optimal parameters comprehensively verified the accuracy of the corn kernel bonded particle model constructed, as well as the calibrated simulation input parameters. The investigations of this study could provide a reliable theoretical foundation for the later construction of corn ear DEM models to simulate the threshing process or research into the crushing problem of corn kernels.