Design and Operating-Parameter Optimization of a Precision Seeder for Chinese Yam Based on Automatic Seed Distribution and Chain-Driven Metering

Abstract

A precision seeder for elongated Chinese yam seed segments was developed by integrating automatic seed distribution with chain-driven metering. The design was based on measured seed segment properties, including mean length, width, thickness, equivalent diameter, density, moisture content, intrinsic mechanical properties, and seed-seed/seed-steel contact parameters. A seed-layer stress model, sprocket-conveying stability condition, seed-dropping trajectory equation, and plant-spacing equation were used to determine the main structural parameters and to select operating speed, seed-dropping height, and seed-box slope angle as optimization variables. Box–Behnken response surface optimization predicted the best parameter combination as an operating speed of 0.20 m s⁻¹, a seed-dropping height of 0.14 m, and a seed-box slope angle of 26.18°, with predicted qualified-seeding, multiple-seeding, and missed-seeding indices of 86.45%, 5.16%, and 8.40%, respectively. Field validation using the rounded seed-box slope angle of 26° produced mean qualified-seeding, multiple-seeding, and missed-seeding indices of 86.16%, 5.13%, and 8.71%, respectively. The results demonstrate a practical design route for oriented precision seeding of elongated tuber segments.

1. Introduction

Chinese yam (Dioscorea opposita Thunb.) is a medicinal and edible crop with clear economic value in China [1]. China is one of the major global production areas for yam, and its annual yam output has remained at approximately 10 million t in recent years, reaching 11.013 million t in 2024. The main production regions include Jiangsu, Zhejiang, Hebei, Henan, Shandong, Guangxi, and other provinces [2]. However, the planting operation in Hebei and other major production areas still relies mainly on manual or semi-mechanized seed placement, resulting in high seasonal labor demand, low operating efficiency, and high production cost. Because Chinese yam is commonly planted within a short spring window and require the oriented placement of slender seed segments, mechanized precision seeding is a key technical bottleneck for standardized and large-scale production.

Existing equipment for yams, sugarcane, cassava, and potato provides useful references for seed feeding, conveying, and metering, but these devices cannot be directly transferred to Chinese yam seed segments. Recent studies on yam minisett planters have developed tractor-operated double-row machines and evaluated their structural safety and metering components [3]. For sugarcane and cassava stem segments, recent studies have developed pre-cutting, conveying, ordered seeding, and spoon-chain metering mechanisms to improve seed placement and seeding uniformity [4,5,6,7]. Potato precision-metering research has emphasized the adaptability of metering devices to cut potatoes with different shapes and sizes, chain-spoon seed discharge, and the control of qualified seeding, multiple seeding, and missed seeding [8,9,10,11,12,13,14,15]. However, sugarcane and cassava seed materials are generally stem segments, while potato seed pieces are closer to round or block-shaped tubers. Chinese yam seed segments have a larger length-to-diameter ratio, more obvious surface irregularity, and stricter orientation requirements during dropping. Therefore, existing metering devices for these crops still have limitations in solving the ordered distribution, single-segment pickup, stable conveying, and oriented dropping of Chinese yam seed segments.

Domestic Chinese yam-related machinery has mainly focused on harvesting, soil breaking, trenching, or semi-automatic double-row planting, and research on automatic seed distribution combined with oriented chain-driven metering remains limited [16,17,18]. Chain-based discrete metering studies for cassava, potato, and garlic indicate that constrained chain conveying and seed-cup structures can improve the transport stability of irregular agricultural materials when the cup geometry is matched to material morphology [7,9,19]. These studies suggest that a chain-driven metering system has potential for handling elongated seed materials, but Chinese yam seed segments still require a dedicated automatic distribution device to convert random seed accumulation in the seed box into an ordered seed supply before metering.

Table 1. Technical and functional comparison of related seed-metering and seed-distribution devices.

Crop/Object	Typical Mechanism	Reported Technical/Functional Characteristics	Limitations for Chinese Yam Seed Segments	References
Yam minisett	Tractor-operated double-row planter	Double-row planting structure; metering and structural safety evaluated	Developed mainly for yam minisetts and local planting conditions, automatic ordered distribution for Chinese yam seed segments remains insufficient	[3]
Sugarcane/cassava stem segments	Pre-cutting, conveying, ordered feeding, spoon-chain or stem-guiding mechanisms	Suitable for elongated stem segments; can improve seed placement and seeding uniformity	Stem-segment mechanisms may not adapt well to short, soft, irregular Chinese yam seed segments	[4,5,6,7]
Cut potato/potato tuber	Precision seed-metering, chain-spoon discharge, excess-seed removing, seed-cutting, and seed-picking mechanisms	Focuses on qualified seeding, multiple seeding, missed seeding, plant-spacing control, and adaptability to seed size and shape	Designed mainly for potato tubers or cut potato pieces; longitudinal posture control for Chinese yam seed segments is limited	[8,9,10,11,12,13,14,15]
Monitoring and compensation devices	Capacitive sensing, seeding monitoring, miss-seeding detection, and compensation control	Improves detection of missed seeding and supports feedback compensation during precision planting	Provides useful monitoring and compensation references, but does not solve ordered seed supply and oriented dropping of elongated Chinese yam seed segments	[10,11,15,20,21,22]
Existing domestic Chinese-yam machinery	Soil breaking, harvesting, trenching, or semi-automatic double-row planting	Improves individual field operations and partially reduces labor input	Manual or semi-manual seed feeding remains common; automatic seed distribution and precision metering are still insufficient	[16,17,18]
Chain-based metering for irregular materials	Chain-driven or spoon-chain metering	Seed-cup constraint and chain conveying can improve transport stability for irregular materials	Structure must be matched to the specific geometry and posture requirement of Chinese yam seed segments	[7,9,19]
This study	Automatic seed distribution + chain-driven metering	Ordered seed supply, constrained conveying, oriented dropping, and operating-parameter optimization	Further validation is needed for different cultivars, seed sizes, soil types, and higher operating speeds	This study

summarizes the technical and functional characteristics of related seed-metering and seed-distribution devices. Existing yam minisett planters provide references for overall machine configuration, while sugarcane and cassava planters provide references for handling elongated stem segments. Potato and cut-potato precision metering studies provide useful information on spacing control and the reduction in multiple and missed seeding. Nevertheless, these devices are not specifically designed for short, soft, irregular, elongated Chinese yam seed segments. Therefore, the scientific problem addressed in this study is how to achieve continuous seed supply, stable constrained conveying, and oriented dropping for Chinese yam seed segments under field operating conditions.

In this study, seed segments of the Xiaobaizui Chinese yam cultivar, a main cultivar in Hebei Province, were used as the research object. A precision Chinese yam seeder combining automatic seed distribution and chain-driven metering was designed and optimized. The objectives were to: (1) measure the geometric, physical, mechanical, and contact parameters of Chinese yam seed segments as the basis for structural design; (2) design the overall machine, automatic seed distribution device, and chain-driven metering system according to the material characteristics and planting requirements; (3) clarify the effects of operating speed, seed-dropping height, and seed-box slope angle on seeding performance through theoretical analysis, single-factor tests, and response surface optimization; and (4) verify the applicability and stability of the optimized parameter combination through bench and field tests.

2. Materials and Methods

2.1. Test Materials and Physical Properties

The test material was seed segments of the Xiaobaizui Chinese yam cultivar collected from Shenze County, Shijiazhuang, China. Damaged, moldy, and obviously deformed samples were removed to improve sample consistency. Geometric dimensions were measured with a vernier caliper; density was measured by the suspension method; moisture content was determined by oven drying; elastic and shear moduli were measured using an HSSD electronic universal testing machine (Jinan Hensgrand Instrument Co., Ltd., Jinan, China); and restitution, static-friction, and rolling-friction coefficients were measured using a 5KF160 high-speed imaging system (Revealer/Qianyanlang; Hefei Zhongke Junda Vision Technology Co., Ltd., Hefei, China), a FASTCAM Mini UX100 high-speed camera (Photron Ltd., Tokyo, Japan), and bench tests. Geometric and mechanical properties were measured using 30 samples, whereas density, moisture content, and contact parameters were determined from 10 repeated tests. The average values are summarized in

Table 2. Main physical and contact parameters of Chinese yam seed segments.

Parameter Category	Parameter	Mean Value	SD	CV/%
Geometric parameter	Length	80.13 mm	4.81	6.00
Geometric parameter	Width	20.72 mm	1.66	8.01
Geometric parameter	Thickness	19.57 mm	1.57	8.02
Geometric parameter	Equivalent diameter	20.00 mm	1.40	7.00
Physical parameter	Moisture content	76.21%	0.76	1.00
Physical parameter	Density	1003 kg m−3	15.0	1.50
Intrinsic parameter	Elastic modulus	3.92 × 105 Pa	0.59 × 105	15.05
Intrinsic parameter	Shear modulus	1.40 × 105 Pa	0.21 × 105	15.00
Intrinsic parameter	Poisson’s ratio	0.40	0.02	5.00
Contact parameter	Coefficient of restitution: seed–seed	0.194	0.023	11.86
Contact parameter	Coefficient of restitution: seed–steel	0.352	0.042	11.93
Contact parameter	Static friction coefficient: seed–seed	0.361	0.043	11.91
Contact parameter	Static friction coefficient: seed–steel	0.273	0.033	12.09
Contact parameter	Rolling friction coefficient: seed–seed	0.169	0.025	14.79
Contact parameter	Rolling friction coefficient: seed–steel	0.047	0.009	19.15

Note: SD and CV denote standard deviation and coefficient of variation, respectively. The SD has the same unit as the corresponding mean value, whereas CV is dimensionless. These values describe the dispersion of the measured material parameters.

. The measurement of geometric, physical, and mechanical parameters followed the general approach used for tuber materials and the measurement principles in relevant planter and seed-tuber standards [23,24,25].

2.2. Overall Structure and Working Principle

The seeder is tractor-drawn and consists mainly of an automatic seed distribution device, seed box, fertilizer box, frame, ground wheel, furrow opener, seed-metering system, and seats (Figure 1). The seed-metering system is installed at the rear of the machine and is powered by the ground wheel through gear and chain transmission. The system performs seed feeding, picking, stable conveying, and oriented dropping. The fertilizer box and furrow opener complete synchronized fertilization and furrow formation. The machine uses a modular arrangement to facilitate adjustment of seed-dropping height, plant spacing, and seed-box slope angle according to different field conditions and seed segment specifications (

Table 3. Main technical specifications of the Chinese yam seeder.

Item	Design Value
Overall dimensions (L × W × H)/mm	2000 × 1200 × 900
Matched tractor power/kW	36.8–51.5
Number of rows	2
Operating speed/km h−1	1.0–3.6
Row spacing/mm	440
Plant spacing/mm	80–120

).

The planting process is shown in Figure 2. Before seeding, deep trench loosening and soil backfilling are completed. During operation, the furrow opener forms a seed furrow as the machine advances. The ground wheel synchronously drives the automatic seed distribution device and the chain-driven metering system. The pickup wheel supplies Chinese yam seed segments from the seed box into the metering system in an orderly manner. The seed cups convey the seed segments and release them at the outlet; the seed segments then fall into the seed furrow under gravity.

2.3. Design of the Automatic Seed Distribution Device and Chain-Driven Metering System

The automatic seed distribution device was designed to replace manual feeding and to provide a continuous and ordered supply of seed segments from the seed box to the chain-driven metering system. The geometric design was based on the measured seed segment dimensions rather than on a full derivation of simple dimensional relations. With a maximum seed segment diameter of 25 mm and a reserved clearance of 10 mm, the pickup-cavity diameter was determined as 35 mm. Considering the average seed segment length of 80 mm and a 20 mm axial allowance, the pickup-wheel width was set to 100 mm. A six-sided pickup wheel with a diameter of 120 mm was adopted to coordinate seed pickup with the subsequent metering process and to avoid excessive seed accumulation in the pickup region. Chain-based discrete metering has been used for potato, cassava, and garlic seed-metering devices, indicating that constrained chain conveying can improve seed transport stability when the seed-cup structure is matched to material geometry [7,9,19].

Because seed-box slope angle affects the flowability of slender seed segments and the stability of seed entry into the pickup cavity, a simplified seed-layer stress model was introduced to provide the theoretical basis for selecting this factor in the subsequent optimization test. The theoretical model was established using equivalent geometric parameters of the seed segments. Because Chinese yam seed segments are irregular biological materials, their geometric variability was not directly introduced into each equation. Instead, the measured mean values and dispersion indicators in Table 2 were used to determine representative structural parameters. The model assumes that the seed segment layer can be simplified as a continuous granular layer, and that a single seed segment can be represented by an equivalent diameter and an average radius during stress and trajectory analysis. The influence of seed segment variability was further evaluated through bench verification and field validation. For a differential seed layer with cross-sectional area A_s and perimeter C at depth y in the seed box, the vertical force balance can be expressed as:

$${\sigma }_y{A}_s+\gamma A_{s}dy-({\sigma }_y+d{\sigma }_y)A_s-f_{s}K{\sigma }_{y}Cdy=0,$$

(1)

where σ_y is the vertical stress, γ is the unit weight of the seed segments, f_s is the static friction coefficient between the seed segment and the seed-box wall, K is the lateral pressure coefficient, and y is the depth below the upper surface of the seed segment layer. Defining the hydraulic radius as R_h = A_s/C, Equation (1) can be rearranged as:

$${\displaystyle \frac{d{\sigma }_y}{dy}} = \gamma - {\displaystyle \frac{K{f}_s{\sigma }_y}{R_h}},$$

(2)

Under the boundary condition σ_y = 0 at y = 0, the vertical stress is obtained as:

$${\sigma }_y = {\displaystyle \frac{\gamma R_h}{K{f}_s}}[1 - exp(-{\displaystyle \frac{K{f}_{s}y}{R_h}})],$$

(3)

The relationship between the horizontal stress σ_x and vertical stress σ_y is described by the lateral pressure coefficient:

$$K = {\displaystyle \frac{{\sigma }_x}{{\sigma }_y}} = {\displaystyle \frac{1 - \mathrm{s}\mathrm{i}\mathrm{n}{\phi }_1}{1 + \mathrm{s}\mathrm{i}\mathrm{n}{\phi }_1}},$$

(4)

where φ₁ is the internal friction angle of the seed segments. The gravitational stress acting on an individual seed segment is expressed as:

$${\sigma }_G = {\displaystyle \frac{G}{{{\pi r}_1}^2}},$$

(5)

where G is the gravity of a single seed segment and r₁ is the average radius of the seed segment. The resultant seed-picking stress affected by the seed-box slope angle θ can then be simplified as:

$${\sigma }_p\left(\theta \right)={\sigma }_G\sin \theta +{\sigma }_y\sin \theta +{\sigma }_x\cos \theta ,$$

(6)

The calculated σ_p(θ) was integrated into the optimization workflow as a physical constraint rather than as a separate response variable. Specifically, the tested seed-box slope angle was limited to the range in which σ_p(θ) was sufficient to maintain a continuous seed supply but not high enough to cause excessive squeezing or mechanical damage according to the compression and shear tests. The response surface optimization then maximized the qualified-seeding index and minimized the multiple- and missed-seeding indices within this physically feasible range.

Equations (1)–(6) therefore explain why the seed-box slope angle was selected as a key factor: small θ values reduce the driving component for seed entry into the pickup cavity, whereas excessive θ values increase inter-segment compression and the probability of multiple picking.

The chain-driven metering system used a 10A chain with a pitch of 15.875 mm. To simplify assembly and improve synchronization, the driving and driven sprockets were designed as the same type. The sprocket center distance and the spacing between adjacent seed cups were set as 18 and 3 times the chain pitch, respectively, corresponding to 285.75 mm and 47.625 mm. Forty-nine seed cups were uniformly arranged on each chain. The pitch diameter of the driving sprocket was 41.23 mm, and the spacing between adjacent metering units was 220 mm.

During conveying over the sprocket, the seed segment is affected by gravity, normal force, friction, and centrifugal force. To reduce the risk of seed detachment or collision, the sprocket radius R should satisfy the following stability condition:

$$R \ge {\displaystyle \frac{v^2}{g(\mathrm{s}\mathrm{i}\mathrm{n}{\theta }_1 + \mu \mathrm{c}\mathrm{o}\mathrm{s}{\theta }_1)}},$$

(7)

where v is the seed-cup linear velocity, g is gravitational acceleration, μ is the friction coefficient between the seed segment and the seed cup, and θ₁ is the central friction angle of the sprocket. With v = 0.50 m s⁻¹ and μ = 0.273, the calculated lower limit of R was approximately 0.023 m; therefore, a larger radius was selected to improve conveying stability.

After the seed segment leaves the seed cup, it has an initial horizontal velocity and then drops under gravity. Neglecting air resistance, the seed-dropping trajectory can be described by:

$$\{\begin{array}{c}x = v_{1}t\\ y={\displaystyle \frac{1}{2}}g{t}^2\end{array},$$

(8)

where x and y are the horizontal and vertical displacements, respectively, v₁ is the seed-cup linear velocity, and t is the seed-dropping time. This relationship explains why seed-dropping height was selected as another key factor in the response surface optimization.

The theoretical plant spacing s_p was determined by the machine operating speed v₀, seed-cup spacing L, and seed-cup linear velocity v₁:

$$s_p = {\displaystyle \frac{v_{0}L}{v_1}},$$

(9)

Within the operating-speed range of 0.10–0.25 m s⁻¹ and seed-cup linear-velocity range of 0.20–0.60 m s⁻¹, the theoretical plant spacing was 90–130 mm, which satisfied the agronomic spacing requirement for Chinese yam planting. Figure 3, Figure 4 and Figure 5 show the seed-picking stress state, conveying force analysis, and the seed guiding/dropping process, respectively.

2.4. Experimental Design and Evaluation Indicators

Based on preliminary single-factor tests and analysis of the operating process, operating speed, seed-dropping height, and seed-box slope angle were selected as the key factors. A three-factor, three-level response surface optimization test was conducted on the seed-metering bench. The qualified-seeding index Y₁, multiple-seeding index Y₂, and missed-seeding index Y₃ were used as the evaluation indicators. The qualified-seeding index was defined as the proportion of seed segments whose plant spacing fell within the agronomically acceptable range. The multiple-seeding index was defined as the proportion of intervals containing two or more seed segments, and the missed-seeding index was defined as the proportion of intervals without a seed segment. The indicators were calculated as follows:

$$Y_1 = {\displaystyle \frac{n_1}{N^{\prime }}} \times 100\%; Y_2={\displaystyle \frac{n_2}{N^{\prime }}} \times 100\%; Y_3 = {\displaystyle \frac{n_0}{N^{\prime }}} \times 100\%$$

(10)

where N is the total number of evaluated intervals, n₁ is the number of qualified intervals, n₂ is the number of multiple-seeding intervals, and n₀ is the number of missed-seeding intervals.

The single-factor tests were repeated three times at each factor level, and the effects of individual factors were evaluated using one-way analysis of variance. For each test run, the evaluated intervals were recorded after the seed-metering system reached stable operation. To reduce systematic bias, the order of treatments was adjusted between repeated runs where possible. The statistical analysis assumed independent observations within each treatment, and differences among factor levels were evaluated using one-way ANOVA in IBM SPSS Statistics (version 27.0; IBM Corp., Armonk, NY, USA). The response surface experimental design, regression analysis, model analysis of variance, and parameter optimization were conducted using Design-Expert software (version 13.0; Stat-Ease Inc., Minneapolis, MN, USA) Fisher’s F-test was used to evaluate statistical significance, with p < 0.05 considered significant and p < 0.01 considered highly significant (

Table 4. Factors and levels used in the response surface test.

Level	Operating Speed X1/(m s−1)	Seed-Dropping Height X2/m	Seed-Box Slope Angle X3/°
1	0.15	0.10	20
2	0.20	0.15	30
3	0.25	0.20	40

).

The bench-test platform consisted of a self-developed precision seeding control terminal, a self-developed automatic seed distribution device, a self-developed seed-metering system, a chain-transmission device, a DM9045 stepper motor driver, a VICTOR 62PG015.048-0486 stepper motor (Shenzhen Victor Motor Co., Ltd., Shenzhen, China), and a high-speed imaging system (Figure 6a). It allowed the operating speed, seed-dropping height, and seed-box slope angle to be adjusted. Field validation was conducted at the experimental base of Hebei Nonghaha Machinery Group Co., Ltd. (Shenze, Hebei, China). The soil was loam with a moisture content of 18–22% and a soil firmness of 1.2–1.5 MPa. The prototype was matched with an SD3004 tractor (Shandong Sadin Heavy Industry Co., Ltd., Weifang, China) for field validation (Figure 6b). The test field had been deep-tilled and leveled before validation. The seed material was Xiaobaizui Chinese yam seed segments from Shenze County, Shijiazhuang, China; approximately 20 kg of seed segments were prepared, with individual segment mass controlled at 30–50 g. To facilitate seed-spacing observation and data recording, the fertilizer box was kept empty and the covering and compaction devices were removed during the validation run. Video recording was used to identify multiple and missed seedings and to cross-check the field measurements.

3. Results and Discussion

3.1. Single-Factor Test Results

The single-factor tests were used to identify the operating variables that strongly influenced metering performance and to define the factor ranges for subsequent response surface optimization. Operating speed, seed-dropping height, and seed-box slope angle had significant effects on the qualified-seeding index (Y₁), multiple-seeding index (Y₂), and missed-seeding index (Y₃). As the operating speed increased from 0.10 to 0.20 m s⁻¹, the coordination between seed entry into the cups and passage through the guide outlet improved, resulting in a higher qualified-seeding index. When the speed was further increased to 0.25–0.30 m s⁻¹, the inertial impact near the sprocket increased, which raised the probabilities of missed and multiple seeding. Seed-dropping height also affected the landing stability of the seed segments: insufficient dropping height limited posture adjustment after release from the cup, whereas excessive height increased landing dispersion. The seed-box slope angle affected seed segment supply and picking stability; a small angle reduced seed supply, whereas an excessive angle increased squeezing and multiple picking. In contrast, the number of seed segments in the seed box had no significant effect on most performance indices within 30–70 pieces. Therefore, operating speed, seed-dropping height, and seed-box slope angle were selected as the three factors for response surface optimization, while the number of seed segments in the seed box was excluded from the optimization factors (

Table 5. Summary of significance results from the single-factor tests.

Factor	Qualified-Seeding Index Y1	Multiple-Seeding Index Y2	Missed-Seeding Index Y3
Operating speed	F = 28.518, p < 0.001	F = 56.048, p < 0.001	F = 16.143, p < 0.001
Number of seed segments in the seed box	F = 1.046, p = 0.431	F = 3.271, p = 0.058	F = 3.056, p = 0.069
Seed-dropping height	F = 6.235, p = 0.009	F = 19.946, p < 0.001	F = 3.804, p = 0.039
Seed-box slope angle	F = 12.067, p < 0.001	F = 22.265, p < 0.001	F = 3.381, p = 0.054

).

3.2. Response Surface Optimization

Using operating speed X₁, seed-dropping height X₂, and seed-box slope angle X₃ as independent variables, and maximizing Y₁ while minimizing Y₂ and Y₃ as optimization objectives, quadratic regression models were established and analyzed by analysis of variance. The optimization test was based on measured bench-test data. In this workflow, the single-factor tests were used to screen the main factors and define their reasonable level ranges; the response surface models were then used to analyze interactions among the selected factors and to obtain continuous optimal parameters. Finally, bench verification and field validation were used to complete the experimental closed loop.

For each response indicator, the second-order response surface model was expressed as:

Y = β₀ + Σβ_iX_i + Σβ_iiX_i² + Σβ_ijX_iX_j + ε,

(11)

where Y is the predicted response, X_i and X_j are coded independent variables, β₀ is the intercept, β_i, β_ii, and β_ij are the linear, quadratic, and interaction coefficients, respectively, and ε is the residual error.

According to the regression fitting and ANOVA results, non-significant terms were removed from the final coded regression equations. The fitted regression models for the qualified-seeding index Y₁, multiple-seeding index Y₂, and missed-seeding index Y₃ were as follows:

$$Y_1=86.14-1.42X_1-1.31X_2-1.88X_3-3.22X_1{X}_2+2.80X_2{X}_3-7.01X_1^2-4.87X_2^2-7.35X_3^2,$$

(12)

$$Y_2=5.37-1.59X_1+2.17X_2+1.47X_3+3.31X_1{X}_2-2.47X_2{X}_3+4.53X_1^2+3.91X_2^2+5.50X_3^2,$$

(13)

$$Y_3=8.49+3.01X_1-0.86X_2+2.49X_1^2+0.96X_2^2+1.85X_3^2,$$

(14)

where Y₁, Y₂, and Y₃ represent the qualified-seeding index, multiple-seeding index, and missed-seeding index, respectively; X₁, X₂, and X₃ are the coded values of operating speed, seed-dropping height, and seed-box slope angle, respectively. The significance of the regression models and individual model terms was evaluated by ANOVA based on Fisher’s F-test. Terms with p < 0.05 were considered significant and retained in the final equations, whereas non-significant terms were removed to improve the conciseness and interpretability of the models.

The ANOVA results in

Table 6. ANOVA and model-fit summary for the response surface models.

Response	Model F-Value	Model p-Value	Lack-of-Fit p-Value	R2	Adjusted R2	CV/%	Significant Terms Based on Fisher’s F-Test
Y1 qualified-seeding index	56.09	<0.0001	0.5282	0.9863	0.9687	1.56	X1, X2, X3, X1X2, X2X3, X12, X22, X32
Y2 multiple-seeding index	51.13	<0.0001	0.6250	0.9850	0.9658	8.40	X1, X2, X3, X1X2, X2X3, X12, X22, X32
Y3 missed-seeding index	53.03	<0.0001	0.8286	0.9855	0.9670	4.74	X1, X2, X12, X22, X32

verify the reliability of the response surface models. For Y₁, Y₂, and Y₃, the model p-values were all below 0.0001, whereas the lack-of-fit p-values were 0.5282, 0.6250, and 0.8286, respectively, indicating significant regression models with non-significant lack of fit. The determination coefficients were all greater than 0.985, and the adjusted R² values were all greater than 0.965. These results show that the quadratic models captured the main effects, interaction effects, and curvature of the operating parameters. The response surfaces indicated that excessive operating speed combined with a large seed-box slope angle increased the probability of multiple picking and landing dispersion, because seed-supply pressure and inertial disturbance acted simultaneously. Similar studies on precision metering for stem segments and irregular cut potatoes also showed that seed morphology, conveying constraint, operating speed, and metering-structure coordination affect qualified seeding, multiple seeding, and missed seeding [4,7,8,9,10,12,14] (Figure 7).

The response surface optimization predicted that the best performance would be obtained at an operating speed of 0.20 m s⁻¹, a seed-dropping height of 0.14 m, and a seed-box slope angle of 26.18°. Under this condition, the predicted qualified-seeding, multiple-seeding, and missed-seeding indices were 86.45%, 5.16%, and 8.40%, respectively. Compared with the single-factor results, the response surface optimization better balanced stable seed supply, seed-cup acceptance, and landing-point distribution (

Table 7. Optimized parameter combination and validation results.

Item	Operating Speed/(m s−1)	Seed-Dropping Height/m	Seed-Box Slope Angle/°	Qualified-Seeding Index/%	Multiple-Seeding Index/%	Missed-Seeding Index/%
Predicted optimum	0.20	0.14	26.18	86.45	5.16	8.40
Field validation	0.20	0.14	26.00	86.16	5.13	8.71

).

3.3. Bench Verification and Field Validation

Bench verification was first carried out under the rounded optimized parameter combination of an operating speed of 0.20 m s⁻¹, a seed-dropping height of 0.14 m, and a seed-box slope angle of 26°. The results confirmed continuous seed distribution and stable chain-driven metering before field validation.

Field validation was then conducted under the same parameter settings using the prototype matched with the SD3004 tractor. Six validation runs were completed. The measured qualified-seeding indices were 85.82%, 87.15%, 86.57%, 86.04%, 85.31%, and 87.08%; the corresponding multiple-seeding indices were 5.28%, 4.92%, 5.13%, 5.01%, 5.39%, and 5.02%; and the missed-seeding indices were 8.90%, 7.93%, 8.30%, 8.95%, 9.30%, and 7.90%. The mean qualified-seeding, multiple-seeding, and missed-seeding indices were 86.16%, 5.13%, and 8.71%, respectively.

Compared with the response surface prediction, the field validation qualified-seeding index decreased by 0.29 percentage points, the multiple-seeding index decreased by 0.03 percentage points, and the missed-seeding index increased by 0.31 percentage points. The small deviations between the predicted and field validation results indicate that the optimized parameter combination had good practical applicability. The deviations were mainly caused by seed segment size dispersion, slight field-surface unevenness, machine vibration, and soil disturbance during seed landing. These destabilizing factors suggest that future prototypes should further improve seed segment consistency control and integrate online seed-posture monitoring and missed/multiple-seeding feedback compensation.

3.4. Comparison with Existing Studies and Manuscript Contribution

Compared with published studies on related seed-metering devices, the proposed Chinese yam seeder targets a more specific object: short, soft, irregular, and elongated seed segments requiring ordered supply and oriented dropping. Arkoh et al. developed a double-row tractor-operated yam minisett planter and reported a hopper deformation of 0.442 mm and a ridger-bottom stress of 9.18 MPa, indicating acceptable structural safety [3]. For stem-segment crops, Chen et al. designed a pre-cut cassava seeding mechanism and reported a seed-filling qualification index of 94.13% [6]. For tuber crops, Wang et al. developed a potato precision seed-metering device and obtained a field-qualified-seeding rate of 91.54%, a miss-seeding rate of 3.08%, and a multi-seeding rate of 5.38% [10,11,12,13,14,15,22]. In comparison, the present Chinese yam seeder achieved a field qualified-seeding index of 86.16%, a multiple-seeding index of 5.13%, and a missed-seeding index of 8.71%. Although the qualified-seeding index of 86.16% was lower than that reported for some potato and cassava seed-metering devices, this performance is practically meaningful for Chinese yam planting because the target seed material is softer, more elongated, more irregular, and more posture-sensitive. In addition, the proposed device replaces manual ordered seed feeding and provides a basis for further improvement through seed segment grading, posture control, and online monitoring.

Although the qualified-seeding index of the present device was lower than that reported for cut potato seed metering, the target material was more elongated and posture-sensitive. Unlike yam minisett planters, cassava seeders, or potato seed-metering devices, the proposed system integrates automatic ordered seed supply, constrained chain conveying, and oriented dropping for Chinese yam seed segments. The main contribution of this study is the establishment of a complete optimization workflow involving material-parameter measurement, seed-layer stress analysis, chain-driven metering design, single-factor screening, response surface optimization, bench verification, and field validation.

The main challenges encountered in this study were seed segment size dispersion, irregular surface morphology, unstable posture during seed entry and dropping, variation in seed-supply pressure caused by the seed-box slope angle, and field disturbances such as slight surface unevenness, machine vibration, and soil disturbance during seed landing. These factors contributed to the remaining missed- and multiple-seeding errors. Future prototypes should therefore further improve seed segment grading, seed-cup adaptability, posture control, and online monitoring of missed and multiple seeding [10,11,15,20,21,22,26,27]. It should also be noted that the field validation was conducted under one soil condition, namely loam soil with a moisture content of 18–22% and a soil firmness of 1.2–1.5 MPa. Therefore, the robustness of the optimized parameters under different soil textures, moisture levels, field surface conditions, and higher operating speeds still requires further verification.

4. Conclusions

(1) A precision seeder for oriented Chinese yam planting was developed by integrating automatic seed distribution, chain-driven metering, and synchronized fertilization. The main structural parameters were determined from seed segment geometry and operating requirements: pickup-cavity diameter 35 mm, pickup-wheel diameter 120 mm, pickup-wheel thickness 100 mm, 10A chain, sprocket center distance 285.75 mm, 49 seed cups, and adjacent cup spacing 47.625 mm.

(2) The seed-layer stress model, sprocket conveying condition, seed-dropping trajectory, and theoretical plant-spacing equation clarified the functions of the three optimized parameters. Seed-box slope angle controls seed-supply pressure and multiple-picking risk; seed-dropping height affects landing dispersion and posture adjustment; operating speed determines the matching relationship between ground travel and cup discharge. The theoretical plant-spacing range of 90–130 mm covered the agronomic requirement of 80–120 mm.

(3) Response surface optimization gave an operating speed of 0.20 m s⁻¹, a seed-dropping height of 0.14 m, and a seed-box slope angle of 26.18°. Under the rounded validation setting of 0.20 m s⁻¹, 0.14 m, and 26°, six field validation runs produced mean values of 86.16%, 5.13%, and 8.71% for the qualified-seeding, multiple-seeding, and missed-seeding indices, respectively.

(4) The novelty of this study lies in the development of a dedicated seed-metering solution for short, soft, irregular, elongated Chinese yam seed segments. Compared with existing tuber or stem-segment metering devices, the proposed machine combines automatic ordered seed supply, constrained chain conveying, and oriented dropping, and establishes an optimization workflow involving material-parameter measurement, mechanism analysis, single-factor screening, response surface optimization, bench verification, and field validation. Future research should further incorporate seed segment posture detection and online monitoring of missed and multiple seeding, because recent studies have shown that optical sensing, rotary encoders, and image-based orientation detection can improve seeding-quality monitoring and posture control [10,11,15,20,21,22,26,27]. Adaptability tests under different cultivars, seed segment sizes, soil types, and higher operating speeds should also be conducted before large-scale application.