Engineering Discrete Element Method-Based Design and Optimization of the Key Components of a Spoon-Wheel Spinach Seed-Metering Device

Abstract

In view of issues affecting manual on-demand sowing of spinach, such as high working intensity, low seeding efficiency, and high cost, a mechanized precision seeding device for spinach seeds is proposed, based on a spoon-wheel including a combined cell with two intersecting surfaces. The key structure was designed, and motion analysis was conducted employing SolidWorks software (2022). Additionally, the seeding process was simulated using EDEM simulation software (2022). The number of spoons, the radius of the spoons, and the speed of the seed wheel were chosen as the influencing factors, while the qualified index, reseeding index, and missing index were utilized as the evaluation indexes. A response surface method test based on the Box–Behnken design was carried out, and the test results were analyzed to obtain the optimal parameter combination for the seed-metering device. The field test results showed that when the rotation speed was 18 rpm, the radius of each seed spoon was 2.5 mm, and the number of seed spoons was 50, the average qualified index was 90.89%, the reseeding index was 8.22%, and the missed seeding index was 0.89%. The test results satisfy the technical production requirements for spinach seed sowing.

1. Introduction

Spinach has a high economic value, and it can be planted both in the north and the south [1,2]. The planting area of spinach in China exceeds 600,000 hm², accounting for about 5% of the vegetable market [3]. With rapid growth, a short cycle, a high replanting index, small investment, high yield, and simple management techniques, spinach is a leafy green vegetable that can be supplied annually and economically and is favored by the majority of vegetable producers and consumers [4,5,6].

High-quality sowing can stimulate the latent vitality of seeds, prompting them to germinate and take root rapidly, providing a powerful guarantee for the subsequent growth stages. Thus, it can ensure the smooth progress of the entire planting process and ultimately lead to a bountiful harvest. Spinach cultivation mainly adopts the direct seeding method, and there are three sowing methods: scattering, strip sowing, and hole sowing. The seeding quantity for scattering is 60–75 kg/hm², that for strip sowing is 45–60 kg/hm², and for hole sowing, it is 1–2 seeds per hole [7]. Manual broadcasting is a common traditional sowing method. Not only is this approach highly labor intensive, consuming a large amount of manpower; also, it is difficult to ensure the uniformity of seed sowing. The mechanized sowing index of spinach is low. Only in a few places such as Inner Mongolia (between 113°21′09″ E–114°07′47″ E, 40°26′ N–41°26′ N) and Hebei (Between 114°10’ E–115°27’ E, 40°57’ N–41°34’ N), where sowing is carried out in April and May, is the operational quality of precision direct sowing machinery for some vegetables better [8].

In developed countries, the seeding technology and equipment are more mature. Precision planters mostly employ air-absorbing seed-metering devices, and through the replacement of the seed trays, precision sowing of various vegetables such as spinach can be achieved [9,10]. Taking the single-grain precision seeders produced by Italy’s Maschio company [11], Agricola Italiana company [12] and the United States’ Monosem company [13] as representatives, these seeders can be adapted to the precision sowing of various vegetables with small and medium-sized seeds. They have a faster operating rotational speed and stable performance, but they require supporting power and have a high price and are not adapted to the vegetable planting model in our country. Based on foreign advanced technology, several types of vegetable precision seeding machines have been developed in China. Most of them employ nesting eye wheel type seed-metering devices, and it is inevitable that phenomena including missing and the reseeding of the seed extractor are encountred. However, some pneumatic seed metering devices that have been developed not only require additional fans but also have complex structures and are difficult to maintain, making them difficult to promote. Wang et al. [14] developed a hand-pushed vegetable seeding machine to achieve the planting of vegetables in greenhouses, replacing the seed-metering device wheels with different types of holes. However, the sowing uniformity was poor due to the influence of vibration during the sowing process. Yu Zhenjun et al. [15] carried out orthogonal experiments on the 2BS-4 electric vegetable seeder. However, the experiment indicated that the rotational speed of the nested-eye barrel-shaped seed discharge should not be too fast; otherwise, it would result in uneven seed discharge and seedling emergence, an increase in the index of seedling breakage, and attenuation of yield. Li Jinwen et al. [16] studied the seed meter of the Maschio pneumatic vegetable seeder, but the imported equipment is costly and not suitable for promotion. Yan Jianwei et al. [17] designed an air-suction white radish seeder. Its sowing speed and single grain index are high, but the structure is complex and the operation is cumbersome. Currently, farmers’ acceptance of its operation is not high.

In view of the current problems of low precision, serious seed leakage, and inconvenient operation during the artificial planting of spinach, a mechanical spoon-wheel spinach sowing device has been designed to improve the efficiency of vegetable sowing. The structure, which includes two intersecting surfaces, solves the problems of seed collection and seed collection accuracy.

This paper introduces the working principle of the seeding device and assesses the corresponding parameters. The seeding process and the structural parameters of the seeding spoon were simulated and analyzed by EDEM (Engineering Discrete Element Method), and the Box–Behnken center test method was used for simulation test. The optimal parameter combination was calculated by Design-Expert 13 software, and verified by bench test and field test, aiming to provide ideas for the design of the spinach seeder.

2. Materials and Methods

2.1. The Overall Structure and Working Principle of the Spoon-Wheel Spinach Seed-Metering Device

2.1.1. Structure of the Spoon-Wheel Spinach Seed-Metering Device

The overall assembly model of the seed meter is shown in Figure 1a,b is an exploded view. The spoon-wheel spinach precision seed-metering device is composed of a front shell, a seed-picking wheel, an adjustment baffle, a seed-carrying wheel, and a rear shell. Among these, the seed-picking wheel and the seed-carrying wheel are the core working parts of the device. The above materials are all 3D printing materials (Acrylonitrile Butadiene Styrene, Future Factory Co. Ltd., Shanghai, China).

2.1.2. Working Principle of the Spoon-Wheel Spinach Seed-Metering Device

The seeding process is mainly categorized into four stages: filling, cleaning, carrying, and dropping [18], as shown in Figure 2. Spinach seeds are loaded from the lid of the device and enter the seed cleaning area by rotating under the action of the seed-picking wheel. The seed-taking wheel adopts a unique cell design instead of the current spoon-like design. On the one hand, the cell of the seed-picking wheel can fulfill the function of seed-picking; on the other hand, it can also disturb the contents and enhance the seed filling index. The brush can clear away the excess seeds and realize the precision sowing of the seeds. The seed-carrying wheel consists of multiple seed grooves. Due to the effect of gravity, spinach seeds slide into the seed-carrying wheel along the dredging hole to meet the requirement of precision seeding in a single seed groove. Compared with existing cell-wheel seed-metering devices, this reduces damage to seeds more effectively. Most existing seed-metering devices are of the cell-wheel type. These devices utilize the speed difference between the seeds and the cell-wheel to roll the seeds into the holes for seeding. However, such devices can cause significant damage to the seeds and are not suitable for sowing large-grained seeds.

The seed-picking wheel and the seed-carrying wheel are fixed together, and each guiding hole corresponds to a seed groove. The adjusting baffle adjusts the seed-carrying interval between the spinach seeds by coordinating with the position of the hole on the shell. When the holes with the seeds leave the seed-filling area, the excess seeds cleared by the brush fall back to the seed-filling area due to their own gravity. When the seed enters the seed groove of the seed-carrying area and turns to the dropping area with the rotation of the seed-carrying wheel, the spinach seed loses the enclosure of the shell and falls into the seed groove under the action of its own gravity and centrifugal force, thereby completing the spinach dibbling operation.

2.2. Design of Key Structural Parameters

2.2.1. Seed-Carrying Wheel Diameter Design

The diameter of the seed-carrying wheel plays a crucial role in determining structural parameters such as the number of spoons, the overall size, the size of the spoons, and the linear velocity of the seed. The diameter of the seed-carrying wheel is generally within the range of 140 to 260 mm [19]. Due to the small particle size of spinach seeds and the overall design of the spinach seed-metering device, 214 mm was chosen.

2.2.2. Design of the Structural Parameters of the Seed Scoop

The spoons are the core operating components of the spinach precision seeder seed-picking mechanism. Their shape and structural parameters have a direct influence on the seed-picking performance. The design of the seed spoon not only needs to ensure that the seeds can be scooped up by the spoons in any attitude during the filling process, but also needs to ensure that the shape of the spoons cannot accommodate excess seeds. The shape of the seed spoon is such that the excess seeds can rely on their own gravity to leave the seed spoon during the seed cleaning process. In order to meet the requirements of seed-picking and seed-clearing, the scoops were designed as a combination of two coherent surfaces. In fact, the design includes two inclined column surfaces coherent with the seed-picking wheel, and the intersecting part has been removed to form the scoops.

The structural dimensions of the scoops were designed as shown in Figure 3.

R₁ represents the cylindrical radius intercepted on the return side of the front side of the seed-picking wheel. α indicates the angle between the cylindrical center axis a and the front side of the seed-picking wheel. R₂ stands for the cylindrical radius intercepted on the opposite side of the seed-picking wheel. β is the angle between the cylindrical center axis b and the opposite side of the seed-picking wheel, which is designed to be 45°. γ is the angle between the cylindrical center axis b and the y axis, and it is also designed to be 45°. G is the gravity of the seed. F_z is the support force of the slope on the seed. F_c is the friction force of the inclined plane on the seed.

2.2.3. Determination of the Size of the Seed Spoons

Spinach seed has a medium particle size, is irregularly round, has no thorns, and its surface is smooth. The three-axis size has the greatest influence on the design of the size of the holes. This study considered Royal Dark Green spinach seed as the research object and randomly selected 100 full spinach seeds after initial sieving for measurement. Their length ranged from 2.19 to 4.22 mm, the width ranged from 2.25 to 3.81 mm, and the thickness ranged from 1.47 to 2.71 mm. The average length of the spinach seeds was 3.19 mm, the average width was 2.87 mm, and the average thickness was 1.95 mm.

The formula [20] for calculating the sphericity of the spinach seeds is shown in Equation (1):

$$S_P={\displaystyle \frac{\sqrt[3]{LWH}}{L}}\times 100\%$$

(1)

where the L is the length of the seed, mm. W is the width of the seed, mm. H is the thickness of the seed, mm.

The design of the spoons needs take into account the three-axis dimensions and sphericity of the spinach seeds. Based on the measured three-dimensional average sizes, it was calculated that the sphericity of the spinach seeds amounted to 81.39%.

The filling side of the seed-picking wheel is considered the reverse side. It has been ensured that the size of the spoon can accommodate only 1 to 2 seeds during the filling process. It is required that the minimum radius of the intercepted cylinder be greater than the maximum radius of the spinach seeds and less than the sum of the radii of two spinach seeds, that is, 3R > R₂ > R. According to the statistical data, the average triaxial sizes of spinach seeds were 3.19 mm, 2.87 mm, and 1.95 mm respectively. To ensure the seeding accuracy, R₂ was tested at 2 mm, 2.25 mm, 2.5 mm, 2.75 mm, and 3 mm respectively.

2.2.4. Angle of Inclination of the Seed-Picking Wheel for the Seed Scoop Holes

In the design of the seed-collecting side, an inclined angle β exists between the spoon and the vertical plane of the clapboard. The aim is to guarantee that the seed can slide to the seed delivery port by its own gravity, without remaining in the holes. This demands that the gravity component of the seed in the θ₁ direction of the inclination angle of the R₂ cylinder be greater than the friction force between the seed and the spoon hole; that is, the value of the minimum angle (θ₁) can be obtained according to Equation (2):

$$\arctan \mu ={\theta }_1$$

(2)

where μ spoon represents the static friction coefficient between spoons. In fact, spinach seeds need to overcome the static friction force first when moving, so the static friction force is included in the actual calculation. In the test, the static friction coefficient of the spinach and ABS resin was 0.464. By calculation, when the seed wheel material is ABS resin (acrylonitrile butadiene styrene), the static friction angle between the seed and the material is 24.9°. Ensuring that the seeds can quickly slide to the seed delivery port, the value of β is designed to be 45°.

2.2.5. Inclination of the Spoon Hole on the Reflux Side of the Seed

In the design of the reflux side on the front of the seed-picking wheel, after the seeds leave the seed-filling area, they make circular motion along with the seed-picking wheel. When the spoons pass through the seed-cleaning area, except for the seeds in the holes, the other seeds slide back to the seed-filling position along the holes on the reflux side of the front of the seed-picking wheel under the action of gravity and centrifugal force, as shown in Figure 3. To ensure that the seeds can fall back to the seed-filling area under the action of their own gravity, force analysis was conducted for the seeds on the reflux side of the holes, as shown in Equation (3).

$$\left\{\begin{array}{c}Mg\mathrm{c}\mathrm{o}\mathrm{s}\theta =F\mathrm{z}\\ F\mathrm{c}=F\mathrm{z}\mathrm{t}\mathrm{a}\mathrm{n}\theta \\ \mathsf{\&}\mathsf{\theta }\ge \mathsf{\phi }\\ ma=mg\mathrm{s}\mathrm{i}\mathrm{n}\theta -F\mathrm{c}\end{array}\right.$$

(3)

where θ is the return-side hole inclination angle, (°); φ is the maximum friction angle between spinach seed and seed-picking wheel material, (°); (N). M is the seed quality.

After measurement, the static friction angle between the seeds and the material of the seed-picking wheel was found to be 24.9°, and θ was determined to be 20° according to the structure of the seed discharger and the pre-test. The angle θ between the axis a and the x axis will increase with the rotation of the seed-picking wheel to meet the design requirements of the return side of the hole.

2.3. Simulation Model and Parameter Setting

The tests comprised three parts. Based on the designed experimental scheme, the structural parameters of the device were optimized by means of a combination of simulation and regression analysis. Simulation tests were conducted with the qualified index, reseeding index, and missed seeding index as evaluation indicators [21,22]. Each group of tests was repeated three times. The test results were recorded and the average values calculated. In the first part, EDEM was utilized to conduct single-factor tests on the number of spoons, the rotational speed of the seed-picking wheels, and the radius of the spoons, to determine the reasonable range of each factor. In the second part, a Box–Behnken test with three factors and three levels was conducted for the above three factors, the results were analyzed, and the regression equation was established [23]. The optimal working parameters of the device were ascertained according to this equation. In the third part, based on the above optimal parameter solution, the corresponding seeding device was processed for bench test and field test, and the corresponding results were analyzed.

Based on the size of the measured spinach seeds, spinach seeds with similar length, width, and thickness were selected. A geometric model was established using Solidworks 2022 and imported into EDEM 2022R1. The simulation model of the spinach seeds was established using the automatic particle-filling function [24]. The minimum radius of the filling sphere was set at 0.3 mm, the smoothing value was 0.5, and 204 particles were filled [25]. The 3D model of the device was established using Solidworks. To improve the simulation efficiency, components that were not involved in the simulation in the 3D model were simplified or omitted. The STL (stereolithography) format file was exported and imported into EDEM, as shown in Figure 4.

The particle factory was constructed. The working surface for the seeds was placed at the entrance of the seed box. Based on the seed storage space in the seed-metering device and the seed box, the total number of generated particles was set to be 5000. Five thousand seeds were generated per minute, with all the particles being of equal size. The simulation time duration was set as 20 s. The step size for the discrete element simulation was 20%, and the output time step size was 0.01 s.

During the operation of the device, the seeds come into contact with the seed-picking wheel and the adjustment baffle. The material of the seed-picking wheel is ABS resin and that of the adjustment baffle is stainless steel. The Hertz–Mindlin (no-slip) model was used as the contact model because there is no adhesion between the spinach species and the seed-metering device [26,27,28]. The simulation parameters were set according to the results of parameter calibration before the test, as shown in

Table 1. Simulation parameters of EDEM.

Parameters	ABS Resins	Stainless Steels	Spinach Seed
Densities (kg·m−3)	1.06	7.865	1.05
Poisson’s ratio	0.39	0.30	0.35
Shear modulus/Pa	8.9 × 108	7.9 × 1010	2.95 × 107
Collision recovery coefficient	0.310	0.346	0.470
Static friction factor	0.467	0.505	0.370
Dynamic friction factor	0.045	0.047	0.040

[29,30,31].

2.4. Experimental Factors and Evaluation Indicators

The rotational speed of the seed-picking wheel n, the radius of the spoon R₂, and the number of spoons δ were selected as the influencing factors, and the qualified index y₁ and the missing index y₂ were taken as the response indicators.

Based on the analysis of the single-factor test results, the horizontal ranges of the number of spoons, the rotational speed of the seed wheel, and the spoon radius were determined. In order to study the optimal combination parameters of these three test factors, the Box–Behnken center combination test [32] was conducted. The factor level coding is shown in

Table 2. Coding of factors and levels.

Codes	Seed Wheel Speed A (rpm)	Radius of Spoon B (mm)	Number of Spoons C
−1	15	2.00	45
0	20	2.25	50
+1	25	2.50	55

.

3. Results and Discussion

3.1. Single-Factor Analysis of Simulation

3.1.1. Influence of the Number of the Seed Spoons on Seed Discharge Effect

For the pre-testing, the rotational speed of the seed-picking wheel was set at 20 rpm, the number of seeds at 5000, and the radius of the seed spoon type hole R₂ at 2.5 mm. In the simulation test, the effect on the working performance of the seed discharger was analyzed when the number of seed spoons was 40, 45, 50, 55, and 60. The average of three tests was calculated for each group, and the test results are shown in

Table 3. Simulation results with different numbers of seed spoons, %.

Numbers	No.	Qualified Index	Missing Index	Reseeding Index	Average Qualified Index	Average Missing Index	Average Reseeding Index
40	1	89.33	0	10.67	87.11	0	12.89
	2	85.33	0	14.67
	3	86.67	0	13.33
45	1	89.33	0	10.67	87.78	0	12.22
	2	87.33	0	12.67
	3	86.67	0	13.33
50	1	91.33	0.67	8.00	91.78	0.44	7.78
	2	91.33	0	8.67
	3	92.67	0.67	6.67
55	1	84.67	0.67	14.67	82.89	0.89	16.22
	2	82.67	0.67	16.67
	3	81.33	1.33	17.33
60	1	86.00	1.33	12.67	82.33	1.33	16.44
	2	81.33	1.33	17.33
	3	79.33	1.33	19.33

.

Through the test analysis, it was observed that with the increase in the number of seed spoons, the qualified index of the seed discharger initially increased and then declined, the missing index initially decreased and then increased, and the reseeding index initially decreased and then increased. When the number of seed spoons was 40, the spacing between each two neighboring seed spoons was too large, and the seed flow was unstable, which was not conducive to seed charging, resulting in the reduction of the qualified index. When the number of seed spoons was 50, the seed spoon spacing was more appropriate. With the increase in the number of spoons, the spacing between spoons became too small, leading to untimely seed filling; then, the qualified index decreased and the reseeding index increased. Considering the interaction between the factors in the subsequent experiments, the appropriate number of seed spoons was taken to be 45–55.

3.1.2. Influence of Rotational Speed on Seed Discharge Effect

The rotational speed of the seed-picking wheel had a highly significant effect on the seed filling effect of the seed discharger. To study the influence of the rotational speed on the operation of the seed discharger, for the pre-testing, the number of seed scoops was set to 50, the number of seeds to 5000, and the radius of the seed scoop R₂ to 2.5 mm. In the simulation test, the effect of the rotational speed of the seed-picking wheel on the performance of the seed discharger was analyzed at 10, 15, 20, 25, and 30 rpm. The test results are shown in

Table 4. Simulation results under different rotational seed wheel speeds, %.

Rotation Speed (rpm)	Serial No.	Qualified Index	Missing Index	Reseeding Index	Average Qualified Index	Average Missing Index	Average Reseeding Index
10	1	85.33	1.33	13.33	87.11	1.33	11.56
	2	88.67	1.33	10.00
	3	87.33	1.33	11.33
15	1	89.33	0.67	10.00	88.89	0.44	10.67
	2	88.00	0.67	11.33
	3	89.33	0	10.67
20	1	91.33	0.67	8.00	91.78	0.44	7.78
	2	91.33	0	8.67
	3	92.67	0.67	6.67
25	1	92.00	0	8.00	91.56	0.44	8.00
	2	90.00	0.67	9.33
	3	92.67	0.67	6.67
30	1	89.33	0.67	10.00	87.33	0.89	11.78
	2	86.00	0.67	13.33
	3	86.67	1.33	12.00

.

It was observed from the test results that with the increase in the rotational speed, the qualified index of the test initially increased and then decreased, the missing index initially decreased and then increased, and the reseeding index initially decreased and then increased. At a rotational speed of 20 rpm, the qualified index was the highest at 91.78%, and at a rotational speed of 10 rpm, the qualified index was the lowest and the reseeding index was higher. This indicated that when the rotational speed was too slow, seed accumulation was more likely to form between the seed scoop and the seed-picking wheel. This means that when the rotational speed was too slow, it was easier for seed to accumulate between the spoons and the seed-picking wheel. When the rotational speed was faster, the backfilling of the seed population between the two neighboring seed spoons in the seed filling area was not timely, causing the missing index to increase and leading to the reduction of the qualified index. Considering the interaction between the factors in the subsequent test, the rotational speed was set at 15–25 rpm.

3.1.3. Influence of the Radius of the Seed Scoop on Seed Discharge Effect

The size of the seed spoon hole determines the success index of seed extraction in the seed-filling process. In order to study the degree of influence of the seed spoon hole on the seed-filling performance of the seed expeller, according to the theoretical design of the seed spoon hole, during the pre-test, the number of seed spoons was set at 50 and the rotational speed at 20 rpm. In the simulation analysis, the influence of the radius of the hole—2.0 mm, 2.25 mm, 2.5 mm, 2.75 mm, or 3.0 mm—on the seed-filling performance of the seed discharger was analyzed. The test results are shown in

Table 5. Simulation results for different kinds of spoon hole radius, %.

Radius of a Spoon (mm)	Serial No.	Qualified Index	Missing Index	Reseeding Index	Average Qualified Index	Average Missing Index	Average Reseeding Index
2.0	1	94.67	0.67	4.67	95.11	0.22	4.67
	2	96.00	0	4.00
	3	94.67	0	5.33
2.25	1	92.00	1.33	6.67	92.22	1.56	6.22
	2	92.00	2.00	6.00
	3	92.67	1.33	6.00
2.5	1	91.33	0.67	8.00	91.78	0.44	7.78
	2	92.67	0.67	6.67
	3	91.33	0	8.67
2.75	1	89.33	1.33	9.33	90.00	1.33	8.67
	2	90.67	1.33	8.00
	3	90.00	1.33	8.67
3.0	1	88.67	0.67	10.67	87.11	0.89	12.00
	2	86.00	1.33	12.67
	3	86.67	0.67	12.67

.

With the increase in the radius of the seed spoon, the qualified index in the test gradually decreased and the reseeding index gradually increased. When the radius of the holes was 2 mm, the qualified index was the highest. When the radius was 3 mm, the qualified index was the lowest. This reflects that in the process of seed-picking, when the seed spoon passes through the contents of the seed-filling area, if the radius of the seed spoon hole is small, the single grain index in the seed spoon hole is higher, but leakage is easily caused. If the radius of the holes is large, with the rotation of the seed-picking wheel, the seed spoon is more likely to carry other seeds outside the holes along with the movement of the seed-picking wheel. The reseeding index will gradually increase and the qualified index will decrease. Considering the interaction between the factors in the subsequent test, the radius of the seed spoon hole was designed to be 2–2.5 mm.

3.2. Box-Behnken Design Experiment

The experimental design scheme and results are shown in

Table 6. Box–Behnken test program and results.

No.	A	B	C	Qualified Index y1 (%)	Missing Index y2 (%)
1	−1	−1	0	95.33	2.67
2	1	−1	0	94.67	3.33
3	−1	1	0	89.33	0.67
4	1	1	0	91.33	0.67
5	−1	0	−1	90.00	1.33
6	1	0	−1	89.33	1.33
7	−1	0	1	91.33	0.67
8	1	0	1	93.33	0.67
9	0	−1	−1	88.67	3.33
10	0	1	−1	87.33	0.67
11	0	−1	1	96.00	2.00
12	0	1	1	86.00	0.67
13	0	0	0	92.00	1.33
14	0	0	0	92.67	1.33
15	0	0	0	92.67	0.67
16	0	0	0	92.67	0.67
17	0	0	0	92.67	1.33

. Design Expert 13 data analysis software was used to analyze the experimental data by multiple regression fitting [33,34], and the regression equations for the qualified index y₁ and the missing index y₂ were obtained as Equation (4) and Equation (5), respectively:

$$\begin{array}{c}{y}_1=92.53+0.3333A-2.58B+1.42C+0.6667AB+0.6667AC-2.17BC\\ +0.8167A^2-0.6833B^2-2.35C^2\end{array}$$

(4)

$$\begin{array}{c}{y}_2=1.07+0.0833A-1.08B-0.3333C-0.1667AB+0.3333BC\\ +0.05A^2+0.7167B^2-0.1167C^2\end{array}$$

(5)

The variance analysis of the regression model is shown in

Table 7. Variance analysis of regression model.

Items	Source of Variation	Sun of Squares	Freedom	Mean Square	F Value	p Value
Qualified index y1	Model	120.45	9	13.38	102.82	<0.0001 **
	A-A	0.8889	1	0.8889	6.83	0.0347 *
	B-B	53.39	1	53.39	410.18	<0.0001 **
	C-C	16.06	1	16.06	123.35	<0.0001 **
	AB	1.78	1	1.78	13.66	0.0077 **
	AC	1.78	1	1.78	13.66	0.0077 **
	BC	18.78	1	18.78	144.27	<0.0001 **
	A2	2.81	1	2.81	21.58	0.0024 **
	B2	1.97	1	1.97	15.11	0.0060 **
	C2	23.25	1	23.25	178.65	<0.0001 *
	Residual	0.9111	7	0.1302
	Lack of Fit	0.5556	3	0.1852	2.08	0.2451
	Pure Error	0.3556	4	0.0889
	Cor Total	121.36	16
Missing index y2	Model	13.11	9	1.46	15.82	0.0007 **
	A-A	0.0556	1	0.0556	0.6034	0.4627
	B-B	9.39	1	9.39	101.98	<0.0001 **
	C-C	0.8889	1	0.8889	9.66	0.0171 *
	AB	0.1111	1	0.1111	1.21	0.3083
	AC	0.0000	1	0.0000	0.0000	1.0000
	BC	0.4444	1	0.4444	4.83	0.0640
	A2	0.0105	1	0.0105	0.1143	0.7452
	B2	2.16	1	2.16	23.49	0.0019 **
	C2	0.0573	1	0.0573	0.6225	0.4560
	Residual	0.6444	7	0.0921
	Lack of Fit	0.1111	3	0.0370	0.2778	0.8395
	Pure Error	0.5333	4	0.1333
	Cor Total	13.75	16

Note: * indicates significant effect (0.01 < p < 0.05), ** indicates highly significant difference (p < 0.01).

.

3.3. Analysis of Box-Behnken Design Experimental Results

In order to visualize and analyze the relationship between factors and experimental indexes, Design Expert 13 was used to draw the response surface, as shown in Figure 5 and Figure 6.

(1)

Influence of various test factors on qualified index

Figure 5a depicts the impact of the interaction between the rotational speed and the radius of the spoons on the qualified index when the number of spoons is at the center level. With the increase in the radius of the spoons, the qualified index gradually declines. When the radius of the spoons is fixed, the qualified index initially decreases and subsequently increases with the rise in rotational speed. Figure 5b presents the effect of the interaction of rotational speed and the number of spoons on the qualified index when the radius of spoons is at the center level. When the rotational speed of the spoons is constant, the qualified index gradually increases and then decreases with the increase in the number of spoons. Figure 5c illustrates the effect of the interaction between the rotational speed and the radius of the scoops on the qualified index when the number of scoops is at the center level. The qualified index decreases as the radius of the scoops increases at a certain rotational speed.

(2)

Influence of various test factors on missing index

Figure 6a demonstrates the influence of the interaction between the rotational speed and the radius of the spoons on the missing index when the number of scoops is at the center level. The missing index reduces with the decrease in the rotational speed and the increase in the radius of the spoons. Figure 6b reveals the effect of the interaction between the rotational speed and the number of spoons on the missing index when the radius of the spoons is at the center level. The missing index decreases as the number of spoons increases at a certain rotational speed. Figure 6c exhibits the impact of the interaction between rotational speed and radius on the missing index when the number of scoops is at the center level. The missing index declines with the increase in the number of spoons at a certain rotational speed. When the number of scoops remains constant, the missing index decreases with the growth of the radius of the scoops.

3.4. Parameter Optimization and Simulation Test Verification

To identify the combination of factors and parameters to enable the optimal performance of the seed-metering device, the objective functions were the highest qualified index and the lowest missing index, and the constraints were the rotational speed (A), the radius of the spoons (B), and the number of spoons (C). The objective function and constraints are shown in Equation (6):

$$\left\{\begin{array}{c}max{y}_1\\ min{y}_2\\ s.t.\left\{\begin{array}{c}15\mathrm{r}\mathrm{p}\mathrm{m}\le A\le 25\mathrm{r}\mathrm{p}\mathrm{m}\\ 2\mathrm{m}\mathrm{m}\le B\le 2.5\mathrm{m}\mathrm{m}\\ 45\text{ ≤ }C\text{ ≤ }55\end{array}\right.\end{array}\right.$$

(6)

According to the Design Expert 13 optimization solver module, the optimal performance of the seed extractor was attained when the rotational speed of the seed-picking wheel was 17.92 rpm, the radius of the seed spoons was 2.16 mm, and the number of seed spoons was 50. The seed extraction index of the qualified index was 93.45%, and the missing index was 1.49%. To verify the reliability of the optimized parameters, three repetitive tests were conducted. The seed extractor achieved the optimal performance of seed extraction with a qualified index of 91.78% and a missing index of 1.56% at a rotational speed of 18 rpm, a seed scoop radius of 2.25 mm, and 50 seed scoops.

3.5. Bench Test and Field Test Validation

3.5.1. Bench Test

In order to verify the influence of the main structural and working parameters of the spinach seed-metering device on the seeding performance, the seed scoop with the optimal parameters was fabricated by 3D printing for bench-test verification. The material was ABS resin, and the accuracy was 0.1 mm. The test was conducted on the JPS-12 seed-metering device performance test bench in the laboratory of Nanjing Institute of Agricultural Mechanization, Ministry of Agriculture and Rural Affairs, as shown in Figure 7. The test was carried out with the rotational speed of the seed wheel set to 18 rpm, while the radius of the spoons was 2.25 mm and the number of spoons was 50. A total of 150 holes were used for each group, and three groups were tested. After the seeding performance became stable, the counting began. After the data were counted, the qualified index, reseeding index, and missing index of each group were calculated in accordance with the standard GB/T 6973-2005 [35] ‘Single Seed (Precision) Seeder Test Method’ and the actual agronomic requirements of direct precision hill seeding of spinach. The test results showed that the qualified index was 91.56%, the missing index was 1.56%, and the reseeding index was 6.89%.

3.5.2. Field Test

In order to explore the field sowing performance of the device, a field sowing test was conducted on 5 May 2023, in Erdong Town, Xinghe County, Inner Mongolia. The test spinach variety was Royal Dark Green; the test field was flat and was not loamy. The device was installed on a DS-type vegetable precision planter for the field test. During the testing, the machine was adjusted to the optimal parameter combination, and multiple tests were carried out. The spacing of the seeding was measured and recorded using a tape measure, and the data were processed. The sowing effect under stable machine operation is shown in Figure 8B. In relation to the bench test and the actual processing accuracy of the seeder, and due to the low germination index of spinach seeds, farmers tend to increase the reseeding index of spinach during planting. Therefore, the actual radius of the spoons on the spinach seed-metering device was designed to be 2.5 mm. The optimal parameter combination was set as follows: the rotational speed of the seed-metering device was 18 rpm, the radius of each spoon was 2.5 mm, and the number of spoons was 50. Details of the field test and sowing effect are shown in Figure 8. The qualified index of sowing (1–2 seeds) was 90.89%, the reseeding index (>2 seeds) was 8.22%, and the missing index was 0.89%. The results verified the rationality of the simulation to meet the actual production requirements for spinach in the field.

3.6. Discussion

At present, in China, most spinach seeding is still carried out manually. Only a few choose the cell-wheel seed-metering device for seeding. However, the cell-wheel seed-metering device causes relatively severe damage to spinach seeds and has low operating efficiency. Moreover, the resulting growth does not meet the current market demand. Some large-scale growers choose some imported air suction seed metering devices. However, they are expensive and not suitable for widespread promotion.

This study innovatively designed a seed-taking spoon structure with a penetrating surface. Through the optimized design of the seed-taking spoon, this structure is suitable for the mechanized planting of spinach, effectively avoiding damage to spinach seeds, making the seeding process more smooth and ensuring the effective discharge of seeds. As was observed from the bench test (with a qualification index of 91.78% and a missed sowing index of 1.56%) and the field test (with an average qualification rate of 90.89% and a missed sowing index of 0.89%), the performance indicators of this seed-metering device basically meet the usage requirements and are close to the operation effect of the air suction seed-metering device [36].

Although on the whole, the seed-metering device can basically meet the requirements of agronomic sowing, it was apparent from the tests on the actual seeder that the effect still needs to be improved. Since there is the problem of machine vibration within the seeder during the actual sowing process, in later stages of this research, the effect of vibration on seed metering will be analyzed and optimized. In-depth research will be conducted on the reliability and durability of the processing materials used in the seed device.

Currently, agricultural technology is evolving in a more intelligent direction. The intelligence of seed-metering devices is mainly manifested in sowing quality monitoring. In subsequent research, we will install monitoring sensors appropriate to the structure and characteristics of the seed-metering device designed in this study, to realize the visualization of the seed-metering process and further improve the seed-metering quality.

4. Conclusions

This study analyzed the seeding mechanism of a spoon-wheel type direct seeding device for spinach and explored its seeding performance through EDEM simulation, bench testing, and field testing, and the following conclusions were drawn:

(1)

Through single-factor tests, with the average qualified rate as the indicator—the larger the value, the better—it was determined that the appropriate range of the seed wheel speed was 15 rpm–25 rpm, the appropriate number of spoons was 45–55, and the appropriate range of radius was 2 mm–2.5 mm. Based on this, a Box–Behnken test was carried out to obtain the optimal solution combination. When the seed wheel speed was 18 rpm, the seed spoon radius was 2.25 mm, and the number of seed spoons was 50, the seed-picking performance of the seed picker reached its best; the seed-picking qualified rate was 91.78%, and the seed-missing rate was 1.56%.

(2)

After optimizing the combination, the bench test results showed that when the rotational speed was 18 rpm, the seed scoop radius was 2.25 mm, and the number of seed scoops was 50, the average qualified index was 91.56%, the reseeding index was 6.89%, and the missing index was 1.56%. The field test results showed that when the rotational speed was 18 rpm, the seed scoop radius was 2.5 mm, and the number of seed scoops was 50, the average qualified index was 90.89%, the reseeding index was 8.22%, and the missing index was 0.89%.

(3)

All performance indicators of the seed-metering device essentially conformed to the operational requisites. Nevertheless, there remains ample scope for optimization and enhancement in the seeding process and seeding stability.

(4)

This research provides ideas for innovative research and the development and efficient application of spinach seed-metering devices and offers a reference for the green mechanized production of spinach.