Farmland particulate pollution seriously affects regional atmospheric quality, and exploring efficient field dust control strategies is an urgent need for agricultural ecological protection. This study employed a wind tunnel and online dust monitoring system to investigate the dust reduction effect of straw return in conservation tillage in Beijing farmland under varying wind speeds and precipitation levels, providing theoretical and technical support for straw coverage configuration and dust pollution control. Given the insufficient understanding of the combined impacts of straw coverage, wind speed and precipitation on farmland particulate emissions, this study examined how these key factors jointly affect fine particulate matter (PM2.5), inhalable particulate matter (PM10), and total suspended particulate (TSP) emissions. A three-factor, three-level response surface experiment modeled these relationships and identified optimal conditions for suppressing PM emissions-51.35% straw coverage, 3.96 m & centerdot;s-1 wind speed, and 32.36 mm precipitation-yielding average PM2.5, PM10, and TSP concentrations of 26.31, 31.71, and 42.43 mu g & centerdot;m-3, respectively. Field data showed that the mean absolute errors (MAEs) between predicted and measured concentrations were 0.52-5.80, 0.46-3.93, and 1.83-5.68 mu g & centerdot;m-3 for PM2.5, PM10, and TSP, respectively, corresponding to relative prediction accuracies of 90.42-97.95%, 95.03-98.52%, and 93.10-97.21%-indicating strong model accuracy. This approach enhances dynamic monitoring of straw return practices and guides rational field management. By integrating meteorological conditions and particulate emission characteristics, the model can quantitatively assess regional straw coverage and screen optimal straw mulching rates. It provides a clear data reference for decision-makers to formulate targeted dust prevention policies, standardize straw return regulation, and advance eco-friendly and sustainable agricultural production.

To support the mechanistic analysis and operating-parameter optimisation of sorghum headers, a coordinate–orientation discrete element modelling framework based on Sphero-Cylinder (SC) particles was developed, and a flexible model of sorghum plants (FMSP) was parameterised and validated using mature Jinza 22 samples. The macroscopic biomechanical parameters were calibrated through physical tests, with the relative errors for peak shear and tensile forces remaining below 5.0%. A central composite design was employed to optimise the header operational parameters. Pareto analysis indicated that H is the dominant factor influencing the harvesting loss. Within the tested design space, the optimal parameter combination was determined as a forward velocity Vct of 0.80 m s−1, a header operating height H of 442.51 mm, a reel tine velocity Vbhc of 2.28 m s−1, and a cutter frequency Fgd of 7.82 Hz. Simulation utilising these optimal parameters yielded a loss rate of 2.75%. Microscopic mechanical analysis revealed that, during the severing process, the reel tines generate a transverse feeding pressure along the Y-direction, peaking at 50.4 N, which mechanically constrains the stalks against the stationary guards to facilitate shear fracture. These results indicate that the coordinate–orientation assembly method can support SC-based full-plant modelling for harvesting simulations, providing a basis for analysing header loss mechanisms and optimising operating parameters.

Terrain-induced chassis motion can disturb the soil-entry attitude and soil engagement of seeding components on compact agricultural robots. This study develops an inverted 2RPU–RPS rallel mechanism for a maize seeding robot to regulate the end effector without levelling the entire chassis. The mechanism supports the disc opener and terminal seed tube and provides one vertical translation and two rotations. A nonlinear inverse-kinematic model, a unilateral penetration–downforce model, constrained electric-cylinder dynamics, and a coordinated feedforward–PI controller are established. The roll and pitch loops combine chassis-attitude feedforward compensation with end-effector error feedback, while the vertical loop regulates the opener downforce using a stiffness-based penetration reference and force feedback. MATLAB/Simulink simulations are conducted under isolated attitude disturbances, vertical terrain excitation, and multi-row operation. With maximum chassis roll and pitch disturbances of 4.49° and 3.35°, the end-effector RMSE values are 0.109° and 0.114°, respectively. At a prescribed downforce of 400 N, the downforce RMSE is 11.07 N and the mean disc-opener penetration is 34.99 mm. During the 300 s multi-row simulation, the mean penetration remains 34.96 mm and the downforce RMSE is 12.65 N. The results indicate that the strategy can attenuate chassis-induced disturbances and maintain stable soil engagement under the adopted modelling assumptions.

This study aims to address the issue of breakage during sowing, threshing, and storage of maize kernels, which reduces the quality of mechanical operations. Using uniaxial compression tests, the peak load of maize kernels was determined under different moisture content, axial ratio, and pressure direction conditions. A constitutive model was developed to describe the moisture content, axial ratio, peak load, and displacement of maize kernels under abdominal pressure. Based on the principle of hexagonal closest packing, a particle model was constructed to precisely represent the outer contour and internal dense structure of maize kernels. The bonding parameters of the Bonding V2 model were calibrated through orthogonal experiments, and the theoretical load peak relative to the actual error was 1.7%. Explored how the required calibration parameters should change under the condition of 0.2-0.5 mm particle radius, the parameters corresponding to the particle with the current radius r0 are p, n, and s. Once the particle radius reaches r0/m, the corresponding parameters should be updated to m2p, m2n, and s. Through response surface analysis, a regression equation was developed that quantitatively relates moisture content, axial ratio, and bonding parameters to the maximum load and displacement of the bridge. The model's relative error was 4.5%. By combining DEM-MBD simulation with bench testing, the model's mechanical characterisation performance was verified. When the seeder's forward speed is 8 km/h, the simulation results indicate that wear is most severe within the range from finger clamp closure to seed rotation of 40 degrees. At this stage, the average resultant force on the kernels was 1.03N, and the bond fracture ratio was 2.49%. It can better characterize the sowing damage process. Research can provide a model to support reducing mechanical losses during maize sowing and threshing.

The strip intercropping of soybean and maize, characterized by planting the two crops alternately in adjacent rows, has been widely promoted in several regions of China due to its potential to enhance resource utilization efficiency and overall yield. Accurate detection of crop rows and missing seedlings is essential for enabling precision field operations such as variable fertilization and targeted spraying. Monocular vision has emerged as a core sensing modality owing to its low cost and high resolution. However, the significant differences in row and plant spacing between maize and soybean, coupled with complex field conditions such as weed interference and uneven emergence, severely limit the effectiveness of traditional image processing techniques based on thresholding and geometric fitting. These methods struggle to accommodate the morphological variability of multiple crops, resulting in poor row detection precision and unreliable identification of missing seedlings. In recent years, deep learning has shown strong performance in crop row detection and object recognition tasks, particularly through multi-task networks that integrate segmentation and localization-related features. Nevertheless, most existing studies focus on single-crop scenarios and often neglect the integration of agronomic knowledge, thereby limiting their robustness and interpretability in real-world field environments. To address these issues, this study proposes a Multi-Task Geometric Regression Network for row extraction and missing seedling detection in maize-soybean intercropping systems, built upon an improved U-Net++ architecture and guided by agronomic priors. The proposed method simultaneously performs crop segmentation, row direction prediction, and generation of a missing seedling heatmap. The geometric features output by the network are subsequently processed using agronomic prior-informed post-processing and geometric fitting to finally achieve row extraction and missing seedling localization. Agronomic constraints, such as row spacing regularity, are embedded in the loss function as prior-informed regularization terms, which further enhance detection accuracy and robustness in intercropped fields. Experimental results demonstrate that the semantic segmentation achieves an average Intersection over Union (IoU) of 0.82, an F1-score of 0.86, and a pixel accuracy of 0.91. Row centerline detection attains an F1-score of 0.86 and a mean offset (MO) of 3.9 pixels. For missing seedling detection, the crop classification accuracy reaches 0.91, the average localization error (ALE) is only 2.5 pixels, and the composite detection score (CD-F1) is 0.89. Compared with single-task methods without agronomic priors, the proposed multi-task framework exhibits significant improvements in both stability and accuracy for row detection and missing seedling localization in intercropping scenarios. These results provide practical guidance for deploying intelligent visual systems in precision agriculture and intercropping management.

To enhance the operational stability and adaptability of soil-covering and compaction devices during no-till seeding on sloped farmland, this study developed a DEM-MBD coupled simulation model suited for conditions involving real-time slope variation and applied it to the design and analysis of a slope-adaptive covering-compacting device (SACCD). By coupling multi-body dynamics with the discrete element method, the model simulated the motion behavior and soil interaction process of the SACCD under different slope change rates, revealing its asynchronous response characteristics and profiling compaction mechanisms during dynamic slope transitions. Field experiments further verified the device's performance and the accuracy of the simulation. At the Lishu test site, the SACCD achieved a coefficient of variation of soil compaction of 20.4% under an average surface slope variation rate of 0.087 rad/s. At the Keshan test site, the coefficient of variation of soil compaction was 18.9%, with an almost constant slope. The differences between simulation and field results for the coefficient of variation of soil compaction were 8.32% (Lishu) and 1.02% (Keshan), confirming the reliability of the model and the rationality of the SACCD structural design. Overall, the simulation and field results demonstrate that the SACCD can maintain effective profiling posture and compaction performance under dynamic slope variation, providing a feasible approach and theoretical basis for the design and performance prediction of soil-covering and compaction devices for complex sloped farmland.

A major challenge in excessive organic fertilisation in field crops is the inability to adjust the amount of fertiliser in real time, resulting in uneven fertilisation. We propose a variable-rate fertilisation technology for fermentation fertiliser using a propeller-blade crushing mechanism and a chain-row variable-rate applicator. A variable-discharge fertiliser device was initially designed, comprising a paddle-based fertiliser-crushing unit and chain fertiliser discharge unit. A theoretical model of fertiliser discharge device was designed, and the force acting on the chain plate and main factors affecting fertiliser discharge were investigated. Based on the theoretical results, a control model considering the sprocket speed, forward speed of the fertiliser box and amount of fertiliser discharged from the device was designed using 32 groups of organic fertiliser ground application tests. The adjusted coefficient of determination (R2) was 0.97 (p < 0.01), indicating significance and applicability of precise control. The combined discrete element method–multi-body dynamics simulation was used to analyse the fertiliser transport process of the chain row in the fertiliser discharge model. The average speed of the fertiliser was 0.280 m s−1, with a theoretical error of 8.9%, while the average error of fertiliser discharge was 9.3%. Results demonstrated that the established model could accurately and reliably control fertiliser discharge. The proportional integral derivative (PID) control parameters were adjusted based on the AMESim-Simulink control model. The fuzzy PID control algorithm was selected using overshoot and adjustment time as indicators. The control algorithm’s parameters were verified using a bench test. The system response time for variable fertiliser discharge was <2 s, meeting the need of field operations. The field test showed that at an operating speed of 4–7 km h−1, the average error in fertiliser discharge was 6.3% and the fertilisation rate variation coefficient was 9.7%. Therefore, the designed variable fertiliser discharge system could achieve variable-rate fertiliser discharge and effectively improve the uniformity of organic fertiliser application in fields. Thus, a multi-sensor fusion variable fertiliser discharge control system was developed for fermentation organic fertilizer in a chain-type variable fertiliser discharge device, providing technical and equipment support for accurate variable application of organic fertiliser.

Straw mulching is an important practice in conservation tillage. Although it improves soil quality and moisture retention, it also creates strong occlusion between crops and weeds, which reduces the recognition accuracy of vision systems and limits the stability of variable rate spraying. To address this challenge, this study proposes an integrated framework that combines occlusion level detection with structural completion for maize weed segmentation and spraying control. A MetaYOLOv8-Level network is developed by incorporating an occlusion level prediction head and light weight architectural optimisation. This design enables the joint estimation of target class and occlusion severity. For both lightly and heavily occluded targets, an Amodal Mask2Former network is applied to generate complete masks through dual branch decoding and multi scale deformable attention. A heuristic post processing module is used when further refinement is needed to improve boundary continuity. Based on these perception results, a stratified variable rate spraying strategy is introduced to adjust dosage according to the estimated area and occlusion level. Experiments on a residue covered dataset and field trials in maize fields in Liaoning Province verified the effectiveness of the proposed method. The detection model achieved an mAP@0.5 of 89.6 % and an occlusion level accuracy of 87.3 %. The completion module reached a Dice of 0.809 under heavy occlusion. With the heuristic post processing module, BF1 increased by 0.088 and HD95 was reduced by up to 2.0 pixels. In field spraying tests, the proposed strategy increased the overall dose compliance rate from 67.0% (T0) and 56.7% (T1) to 82.3%, reduced ODR and UDR to 5.0% and 12.7%, respectively, and improved compliance by 16.0–28.0 percentage points under L1 conditions and by 24.0–36.0 percentage points under L2 conditions compared with the baseline strategies.

The resolution of the ground wheel encoder is a critical parameter determining the speed measurement accuracy and seeding quality of electrically driven precision planters. However, the intrinsic influence mechanism of this parameter has not been systematically quantified. A corn electrically driven seeding control system compatible with multi-resolution encoders was developed. Comprehensive bench and field tests were conducted to evaluate the effects of six resolutions (50, 100, 300, 500, 1000, and 1500P & sdot;R- 1) on seeding quality indicators, including the qualified seeding index (QFI) and the coefficient of variation for seed spacing (CVd). Results revealed a nonlinear relationship between resolution and seeding quality. The resolutions of 50, 100, 300, and 1000P & sdot;R- 1 were identified as the "performance sweet spot," where QFI consistently exceeded 97%. Within this range, the 300P & sdot;R- 1 encoder demonstrated optimal robustness, achieving a steady-state speed measurement relative error (Delta e) of only 0.31% and maintaining the lowest field QFI fluctuation. Conversely, significant performance degradation was observed in two specific zones: the "critical zone" (500 P & sdot;R- 1), where specific frequency response characteristics amplified system noise, causing high speed-measurement deviation (sigma e= 0.774); and the "bottleneck zone" (1500P & sdot;R- 1), where hardware processing limits led to severe pulse loss and a systematic error of 8.12%. Contrary to the assumption that higher resolution yields better control, this study confirms that the 50-300P & sdot;R- 1 range offers the optimal balance of accuracy and stability for small and medium-sized planters using standard microcontrollers. This selection strategy effectively avoids instability and hardware-limited bottlenecks, facilitating the cost-effective adoption of precision seeding technology.

The depth of seed burial and impact damage are critical indicators of sowing quality in wheat accelerated seeding technology. To investigate the factors influencing seed burial depth and impact damage, a simulation model of wheat seed impact and soil penetration was developed using EDEM (2018) software, and the motion of wheat seed impact into soil was simulated and analyzed to identify the main influencing factors of wheat seed impact into soil. Seeding velocity, wheat seed equivalent diameter, and soil surface energy were selected as experimental factors, while burial depth and maximum impact force were chosen as response indicators. Both single-factor tests and three-factor, three-level orthogonal tests were conducted. Single-factor simulations showed that burial depth increased with seeding velocity and seed diameter, but decreased with soil surface energy. In contrast, maximum impact force increased with velocity and diameter, peaking at low soil surface energy before declining beyond a threshold. The orthogonal test results indicated that a maximum burial depth of 26.37 mm and a maximum impact force of 0.0704 N were achieved when the wheat seed diameter was 4 mm, the seeding velocity was 65 m/s, and the soil surface energy was 0.5 J/m2. Bench tests were conducted to validate the simulation results further. The results of the bench tests were consistent with the simulation results, with relative deviations of less than 5%, indicating the reliability of the simulation outcomes. This experimental study has provided data and a theoretical basis for the selection of technical parameters and the design and application of accelerated sowing technology for wheat.

In light of the growing conflicts between human activities and land use, enhancing the precision of seeding control throughout the entire maize seeding process is crucial for ensuring seeding quality and increasing yield per unit area. However, current research on maize seeding control methods has predominantly focused on the uniform speed seeding stage. Little attention has been paid to the acceleration stage, where speed variations are more complex and impose higher requirements for speed measurement and real-time control. To address this issue, this paper develops a maize seeding control system based on the Tracking Differentiator Filter-Optimal Tracking Control (TDF-OTC) method, aiming to improve the seeding quality during the acceleration stage from both input and output perspectives of the control system. An electric-driven seeding system was built on a pneumatic precision high-speed maize planter to provide the hardware platform for implementing the TDF-OTC method. A nonlinear tracking differentiator (NLTD) based on TDF was designed to address the filtering problem of oscillatory speed measurement signals, leveraging its ability to balance tracking speed and noise reduction. This ensures accurate forward speed input for the control system. Additionally, a linear quadratic tracker (LQT) based on OTC was designed to minimize error performance metrics and compel the system's actual output to track the target output trajectory. This resolved the rapid tracking of the drastically changing target rotational speed of seed metering drive motor, ensuring accurate motor speed output for the control system. Considering the real-world conditions of accelerated seeding operations, the parameters of NLTD and LQT were determined using MATLAB Simulink to ensure optimal performance. A series of tests was conducted to evaluate the performance of the proposed method. The TDF test results demonstrated that the NLTD effectively filtered and reduced noise from oscillatory speed input signals. The accelerated response test results of OTC showed that the designed LQT outperformed PID controllers in acceleration tracking capability. Accelerated seeding test in the field, where the planter accelerated from a standstill to approximately 3.5-4.0 m/s, revealed that the TDF-OTC method achieved an average seeding qualification rate (ASQR) of 90.63% and an average coefficient of variation of seeding spacing (ACVSP) of 21.66%. Compared to the PID method, these results represented a year-on-year improvement of 12.28% in ASQR and a reduction of 14.99% in ACVSP, affirming the effectiveness of the proposed method in improving seeding quality during the acceleration stage. This study provides a valuable reference for advancements in precision seeding.

During contour no-tillage sowing, traditional no-tillage seeders face challenges due to large transverse surface undulations. To address this, a slope-adaptive covering-compacting device (SACCD) was developed. This study aimed to accurately predict the operational performance of SACCD. The Edinburgh Elasto-Plastic Adhesion (EEPA) model was calibrated using field data on soil compaction and bulk density from the black soil region of northeastern China. Plate compression tests and cone penetration tests were conducted to determine the optimal EEPA model parameters. The calibration errors for soil bulk density were 0.2 %, while the errors for soil compaction at depths of 2.5 cm and 5 cm were 1.8 % and 2.0 %, respectively. A DEM-MBD coupled simulation model was then established. It simulated the interaction between SACCD and soil particles under varying slope conditions after furrow opening. The study evaluated the effects of slope angles and the elastic coefficient of the adaptive spring on soil compaction. Optimal operating parameters for SACCD were identified. Field tests showed that SACCD significantly outperformed the longitudinal profiling covering-compacting device (LPCCD). It reduced the coefficient of variation (CV) of soil compaction by 44.9 %, improved the CV of sowing depth by 7.3 %, and increased the emergence rate by 11.6 %. These results highlight the SACCD's bidirectional profiling capability and its potential to enhance crop yields. In field tests, the soil compaction at a 5 cm depth differed by 5.2 % from simulation results, while the CV of soil compaction differed by 7.9 %.

Granulating surface straw and returning it to the field in conservation tillage can prevent straw blockage, effectively improve sowing quality, and enhance soil fertility. However, straw modelling methods must be updated for accuracy to guide the optimal engineering of mechanised strip applications. This study elucidates granulated straw's particle size and length distribution through mathematical statistical analysis. The differential in tangential and normal compression loads is ascertained via uniaxial compression, three-point bending, and shear examinations. The contact parameters amongst straw particles are calibrated using the cylinder lifting technique. A novel approach is introduced for constructing and calibrating an anisotropic bidirectional bonding model tailored explicitly for materials exhibiting significant discrepancies in tangential and normal load. Utilising the Box-Behnken test design, mathematical models correlating bonding parameters with tangential and normal loads are developed. The study further investigates the effects of variations in interaction parameters on the assessment of the load, culminating in the identification of an optimal set of bonding parameters. Simulation outcomes from uniaxial compression and three-point bending tests revealed deviations from actual target values by 5.9 % and 2.4 %, respectively, with consistent deformation patterns observed. During the validation phase of mechanised straw-breaking strip application, the average discrepancy of test outcomes is recorded at 8.6 %, affirmatively simulating the roller device's field mechanised straw-breaking strip application process. This research offers technical backing for the swift establishment of anisotropic material models and the optimisation design of key components of agricultural machinery.

Returning the entire straw to the field can lead to excessive buildup, obstructing the seeding process and reducing quality. To address this, technologies like straw granulation and strip application enhance straw decomposition and soil organic matter. However, differing sizes of granulated straw can impact seeding quality. This paper presents a straw-crushing device with a differential counter roller that breaks long straw particles during rotation. It also includes trenching shovels, covering discs, and compaction wheels for practical mechanised application. The discrete element method (DEM) was used to simulate the straw crushing capacity and analyse the factors affecting the crushing rate. The work determined that the centre distance between the two rollers is 99 mm, the teeth height H is 15 mm, teeth width b1 is 10 mm, and teeth thickness b2 is 10 mm. It was found that the minimal variation in the straw crushing rate varied between 0.77 and 1.77 kg s- 1. Optimal crushing rates are achieved when the upper roller's rotation speed ranges from 100 to 200 r & sdot;min- 1 and the lower roller's speed ranges from 300 to 400 r & sdot;min- 1. A simulation model is also developed to analyse the mechanised strip application process. Through single-factor tests and orthogonal test methods, the operational parameters were optimised. The findings indicated that an upper roller speed of 150 r & sdot;min- 1, a lower roller speed of 300 r & sdot;min- 1, a strip application depth of 150 mm, and a forward speed of 5 km h- 1 resulted in a straw crushing rate of 33.6 %. The uniformity variation coefficient of strip application is determined to be 14.4 %, which complies with the strip application requirements. Field validation of the test parameters yielded an average coefficient of variation of 14.9 %; the average crushing rate of granulated straw is 41.5 %, with an error margin of 0.5 % and 7.9 % compared to field tests. The optimised parameters achieve the necessary standards for mechanised strip application, providing valuable technical support for developing new granulated straw strip application methodologies.

Parameter calibration is a key component of the discrete element method (DEM). Limitations on the predictive accuracy of DEM parameters calibrated by in-situ simulation approaches exist but are always ignored by researchers. To determine a DEM parameter combination for accurately predicting the seeding depth in non-contact seeding, under a condition of varying vertical velocities, this study constructed a two-dimensional look-up table (LUT) of seeding depths with DEM parameters as inputs using the in-situ simulation. According to the experimental results of seeding depths in five soil bins with different water content, multiple parameter combinations for each soil bin were found in the LUT with their prediction accuracies higher than 95 % under a fixed seed vertical velocity. Although the vertical velocity was different from it in the calibration, the parameter combinations for each soil bin were expected to have at least one that could accurately predict seeding depth with the same parameter combination. Thus, the experimental results of seeding depth at varying vertical velocities for each soil bin and their simulated results were compared. The results demonstrated strong simulation-experiment correlations (R-2 > 0.88) for parameter combinations in each soil bin under varying vertical velocities, though the correlation weakened with increasing velocity. Optimal parameter combinations were uniquely identified per soil bin through correlation maximization. The validated DEM parameters showed prediction errors of 2.42-21.23 % related to vertical velocities. This research provides foundational insights for developing real-time pressure adjustment systems in non-contact wheat seeding using spatial soil property databases.

The shape of particles is a critical determinant that significantly influences the accuracy of discrete element simulations. To reduce the discrepancies between the discrete element model of wheat seeds and the actual particle shapes, and to enhance the accuracy of Computational Fluid Dynamics-Discrete Element Method (CFD-DEM) coupling simulations in gas–solid two-phase flow studies, We employed laser scanning and inverse modeling techniques to develop a three-dimensional (3D) reconstruction of the wheat seed. Subsequently, we employed Rocky DEM simulation software to develop a polyhedron model and an Angle of Repose (AOR) test model. The interval range of material parameters was determined through a series of physical experiments and subsequently employed to delineate the high and low levels of parameters for the simulation tests. The simulation parameters were calibrated using data from AOR simulation tests. The Plackett–Burman test, Steepest-Ascent test, and Box–Behnken test were conducted sequentially to determine the optimal parameter configuration. A test bench for wheat gas-assisted seeding was constructed, and a semi-resolved CFD-DEM coupling simulation model was developed to perform comparative analysis. The results demonstrated that the optimal parameters were as follows: the static friction coefficient of wheat seed was 0.15, the dynamic friction coefficient of wheat seed was 0.11694, and the dynamic friction coefficient between wheat seed and resin was 0.0797. In this scenario, the relative error of AOR was 2.3% and the maximum relative error of ejection velocity observed was 4.1%. The reliability of the polyhedron model and its calibration parameters was rigorously validated, thereby providing a robust reference for studies on gas–solid two-phase flows.