Due to the lack of accurate discrete element simulation model parameters in the design optimization process of key agricultural machinery components in the whole mechanization technology system of cotton generation, the optimization and improvement of the machine is restricted to a certain extent. Taking a cotton rootstalk–soil mixture at harvest stage in a Xinjiang cotton field as the research object, the discrete element simulation model of a cotton rootstalk–soil mixture was constructed, and the contact parameters of discrete element simulation were calibrated by combining simulation analysis with a physical test. The discrete element significant-influence parameters of cotton rootstalk–soil mixture were screened by Placket–Burman test, and the optimal range of significant-influence parameters was determined by the steepest climbing test. According to the principle of Box–Behnken test, the quadratic regression model of repose angle and significant parameters was established with repose angle as the response value. Taking the actual repose angle as the target, the Design-Expert software was used to optimize the parameters with significant influence and obtain the optimal combination of parameters. The optimal parameter combination was compared and verified by simulation experiments. The relative error between the simulated repose angle and the physical test was 2.36%. The results showed that the calibrated parameters were true and reliable, which could provide a theoretical reference for the discrete element simulation of cotton rootstalk–soil mixture in a Xinjiang cotton field.

To address the challenges of multi-scale missed detections, false positives, and incomplete boundary segmentation in cotton field residual plastic film detection, this study proposes the RSE-YOLO-Seg model. First, a PKI module (adaptive receptive field) is integrated into the C3K2 block and combined with the SegNext attention mechanism (multi-scale convolutional kernels) to capture multi-scale residual film features. Second, RFCAConv replaces standard convolutional layers to differentially process regions and receptive fields of different sizes, and an Efficient-Head is designed to reduce parameters. Finally, an NM-IoU loss function is proposed to enhance small residual film detection and boundary segmentation. Experiments on a self-constructed dataset show that RSE-YOLO-Seg improves the object detection average precision (mAP50(B)) by 3% and mask segmentation average precision (mAP50(M)) by 2.7% compared with the baseline, with all module improvements being statistically significant (p < 0.05). Across four complex scenarios, it exhibits stronger robustness than mainstream models (YOLOv5n-seg, YOLOv8n-seg, YOLOv10n-seg, YOLO11n-seg), and achieves 17/38 FPS on Jetson Nano B01/Orin. Additionally, when combined with DeepSORT, compared with random image sampling, the mean error between predicted and actual residual film area decreases from 232.30 cm² to 142.00 cm², and the root mean square error (RMSE) drops from 251.53 cm² to 130.25 cm². This effectively mitigates pose-induced random errors in static images and significantly improves area estimation accuracy.

In order to design the structure and optimize the working parameters of a straw-crushing device, and to improve the qualified rate of straw-crushing length and the rate of straw fluidity, the influence of the airflow characteristics and pressure distribution in the working state of the crushing chamber and the different structure and working parameters of the straw-crushing device on the airflow field were studied based on the computational fluid dynamics (CFD) software Fluent 19.2. The simulation results show that by changing the rotation speed of the cutter shaft, the negative pressure and mass flow rate of the inlet surface of the flow field reach the maximum values of 63.6 Pa and 1.64 kg/s, respectively, when the rotation speed of the cutter shaft is 2000 rpm, which proves that the rotation speed of the cutter shaft increases, and the feeding performance also increases. In the case of 0, 8 and 16 air knives, the peak flow velocity in the crushing chamber is 43.1 m/s, 50.1 m/s and 54.48 m/s, respectively, and the airflow in the crushing chamber is improved to a certain extent, which proves the feasibility of the air knives’ structure. In order to verify the simulation results, a field experiment was carried out. The results showed that the qualified rate of straw crushing was 94.13% and the drop rate was 4.26% under the conditions of a cutter shaft speed of 1800 rpm, a machine forward speed of 6 km/h and a height of 80 mm of the cutter off the ground. The field experiment and simulation show that the qualified rate of crushing increases with the increase in the rotation speed of the cutter shaft, but the increase rate slows down, and the random speed decreases. The drop rate decreases with the increase in tool height and tool speed. The air knife structure improves the crushing qualification rate and significantly reduces the drop rate, which verifies the simulation accuracy. This study provides a theoretical basis for the optimization of straw-crushing devices.

In view of the large difference in moisture content of cotton stalk in autumn in Xinjiang, the existing process of obtaining discrete element simulation parameters of cotton stalk is low in accuracy and complicated in operation, leading to the problems of poor universality and low accuracy in regard to the discrete element simulation parameter-calibration method in the process of mechanized transportation, throwing and returning to the field. Therefore, the experimental study on cotton stalk with different moisture content was carried out with the accumulation angle as the response value, so as to construct a parameter model that can quickly and accurately calibrate cotton stalk with different levels of moisture content. The model has high applicability and flexibility, and it can be widely used in the simulation test of various cotton field-operation machinery, such as a residual film-recycling machine, cotton picker, crushing and returning machine and other equipment. The water content–accumulation angle model was established by the cylinder-lifting method, and the correlation coefficient of the model was 0.9993. Based on EDEM 2020 software, the Hertz–Mindlin model was used to simulate the stacking angle of cotton stalk, and the rolling friction coefficient, static friction coefficient and collision recovery coefficient between cotton stalk and cotton stalk–steel were obtained. Through the Plackett–Burman test, climbing test and Box–Behnken test, three significant parameters, namely the rolling friction coefficient, static friction coefficient and static friction coefficient between cotton stalk and steel, were selected from discrete element simulation parameters to characterize the moisture content of cotton stalk, and the accumulation angle–discrete element parameter model was established. The p-value of the model was less than 0.0001, and the relative error was only 2.67%. Based on the moisture content–stacking angle model and the stacking angle–discrete element parameter model, the moisture content–discrete element parameter model was constructed. The model was verified by the cylinder-lifting method and the plate-drawing method, and the relative error was only 2.79%. Finally, the model was further verified by comparing the effect of the throwing uniformity between the mechanical simulation test and field test, and the relative error was only 4.75%. The test proves that the moisture content–discrete element parameter model is accurate and reliable, not only providing the design basis and support for the mechanization research of cotton stalk conveying and returning to the field in Xinjiang but also providing ideas for the calibration of discrete element simulation parameters of other crop straws.

Compared with the conventional horizontal conveyor-chain-type plastic-film-picking device, the longitudinal nail-tooth-chain-plate-type plastic-film-picking device developed by our team in the early stage has little tearing effect on the mulching film, and the separation effect of the plastic film and the impurity is better. With a view to further enhancing the performance of the plastic-film-picking device, this study optimized it and designed a friction-type nail-tooth-chain-plate plastic-film-picking chain that does not overload or slip and also facilitates the installation of functional components. The kinematic analysis of the picking nail teeth on the film-picking chain was carried out, and the motion equation and trajectory of the nail teeth during the operation were determined, as well as the requirements for the nail teeth to complete the mulching-film pickup. The key parameters of the plastic-film-picking system were determined by analyzing the no-leakage condition of the plastic film and the force. Moreover, the structural design and key parameters of the shovel-type film-lifting device were determined. According to the design results, a prototype was developed, and a multi-factor test of the operating parameters was carried out. The operating speed, spacing of the pickup nail teeth, depth of the film shovel into the soil, and distance between the tip of the film shovel and center of the picking drum were used as the experimental factors. The plastic-film pickup rate, impurity rate of the recovered mulching film, and traction resistance were used as the test indicators. A four-factor, five-level quadratic regression orthogonal combination experiment was conducted using the Central Composite Design (CCD). The effect of each test factor on the test index of the plastic-film-picking system was studied, the regression models were established, and the optimal parameter combination was acquired by using the multi-objective optimization method. When the working speed was 6 km/h, the pickup-nail-tooth spacing was 228.6 mm, the depth of the film-lifting shovel into the soil was 37 mm, and the distance between the tip of the film shovel and the center of the pickup drum was 130 mm, the field experiment shows that the plastic-film pickup rate was 90.12%, the impurity rate of the recovered mulching film was 8.96%, and the traction resistance was 19.905 kN. The relative errors between the test results and the predicted values of the regression models were less than 5%, indicating that the parameter optimization regression models were reliable, and the designed friction-type nail-tooth-chain-plate plastic-film-picking system met the technical requirements of agricultural-plastic-film recycling. The research results can provide a technical reference for the development of mulching-film collection machines.

To address the problems of the inaccurate seeding rate and uneven seeding in the process of dense planting of winter wheat in Xinjiang, according to the physical characteristics of the wheat seeds and the agronomic requirements of the high-yield cultivation techniques for the winter wheat “well” type, a double-hole wheel-type densely planted wheat hole sower was designed and produced. Through theoretical design and research, the structural design of the overall hole seeder and its key components was completed. The findings indicated that 5–7 wheat seeds could be planted in each hole at a 9.2 mm nest depth and 610 mm³ nest volume, which was consistent with the “well”-type high-yield dense-planting cultivation technology’s need for 400,000–500,000 basic seedlings per mu. The rotation speed and the quantity of the wave guide teeth were used as test factors and the qualifying index, replay index, and missed sowing index were used as test indicators to create the two-factor, three-level central composite design center combination test. It was possible to derive the mathematical model connecting the test factors and test indexes. The regression model underwent multi-objective optimization using the Design-Expert 13 program to determine the optimal parameters: the qualifying index was 91.24%, the replay index was 6.14%, and the missed seeding index was 2.62% when the wave guide rail had four teeth and the seed drill rotated at a speed of 40 revolutions per minute. The best parameter combinations were used for a bench verification test, and the test indicated that the qualified index was 90.25%, the replay index was 4.59%, and the missed broadcast index was 5.16%. The results demonstrated that the densely planted wheat hole seeder performs well, satisfies the requirements for winter wheat dense-planting and sowing operations, and serves as a model for the densely planted wheat hole seeders that will be optimized in the future.