Research on the Flexible Gradual Seed-Cleaning Method of the Brush-Type Single-Seed Soybean Planter

Abstract

In response to the poor cleaning effect of the brush-type soybean seeder, a lightweight mechanism of the brush-type soybean seeder was optimized and designed. We proposed a progressive seed-cleaning method. Firstly, mechanical analysis was conducted on the soybean seeding process to prevent excessive clearing of individual soybean seeds initially filled into the orifices by the seeding brush. A formula for expressing the force exerted by the seeding brush on stabilizing individual soybean seeds in the orifice was derived. Secondly, an analysis of the progressive seeding principle was performed, and theoretical calculations were carried out for the progressive seeding curves within the x-y plane and along the direction of the z-axis. The progressive seeding curves were determined. Factors affecting the seeding efficiency were analyzed, including the plate’s speed, the initial angle of the seeding brush, and the initial distance between the seeding brush and the seeding plate. Based on the Box–Behnken central composite design theory, using the seeding tray’s rotational speed, the initial angle of the seeding brush, and the initial distance between the seeding brush and the seeding tray as experimental factors, orthogonal experiments were conducted with the leakage rate and the over-cleaning rate as the indicators for the evaluation, and the working parameters were optimized. Validation experiments showed that with a plate speed of 29 rpm, an initial angle of the seeding brush of 54°, and an initial distance between the seeding brush and the seeding plate of 11 mm, the results were as follows: a skip-seeding rate of 1.23% and an over-seeding rate of 0.66%, indicating a significant improvement in seeding performance. A comparative experiment was conducted with non-progressive cleaning brushes, and the test results showed that the progressive cleaning brush reduced the leakage rate by more than 1% and the over-cleaning rate by more than 0.5%.

1. Introduction

The seed metering unit is the core component of seeding equipment, and its performance has a crucial impact on the efficiency of seeding [1,2,3]. Mechanical seed metering units are widely used worldwide due to advantages such as their simple structure, low cost, and good versatility [4,5,6,7]. Seed cleaning refers to the removal of excess seeds from the seed cell to ensure single-seed cells for single-grain seeding and the improved accuracy of seeding. However, the current seed-cleaning mechanisms of seed metering units still have many defects, mainly manifested in excessive cleaning force, which can cause damage to the seeds and over-cleaning, leading to missed planting. On the other hand, insufficient cleaning force makes it difficult to achieve the desired cleaning effect, resulting in reseeding [8,9,10,11,12].

Researchers from China and other countries have conducted extensive research on the cleaning performance of seed metering units. Li Chuan et al. [13] used the principle of the centrifugal force generated by the high-speed circumferential motion of corn seeds to design a precise high-speed seed packaging device for centrifugal filling and cleaning of corn, achieving clean production. Gao Xiaojun et al. [14] designed a high-speed seed metering unit device using centrifugal force for seeding, cleaning, and air movement, achieving efficient and stable single-grain seeding. Tang Han et al. [15] designed an internally charged air-assisted precise seed metering unit device, achieving accurate seeding. In most of these studies, optimized cleaning mechanisms and multiple cleanings were used to improve the cleaning effect of the seed meter. However, most are applicable to pneumatic seed metering units, with relatively few optimizations for the cleaning mechanisms of the mechanical seed metering unit [16,17,18,19,20].

This article presents a progressive cleaning brush design that uses a cleaning method with an elliptical curve in the x-y plane of the cleaning area and a linear cleaning method in the direction of the z-axis. This progressive seed-cleaning method effectively removes excess seeds from the seed cells while protecting the soybean seeds, ensuring a single seed per seed cell, and improving the cleaning performance of the brush-type plan.

The progressive seed-cleaning method can effectively improve the cleaning effect, thereby ensuring accurate sowing. This method achieved precise single-seed sowing through a slowly changing seed-cleaning method, providing a reference for improving the sowing accuracy of mechanical seeders. The improvement of sowing accuracy can effectively increase crop yield.

2. Structure and Working Principle of the Seed Planter

2.1. Structure of the Seed Planter

The internal structure of the brush-type soybean planter is shown in Figure 1. It is mainly composed of the drive shaft, the housing, the seed-carrying brush, the seed-carrying brush holder, the seed-cleaning brush, the flange plate, and the planter plate. The transmission shaft is installed in the middle of the shell and connected to the flange plate. The flange plate is synchronously connected to the planter plate. The seed-cleaning brush is installed in the upper right of the filling area. The seed-carrying brush is installed inside the shell, flexibly pressing the beans into the seed cells to prevent them from falling into the seed box. The seed-cleaning brush uses a progressive seed-cleaning method and flexible material to remove excess seeds from the seed cells, ensuring single-seed sowing and precise seeding.

The diagram on the right-hand side of Figure 1 shows the progressive cleaning brush, which can achieve progressive cleaning in space and ensure the accuracy of sowing. In the plane, the blue line represents the inner curve of the progressive cleaning brush, and the red line represents the outer curve of the progressive cleaning brush. In space, the red line represents the spatial curve of the progressive cleaning brush.

2.2. Working Principle of the Seeder

The working principle of the brush-type soybean planter is shown in Figure 2, which is mainly divided into the filling area, the cleaning area, the carrying area, and the sowing area. Firstly, the seeds in the seed box are filled into the space formed by the shell and the seed plate under the action of gravity and the inter-seed force, and further filled into the seed cells of the seeding plate. As shown in the schematic diagram on the right side of Figure 1, the seeds filled into the seed cell are rotated to the cleaning area by the seeding plate. Under the action of a progressive cleaning brush, excess seeds are removed from the seed cell, ensuring a single seed cell and single grain. When they reach the carrying area, the soybeans are stabilized in the area formed by the carrying brush, seed cell, and shell, ensuring that they will not drop into the seed box during the carrying process. When they reach the sowing area, the soybeans are released from the carrying brush and sown under the action of their own gravity.

3. Progressive Cleaning Process Analysis

The black arrow in Figure 3 represents the direction of the progressive seed-cleaning brush. If a full area cleaning brush is used, it will increase manufacturing costs. Moreover, when the bean seeds first enter the clearing area, the gravity inclination angle is relatively small, and overcharged seeds cannot be cleared by their own weight. As the seeding tray rotates to the end of the cleaning area, the gravity tilt angle increases. At this point, even without a cleaning brush, overcharged seeds will still use their own weight to complete the cleaning process. The only beans that need to be cleared are those that cannot be cleared by their own weight. And the cleaning brushes in the entire region are prone to having a “protective” effect on overcharged seeds. It covers the overcharged seeds under the cleaning brush, causing replays. Therefore, in order to accurately remove soybean seeds that cannot be cleared by their own weight, a progressive clearing method is adopted to improve sowing accuracy.

3.1. Analysis of Over-Clearance Prevention

To ensure that the single seed entering the seed cell is not overly cleared, the force analysis is conducted on the single-seed bean. Authors establish a three-dimensional coordinate system. The negative direction of centrifugal force on the seed is the positive x-axis. The direction tangent to the rotation of the sowing plate (opposite to the air resistance direction of the bean) is the positive direction of the y-axis, and the direction perpendicular to the type orifice (facing away from the type orifice) is the positive direction of the z-axis. The force analysis diagram is shown in Figure 4. In Figure 4, the red A represents the centroid of the bean seed.

The translation theorem of force allows us to disregard the influence of seed cell friction on the equivalent torque of the seed. We translate the frictional force on the bean seeds to the center of mass of the seeds. We can obtain the following equation.

$$\left\{\begin{array}{l}{{\Sigma F}_x=0\to F}_{N2}+f_2+J=G\sin \alpha +F_1+F_{N3}\\ {{\Sigma F}_y=0\to F}_2+f_3+G\cos \alpha =F_{N1}+f_1\\ {\Sigma F}_z=0\to F_3=F_{N4}+f_4\end{array}\right.$$

(1)

Equation (1) can be explained by Equation (2).

$$\left\{\begin{array}{l}{f}_1=u_1{F}_{N2}\\ f_2=u_1{F}_{N1}\\ J=m{\omega }^{2}r\\ f_4=u_1{F}_{N4}\\ F=\sqrt{F_1^2+F_2^2+F_3^2}\end{array}\right.$$

(2)

The combination of Equations (1) and (2) yields the following result.

$$\left\{\begin{array}{l}{F}_1=F_{N2}+f_2+J-G\sin \alpha -F_{N3}\\ F_2=F_{N1}+f_1-f_3-G\cos \alpha \\ F_3=F_{N4}+f_4\end{array}\right.$$

(3)

In the formula, G = the gravitational force acting on the bean, N; J = the centrifugal force acting on the bean, N; F_N1 = the support force on the lower wall surface of the bean seed cell, N; F_N2 = the support force on the side wall surface of the bean seed cell, N; F_N3 = the support force on the side wall surface of the bean, N; F_N4 = the support force on the bottom surface of the bean seed cell, N; f₁ = the frictional force on the y-axis direction of the bean seed cell, N; f₂ = the frictional force on the x-axis direction of the bean seed cell, N; f₃ = the air resistance acting on the bean, N; f₄ = the frictional force on the z-axis direction of the bean seed cell, N; F₁ = the component force of the clearing brush on the x-axis direction of the bean, N; F₂ = the component force of the clearing brush on the y-axis direction of the bean, N; F₃ = the component force of the clearing brush on the z-axis direction of the bean, N; u₁ = the dynamic friction coefficient between the bean and the seed cell; m = the mass of the bean, g; r = the distance between the center of the sowing plate and the centroid of the bean, mm; ω = the angular velocity of the sowing plate, rad/s; α = the angle between the line connecting the center of the sowing plate and the centroid of the bean and the horizontal direction, degrees.

According to Equation (3), it can be seen that the cleaning brush exerts balanced supporting forces in the three planar directions. The cleaning brush can protect the single soybean seeds stabilized in the seed cell. The single bean seeds that are first accommodated by the seed cells are stably enclosed in the space formed by the seed cells, shell, and brush, effectively preventing over-cleaning.

3.2. Progressive Clear Seeding Brush Design

The cleaning brush for the entire area makes it difficult for overcharged seeds to fall out of the seed cell. At this time, after the sowing plate enters the clear-seed area, the clear-seed brush not only fails to perform the clear-seed function, but instead easily covers the over-filled seeds into the seed cells, playing a “protective” role for the over-filled seeds. If only a clear-seed brush covering the seed cells is used, it is difficult to simultaneously consider the over-filled seeds inside the seed cells and the guide groove. Therefore, a progressive clear-seeding method is proposed. When the beans have just entered the clear-seed area, a small clear-seeding force is used, which can only remove the over-filled seeds in the guide groove. With the rotation of the seed plate, the cleaning range of the cleaning brush gradually increases, and the smooth transition to the seed cell area completes the cleaning process.

3.2.1. Progressive Cleaning Brush with x-y

The diameter of the seeding plate is designed to be 200 mm [21,22]. The sum of radial dimensions of the guide groove and the seed cell is 30 mm. In Figure 5, the length of segment EF represents 30 mm. The design of the progressive seed-cleaning brush plane curve is shown in Figure 5.

The progressive planting of bristles on the outer curve of the x-y surface is designed by the following equation.

$$\frac{x^2}{a^2}+\frac{y^2}{b^2}=1$$

(4)

In the formula, a is the major axis of the ellipse; b is represents the minor axis of the ellipse.

The coordinates of point M are (−100, 0). The coordinates of point N are (−70cosβ, −70sinβ). The curve passes through points M and N, and by substituting into Equation (4), the following equation can be obtained.

$$\frac{x^2}{{100}^2}+\frac{\left({100}^2-{70}^2{\cos }^2\beta \right)y^2}{{\left(70\times 100\sin \beta \right)}^2}=1$$

(5)

In the formula, β is represents the initial angle of the cleaning brush, in degrees.

The progressive planting brush x-y inner curve design is represented by the following equation.

$$x^2+y^2={70}^2$$

(6)

The overcharged soybean seeds just entering the cleaning area cannot complete the cleaning process by their own weight due to the small gravitational inclination. Soybean seeds that cannot be cleared using their own weight despite rotating the seed tray to a certain angle will touch the cleaning brush and complete the cleaning process under the force of the cleaning brush. Therefore, the initial angle of the cleaning brush is crucial. Determining the initial angle β of the cleaning brush can effectively improve the progressive cleaning effect of the cleaning brush. We conducted preliminary experiments on the starting angle. We keep other factors unchanged, including the speed of the seeding plate, starting distance, etc. We only change the numerical value of the starting angle. Through preliminary experiments, it can be determined that when 50° ≤ β ≤ 60°, the cleaning effect is better.

3.2.2. Progressive Planting of Bristles Designed for the z-Axis

To ensure that the soybean seeds have sufficient space to use their weight to complete the sowing, the design of the cleaning brush curve along the z-axis is shown in Figure 6.

In the z-axis direction, the distance between the cleaning brush and the seed disk must be greater than the bean seed size. When leaving the seeding area, the distance between the cleaning brush and the seed disk should be less than or equal to zero. So we obtained the following equation.

$$\left\{\begin{array}{c}h\ge e\\ z=ky\end{array}\right.$$

(7)

In the formula, e = the longitudinal dimension of the bean, mm; h = the initial distance in the z-axis direction between the progressive cleaning brush and the seed plate, mm; k = the coefficient of the z-axis curve of the progressive cleaning brush.

The coordinates of N₁ are (−70sinβ, h). This is because the x-axis of point N in Figure 5 is “−70sinβ”. The equation can be obtained by substituting the coordinates of N₁ into Equation (7).

$$z=-\frac{h}{70\sin \beta }$$

(8)

The longitudinal axis dimension “e” of the Zhonghuang 13 soybean variety is 8.6 mm (measured as the average value of 100 seeds). Combined with the preliminary experiments, it is determined that the progressive sowing brush has better sowing effect when the initial distance “h” is in the range of 9 to 13 mm.

4. Test Stand Experiment

4.1. The Experimental Materials and Apparatus

We selected “Zhonghuang 13” soybean as the experimental sample. Its average geometric size was 8.62 × 5.56 × 6.21 mm. This was the average value obtained by measuring 100 seeds. As shown in Figure 7, the experimental equipment mainly included an aluminum profile stand, a brush-type seeder, a motor, and a seed guide tube. The brush-type seed planter was made by 3D printing technology, with the seed tray made of transparent photosensitive resin material, allowing effective observation of the sowing performance of the seed planter. Three-dimensional printing technology is the technique of printing three-dimensional drawings into physical objects. The starting point was the gradual seed clearing brush area where the beans entered, and the endpoint was the gradual seed clearing brush area where they exited. The high-speed slow-motion camera was used to record the sowing effect of the gradual seed clearing brush [23].

4.2. Experimental Design

Based on the theoretical research on the parameters about influencing the cleaning and sowing process, authors obtained the experimental factors. Experiments were conducted using the rotational speed of the sowing plate (n), the initial angle of the cleaning brush (β), and the initial distance between the cleaning brush and the sowing plate along the z-axis (h) as independent factors. The brush-type sowing apparatus in this study aimed to achieve high-speed operation of over 8 km·h⁻¹. Based on the requirement of soybean planting spacing [8], we had designed the seed tray speed n to be at three levels: 25, 30, and 35 rpm. Based on theoretical analysis and preliminary experiments, the initial angle of the cleaning brush (β) was set at three levels: 50°, 55°, and 60°, and the initial distance between the cleaning brush and the sowing plate along the z-axis (h) was set at three levels: 9 mm, 11 mm, and 13 mm. The levels of experimental factors are given in

Table 1. Experiment factor coding.

Level		Factor
	Seeder Plate Rotation Speed A/(rpm)	Clearing the Starting Angle of the Bristles B/(°)	The Initial Distance between the Cleaning Brush C/(mm)
−1	25	50	17
0	30	55	19
1	35	60	21

.

The leakage rate was defined as the ratio of the number of missed seed cells to the total number of seed cells. The over-cleaning rate was defined as the ratio of over-cleaned seed cells to the total number of seed cells. The number of test seed cells detected was greater than three hundred and sixty, repeated three times. The calculation formula was the following equation.

$$\left\{\begin{array}{c}{M}_p=\frac{m_p}{N_p}\times 100\%\\ H_p=\frac{h_p}{N_p}\times 100\%\end{array}\right.$$

(9)

In the formula, M_p = the leakage rate, %; H_p = the clearance rate, %; m_p = the number of missed seed cells, unit; h_p = the number of over-cleaning seed cells, unit; N_p = the total number of seed cells counted, unit.

4.3. The Experimental Results and Analysis

4.3.1. The Statistical Analysis of the Experiment

Box–Behnken design is a synthesis of statistical design and experimental techniques; we utilized Box–Behnken experimental design and obtained certain data through experiments. The Box–Behnken design uses a multivariate quadratic equation to fit the functional relationship between factors and effect values, and seeks the optimal process parameters through analysis of the regression equation, which is a statistical method for solving multivariate problems.

According to the Box–Behnken experimental principle, the experimental design and analysis were conducted with a total of 17 experimental points [24,25,26,27]. The experimental results are shown in

Table 2. Bench test results.

Serial Number		Test Factors		Test Metrics
	A	B	C	Missed Clearances Y1	Over Clearances Y2
1	0	0	0	1.29	0.62
2	1	0	−1	1.82	1.25
3	0	0	0	1.29	0.71
4	−1	−1	0	1.43	1.04
5	1	1	0	1.86	1.27
6	0	1	1	1.72	1.51
7	0	−1	1	1.42	1.19
8	−1	0	−1	1.65	1.03
9	−1	0	1	1.29	1.22
10	1	−1	0	1.66	1.37
11	0	0	0	1.15	0.59
12	1	0	1	1.79	1.28
13	0	−1	−1	1.71	1.17
14	0	0	0	1.31	0.69
15	0	0	0	1.23	0.62
16	0	1	−1	1.86	1.26
17	−1	1	0	1.65	1.29

.

4.3.2. The Establishment of the Regression Model and the Test of Significance

The authors used Design Expert 13.0 software to conduct a multiple regression fitting analysis of the experimental results, and established response surface regression models of Y₁, Y₂, Y₃ to A, B, C. Among them, Y₁ referred to the clearance rate, and Y₂ referred to the clearance rate. A referred to the speed of the seeding plate, B referred to the initial angle, and C referred to the initial distance.

$$Y_1=1.25+0.1388A+0.1088B-0.1025C-0.005AB+0.0825AC+0.0375BC+0.178A^2+0.218B^2+0.2055C^2$$

(10)

$$Y_2=0.646+0.0738A+0.07B+0.0612C-0.0875AB-0.04AC+0.0575BC+0.2545A^2+0.342B^2+0.2945C^2$$

(11)

The regression equation analysis of variance is shown in

Table 3. Analysis of variance.

Source of Variation	Leakage Rate				Pass Rate
	Sum of Squares	Degree of Freedom	F	p	Sum of Squares	Degree of Freedom	F	p
	0.9365	9	30.29	<0.0001	1.42	9	57.09	<0.0001
A	0.1540	1	44.84	0.0003	0.0435	1	15.70	0.0054
B	0.0946	1	27.54	0.0012	0.0392	1	14.15	0.0071
C	0.0840	1	24.47	0.0017	0.0300	1	10.83	0.0133
AB	0.0001	1	0.0291	0.8693	0.0306	1	11.05	0.0127
AC	0.0272	1	7.93	0.0259	0.0064	1	2.31	0.1724
BC	0.0056	1	1.64	0.2414	0.0132	1	4.77	0.0652
A2	0.1334	1	38.84	0.0004	0.2727	1	98.43	<0.0001
B2	0.2001	1	58.25	0.0001	0.4925	1	177.74	<0.0001
C2	0.1778	1	51.76	0.0002	0.3652	1	131.80	<0.0001
Residuals	0.0240	7			0.0194	7
Misfitting	0.0069	3	0.5393	0.6804	0.0089	3	1.12	0.4386
error	0.0171	4			0.0105	4
sum	0.9606	16			1.44	16

Note: 0.01 < p < 0.05 indicates a significant difference. The expression “p < 0.01” represents an extremely significant difference.

.

An analysis of Table 3 shows that the p-values for the leakage rate Y₁ and the excess clearance rate Y₂ were both less than 0.0001 (less than 0.05), indicating that these two models had a statistically significant impact. Through analysis of variance, the missed cleaning rate Y₁ and the over-cleaning rate Y₂ had a significant impact on the rotation speed A, starting angle B, and starting distance C. The p-values for the misfit items were 0.6804 and 0.4386, respectively (both greater than 0.05), indicating that the fitting degree of these three models was relatively high.

4.3.3. Response Surface Analysis

Based on the experimental results, authors studied the effects of the rotational speed n of the seeding disc, the starting angle β of the cleaning brush, and the starting distance h between the cleaning brush and the seeding disc along the z-axis direction on the leakage rate and over-cleaning rate of the seeding device. The authors used Design Expert 13.0 software to draw response surfaces.

As shown in Figure 8a, when the rotation speed of the seeding plate was low, the leakage rate was relatively small. With an increase in the rotation speed of the seeding plate, the leakage rate sharply increased. The reason was that as the speed of the seeding disc increased, the inertia force on the excess bean seeds increased. Excess bean seeds were difficult to remove with a cleaning brush, resulting in an increase in missed cleaning rates. When the initial angle was small, the effect on the leakage rate was relatively small. With an increase in the initial angle, the leakage rate sharply increased. The reason was that when the starting angle was too large, the coverage area of the cleaning brush was too large. Overcharged seeds were protected by the cleaning brush, resulting in an increased leakage rate. The interaction between the rotation speed of the seeding plate and the initial angle had no significant effect on the leakage rate. As shown in Figure 8b, increasing the initial distance helped reduce the leakage rate. The reason was that when the starting distance was small, the overcharged bean seeds were covered by the cleaning brush. Overfilling soybean seeds could not be cleared by their own gravity, resulting in an increased leakage rate. The interaction between the rotation speed of the seeding plate and the initial distance had a significant effect on the leakage rate. As shown in Figure 8c, the interaction between the initial angle and the initial distance had no significant effect on the leakage rate.

As shown in Figure 9a, when the rotation speed of the seed-feeding plate was low, the over-cleaning rate was minimally affected. With an increase in the rotation speed of the seed-feeding plate, the over-cleaning rate sharply rose. The reason was that as the speed of the seeding disc increased, the inertia force on a single bean seed inside the holes increased. Single bean seeds were prone to slipping out of the mold holes, resulting in an increased clearance rate. The over-cleaning rate first decreased and then increased with an increase in the initial angle. The reason was that when the starting angle was small, the cleaning brush with a smaller starting angle could not cover a single bean seed inside the mold hole. Single bean seeds were prone to falling into the seed box. As the initial angle increased, the over-cleaning rate gradually decreased. With a further increase in the initial angle, the oversized cleaning brush easily removed the single soybean seed inside the shaping seed cell, leading to an increase in the over-cleaning rate. The interaction between the rotation speed of the seed-feeding plate and the initial angle significantly affected the over-cleaning rate. As shown in Figure 9b, the over-cleaning rate first decreased and then increased with an increase in the initial distance. This was because when the initial distance was small, the single soybean seed inside the shaping seed cell was removed by the cleaning brush, leading to an increase in the over-cleaning rate. After the initial distance increased, excess bean seeds inside the pore were cleared. Single bean seeds could stably migrate to the seed carrying area under their own gravity, pore size, and cleaning brush force, and the clearance rate decreased. As the starting distance continued to increase, single bean seeds within the pore were more likely to slide out of the pore under their own gravity, resulting in an increase in clearance rate. The interaction between the rotation speed of the seed-feeding plate and the initial distance did not significantly affect the over-cleaning rate. As shown in Figure 9c, the interaction between the initial angle and initial distance did not significantly affect the over-cleaning rate.

4.4. Parameter Optimization and Confirmatory Experimentation

4.4.1. Parameter Combination Optimization

To achieve the lowest leakage rate and over-cleaning rate, an optimized mathematical model was established [28,29,30].

$$\left\{\begin{array}{l}min{Y}_1(A,B,C)\\ min{Y}_2(A,B,C)\\ s.t.\left\{\begin{array}{c}25\mathrm{rpm}\le A\le 35\mathrm{rpm}\\ 50°\le B\le 60°\\ 9\mathrm{mm}\le C\le 13\mathrm{mm}\end{array}\right.\end{array}\right.$$

(12)

To achieve the optimal cleaning effect, the planter needed the minimum missed cleaning rate Y₁ and over-cleaning rate Y₂. Meanwhile, according to the analysis in Section 4.2 above, the speed range of the seeding plate A was 25–35 rpm, the starting angle range of B was 50–60 degrees, and the starting distance range of C was 9–13 mm. Formula (12) shows the result. Using the Design Expert 13.0 software optimization module, the optimal parameter combination that meets the constraint conditions for minimum under-seeding and over-seeding rates was obtained. The optimal parameter combination for the solution was as follows: sowing disk rotational speed of 28.6 rpm, initial angle of the sowing brush 54°, initial distance of the sowing brush 11.2 mm, corresponding miss-sowing rate of 1.2%, and over-sowing rate of 0.65%.

4.4.2. Confirmatory Experiment

The validation test adopted the same method of bench testing for validation. Due to the difficulty in achieving the optimized values of the theoretical solution for the actual manufacturing process requirements, a set of parameters close to the optimized values was selected for validation testing. The parameter values were as follows: sowing plate rotation speed 29 rpm, initial angle of the cleaning brush 54°, initial distance of the cleaning brush 11 mm. The test results were as follows. The leakage rate was 1.23%, and the over-cleaning rate was 0.66%. Compared to a 3.6% missed cleaning rate and a 1.9% over-cleaning rate of a regular cleaning brush, the progressive cleaning brush significantly improved its cleaning performance.

4.5. Controlled Trial

At the same speed, authors conducted a comparative experiment on the cleaning effect between progressive cleaning brushes and non-progressive cleaning brushes. Using the rotational speed of the seeding machine as the experimental factor, a single-factor comparative experiment was conducted with the leakage rate and over-clearing rate as evaluation indicators. The experimental results are shown in Figure 10.

Figure 10a shows the comparison test results of the leakage rate. Figure 10b shows the comparison test results of the heavy cleaning rate. The comparative test results show that compared to non-progressive seeding brushes, the progressive seeding brush reduced the leakage rate by more than 1%. When the seeding plate rotation speed were between 30~35 rpm, the over-cleaning rate of the progressive seeding brush decreased by more than 0.5%, and the gap gradually increased with the increasing speed.

5. Conclusions

(1) This article designed a progressive seed-cleaning brush. Firstly, authors conducted an analysis on the prevention of over-cleaning of soybean seeds during the cleaning process. The purpose of doing this was to prevent the cleaning brush from removing single bean seeds from the mold hole. Secondly, a progressive clearing method was adopted, and a progressive clearing brush was designed to gradually clear the bean seeds on the x-y plane and along the z-axis direction. The main factors affecting the clearing effect were analyzed, including the rotation speed of the planting plate, the initial angle of the clearing brush, and the initial distance between the clearing brush and the planting plate.

(2) Based on the Box–Behnken center combination design theory, authors took the rotational speed of the seeding disc, the starting angle of the cleaning brush, and the starting distance between the cleaning brush and the seeding disc as experimental factors. Authors used Design Expert 13.0 software to optimize the working parameters and conducted validation experiments. Under the condition of seeding tray rotation speed of 29 rpm, seeding brush initial angle of 54°, and seeding brush initial distance of 11 mm, the experimental results showed that the leakage rate was 1.23% and the over-clearance rate was 0.66%, indicating a significant improvement in seeding performance. Authors conducted comparative experiments with non-progressive sowing brushes. The experimental results showed that compared to non-progressive sowing brushes, progressive sowing brushes reduced the missing sowing rate by more than 1% and the over-sowing rate by more than 0.5%.

6. Patents

The authors applied for a Chinese invention patent based on the content of this article. The patent was titled “A Dual-Row Soybean Precision Seeding Device and Seeding Method”. The application number was 2023101835248, with a filing date of 28 February 2023. The publication number was CN 116210409 A, and the publication date was 6 June 2023. It entered the substantive examination stage on 9 June.