Abstract:Tea industry is a traditional characteristic advantageous industry in China, and the integration of new generation information technology and agriculture, such as big data, Internet of things, cloud computing, etc., promotes the transformation and upgrading of the tea industry to intelligence, and plays an important role in empowering and increasing the efficiency of the whole tea industry chain. On the basis of an overview of the tea industry intelligent technology system, the domestic and international research was summarized based on the application of information technology in the tea industry intelligence around the four aspects of planting, processing, testing and sales, and the key technologies to achieve intelligence in the tea industry was analyzed. Finally, it was looked forward to the future development direction of tea industry intelligence, and suggested to enhance the tea industry information technology infrastructure construction, strengthen the research and development of intelligent tea machine equipment for human-machine collaboration, pay attention to the development of tea planting and processing models, and enhance the ability of big data analysis to help tea sales, in order to lay the foundation for better use of information technology for tea industry upgrading.

Abstract:Facing the problem of difficult maneuvering of large machines during field operations and poor fitting accuracy of navigation paths in complex scenarios, a method of extracting navigation paths under the maize canopy was proposed based on deep learning and Gaussian process regression. Firstly, based on the quadruped robot collecting images of crop rows under the corn canopy, the Mask R-CNN instance segmentation method was improved, and the simple path aggregation network (Simple-PAN) was introduced into the feature fusion network, and the bottom-up path augmentation module and the feature fusion operation module were increased to improve the image context feature extraction module and the fusion capability of image context features. Secondly, the dividing line between the two sides of the area was constructed on the basis of the crop row target under the crown identified by the model, the distribution of the drooping leaves on both sides of the passable area was calculated, and the navigation path algorithm was optimized based on weighted average. The Gaussian process regression (GPR) algorithm was improved, and the DotProduct linear kernel was added to optimize the curve fitting and improve the straight line fitting effect of the GPR method. Finally, the navigation path recognition was performed on the validation set, and the average pixel deviation of the navigation paths fitted by different methods was calculated. The experimental results showed that the algorithm was able to adapt to the situation of leaf-obscuring rhizomes in corn fields, the optimized Mask R-CNN model possessed higher target segmentation accuracy under the canopy, the average deviation of the navigation line fitted based on the improved GPR algorithm was 0.7 pixels, and the average time consumed for processing a frame with a resolution of 1280 pixels×720 pixels was 227ms. The algorithm can provide navigation paths with some obstacle avoidance capability under the maize canopy to meet the requirements of real-time and accuracy of navigation. The research results can provide a technical and theoretical support for the research of navigation algorithms for intelligent agricultural equipment in the field.

Abstract:Aiming at the problems of low accuracy, poor stability, and packet loss of agricultural machines indoor localization when the signal transmission of global navigation satellite system (GNSS) was obstructed. The research of indoor and outdoor cooperative agricultural machines localization algorithm for ultra-wideband (UWB) frequency modulation technology was carries out. Firstly, the three base-station multilateral ranging and localization model was constructed, the absolute localization solve of the main base station and the transformation of the coordinates of the auxiliary base station were realized. Meanwhile, the weighted least squares high double sided two-way ranging (WLS-HDS-TWR) with the full center mass was proposed, and the WLS estimation algorithm based on Taylor series expansion solved for the center of mass of the triangle formed by the intersection of the base station ranging circles, i.e., the localization of the active vehicle. The indoor deployment mode of UWB positioning module for multi-state base station combination was innovatively proposed. The distance information of master and slave vehicle was obtained by time of flight (TOF) method. By integrating GNSS calibration position information, master vehicle coordinates information, and distance measurement information, master-slave vehicle cooperative localization was achieved. Finally, a joint simulation platform was built based on Prescan/Simulink to verify the reliability of the proposed algorithm. The indoor and outdoor collaborative localization experiments were carried out by agricultural tracked vehicles. The experiments showed that the full center mass of WLS-HDS-TWR collaborative positioning algorithm can effectively solve the problem of indoor GNSS signal loss, and the positioning accuracy was improved by 26.98% and 22.03% compared with the HDS-TWR and the full center mass of LS-HDS-TWR algorithms, meeting the needs of intelligent agricultural machines collaborative positioning.

Abstract:In order to address the issues of low control accuracy and poor anti-interference capability of farmland leveller’s hydraulic systems when operating on complex terrains, a backstepping sliding mode controller based on a hybrid extended state observer (HESO) was designed. The HESO estimated the system’s unknown states and total disturbances based on output feedback signals and compensates for disturbances in the feedforward channel. The backstepping sliding mode controller, designed based on the fast reaching law, outputed continuous and smooth control quantities, enhancing the systems robustness and overcoming the issues of system nonlinearity and parameter uncertainty. The stability of the proposed observer and controller was proven by using Lyapunov stability theory, resulting in a conclusion of uniformly bounded stability of errors. The effectiveness and superiority of the control algorithm proposed were verified through joint simulations with AMESim and Matlab/Simulink, as well as field experiments. Single flattening experiments on two wavy terrains with relatively complex geomorphology and three traversal experiments on two similar terrains showed that using the proposed control method, the average absolute error, maximum absolute error, and standard deviation of the absolute error of the elevation after flattening compared with the pre-flattening elevation were 0.053m, 0.146m, 0.037m, and 0.02m, 0.041m, 0.011m, respectively. These values were reduced by 36.35%, 28.32%, 31.37%, and 62.6%, 50%, 51.83% respectively, compared with that of the PID algorithm.

Abstract:The traditional fruit tree pruning process has problems such as high labor intensity, low pruning efficiency, and difficulty in ensuring pruning quality. An intelligent pruning robot arm for fruit trees was designed, and an intelligent pruning system for fruit trees was developed based on solid-state LiDAR and programmable logic controller, achieving automatic pruning of fruit trees. In order to verify the control accuracy of the pruning arm, independent accuracy tests and pruning target position accuracy tests were conducted on the swinging mechanical arm, lifting mechanical arm, and pruning cutting assembly of the pruning machine. The independent accuracy test results showed that the average control accuracy errors of the swinging mechanical arm, lifting mechanical arm, and pruning cutting assembly were 2.32%, 3.75%, and 2.50%, respectively. The pruning target position accuracy test results showed that the average length errors of the target positions Xb and Zb were 2.98% and 1.85%, respectively. The operating inclination angle of the pruning assembly was also determined, the average error was 4.35%, which met the accuracy requirements for fruit tree pruning. A fruit tree pruning experiment was conducted at the Aksu fruit tree planting base in Xinjiang. The results showed that the fruit tree pruning machine equipped with solid-state LiDAR can obtain real-time three-dimensional spatial information of the fruit tree. The pruning machine can formulate pruning strategies based on the information of the fruit tree crown detected by LiDAR. The excellent pruning rates of pear orchards and apple orchards were 93.3% and 86.6%, respectively. This system can effectively improve the efficiency of fruit tree pruning and reduce the labor intensity of pruning personnel.

Abstract:In view of the difficulties in picking chestnuts in hilly areas, the high intensity of manual harvesting and the low efficiency of harvesting, a small shaking branch type chestnuts dropping device with the vibration of flexible wire rope was designed. A single pendulum model was established to analyze the conditions of fruit shedding. The tangential acceleration of fruit shedding was 285.36m/s2, and the main factors affecting fruit shedding were the frequency, amplitude and time of branch vibration. The vibration characteristics of chestnut fruit branches were tested. The frequency range of chestnut fruit branches in the vertical direction was determined by sweeping frequency mode of 0~30Hz with Default Shaker hydraulic shaking table, and the standing frequency test was carried out in the frequency range of 7~15Hz. The optimal frequency range of vibrating fruit fell was found to be 10~12Hz. According to the results of the fruit branch vibration test, ADAMS was used to simulate the shaking mechanism. The results showed that when the rocker amplitude was greater than 95mm, the acceleration of the end of the fruit branch could meet the acceleration conditions of the chestnut fruit vibration, and the rationality of the parameter design of the fruit dropping device was confirmed. A three-factor and three-level field orthogonal test was designed, and the results showed that the highest fruit drop rate was the shaking frequency of 11Hz, the amplitude of 135mm, and the shaking time of 30s. In the experiment, the highest fruit dropping rate of chestnut was 91.9%, and the bark damage rate was 8.8%, which met the bark damage standard of agricultural machinery popularization and identification.

Abstract:In order to solve the problem of fast and stable spherical fruit grasping, a novel active three-finger gripper was proposed. A spherical active roller was installed at the end of the finger mechanism. A flexible-silicone membrane was attached to the active roller to increase the compliance and friction of the gripper. Each finger mechanism had two degrees of freedom, which can realize the opening and closing of the finger and the rotation of the active roller. The three fingers cooperated with each other during work, and the spherical fruit can be translated toward the gripper under the friction. The spherical fruit can be grasped quickly and stably by only touching it, without the need to precisely control the position and posture of the gripper. In order to illustrate the interactive relationship between the designed active gripper and the spherical fruit, an interaction model between them was derived. Finally, in order to verify the effectiveness of the gripper, tomatoes, apples and oranges were selected as typical spherical grasping objects, and a grasping experiment based on three types of grippers (the active gripper, the fin-ray soft gripper and the parallel rigid gripper) was conducted. The experimental results showed that the active gripper had an average first-time grasping success rate of 96.7%, an average grasping intact rate of 98.3%, and an average task time of 5.9 s, achieving relatively good performance in terms of grasping success rate, grasping quality and grasping efficiency.

Abstract:An online detection method of seeding distribution during wheat sowing based on image processing was proposed to address problems such as low manual calculation efficiency of seeding performance parameters and lack of online detection software. A criterion for adhesive seeds based on connected region area and contour perimeter was established, and an improved concave point segmentation adhesive seed method was created to count and coordinate the segmented seeds, achieving calculation and detection of seeding uniformity, accuracy, and dispersion. A seeding test bench was built and detection software was developed. The testing results showed that at different seeding rates and seeding travel speeds, the average accuracy of the improved concave point segmentation algorithm was above 95%, which was significantly higher than that of the concave point segmentation algorithm, indicating that the method had high recognition accuracy for the total number of seed particles;as the seeding rate was increased, the probability of seed adhesion was increased, and the chance of false concave points was increased, resulting in lower algorithm accuracy;as the travel speed was increased, the probability of seed deformation and distortion in the image was increased, leading to some adhered seeds being difficult or incorrectly segmented, and the algorithm accuracy also was decreased;seeding rate and seeding travel speed had no significant effect on seeding uniformity, accuracy and dispersion, which agreed with the manual calculation and measurement results, demonstrating the feasibility of this online detection method for seeding performance.

Abstract:In order to improve the quality of straw deep burial, a direct-injection straw deep burial method was proposed to address the problems of poor straw picking effect, single conveying method, low conveying efficiency, and low deep burial rate during the straw deep burial process. A direct injection straw picking, crushing, and deep burial machine was designed, and the overall structure and working principle were explained. Design and analysis were conducted on key components, and a nonprotective ring-type pickup device was designed. The movement stages of the pickup device and the trajectory of the tines were analyzed, and the phase angles corresponding to the four-movement stages were determined. A mechanical-conveying device consisted of straw conveying screws and blades, and a pneumatic conveying device mainly consisted of straw ejection fan blades and straw conveying ducts. The straw conveying capacity was analyzed, the straw conveying screw’s structural parameters were determined, and the straw conveying screw’s minimum speed was obtained. An analysis of straw conveying blades, straw ejection fan blades, and straw conveying ducts clarified that the key factors affecting straw transport effectiveness were machine forward speed, straw conveying screw rotational speed, straw ejection fan blades rotational speed, angle of straw conveying blades, and angle of straw ejection fan blades. The simulation experiment results showed an interaction between the machine’s forward speed and the rotational speed of the straw conveying screw. The optimal operating parameters were machine forward speed of 3km/h, straw conveying screw rotational speed of 1200r/min, and straw ejection fan blades rotational speed of 1600r/min. Field experiments were conducted on a direct injection straw picking and crushing deep burying machine. The results showed that the average straw picking rate and average deep burying rate were 91.02% and 90.03%, respectively. The range of wind speed and operating torque at the outlet of the straw conveying duct was 1.78~26.83m/s and 61.55~214.78N·m, respectively. All working components operated well, and the operating quality was stable. The experimental results met the design requirements and can achieve the operation of straw-deep burial and returning to the field, providing a basis for developing and improving straw-deep burial and returning machines.

Abstract:To enhance the chopping quality of the maize straw chopping machine and address issues such as the inability to adjust traditional straw chopping parameters, a straw variable-speed chopping device based on a programmable logic controller (PLC) controlled mechanism utilizing an equal diameter cam mechanism for reciprocating motion was designed. Through analyzing force variations during straw chopping underground support conditions, the primary factors influencing the effectiveness of straw chopping were identified, and the LS-Dyna software was employed to determine the range of values for these factors. Through theoretical calculations and dynamic simulation analysis, the motion trajectory of the chopping blades was optimized, and key component parameters of the chopping mechanism were determined. To improve the chopping pass rate of maize straw chopping process, a “pump-valve-motor” speed regulation model was established. A human-machine interface was designed for operational parameter settings, facilitating rapid matching of chopping device operational parameters under different straw cover conditions. Experimental trials were conducted, with forward speed and straw cover as test factors and straw chopping pass rate as the evaluation index. The results demonstrated that there was a significant increase in chopping pass rate with the increase of chopping speed when the field straw cover was constant. Conversely, with a constant chopping speed, the chopping pass rate was decreased as field straw cover was increased. Beyond chopping speed of 200r/min, the chopping pass rate reached over 90% and remained relatively constant. To further validate the experimental results, a comparative test was conducted between variable-speed and constant-speed straw chopping methods, revealing straw chopping pass rate of 92.17% for the variable-speed method compared with the constant-speed method. This device achieved precise control of operational parameters for straw chopping, providing technical support for the intelligent development of straw chopping machines.

Abstract:Aiming at the issues of large inventory but low utilization of combined harvesters in the mid-lower reaches of the Yangtze River, as well as the problems of the longitudinal dimension of the conventional three-point hitch of tractors being large and the structure being complex, resulting in a large turning radius and unstable center of gravity for the unit, a guide rail hitch device suitable for connecting a combined harvester with a rapeseed direct seeding machine was developed by using a modular design approach. Based on the kinematics analysis of the conventional three-point hitch, the overall structure and working principle of the guide rail hitch device were determined from the aspects of compact structure and stable center of gravity. According to the matching relationship between the guide rail hitch device and the combine harvester and rapeseed direct seeding machine, theoretical analysis was carried out from the matching of width, center of gravity and tillage depth, and the operating parameters and structural parameters of the hitch device were determined. The ANSYS Workbench simulation software was used to carry out the modal analysis of the hitch device in a free state. By analyzing the characteristics of the external excitation frequency of the frame and the simulation results, an improvement plan for the structure of the hitch device was proposed. And the first six frequencies of the optimized device were 54.09Hz, 66.35Hz, 83.16Hz, 130.01Hz,143.52Hz and 174.88Hz, which were not within the range of the external excitation frequency. The static analysis of the optimized hitch device showed that the stress values of the suspension device at the highest lifting point and operating state were 50.531MPa and 140.56MPa, respectively, which were less than the allowable stress of 156.67MPa. The field test results showed that the whole machine ran stably. When the field test was carried out with the forward speed of 3.6 km/h, the stability of the ploughing depth was 87.42%, the soil breaking rate was 84.41%, the straw burying rate was 76.32%, and the flatness of the cage was 22.93mm. The test results showed that all the test indexes met the quality requirements of rape sowing.The research result can provide an idea for the research of a multi-purpose device.

Abstract:Aiming to address the lack of an accurate simulation model for the compression and crushing process of cottonseed, based on the discrete element method, the Xinluzao 84 cottonseed was used as the research object, and its parameters were calibrated through the combination of physical tests and simulations. According to the actual morphological characteristics of cottonseed, the volume of cottonseed was calculated by using the average particle size to represent the size of cottonseed in the vertical length direction, and the result was less error with the volume obtained by using the triaxial dimensions, which proved that it was reliable to use the average particle size to describe the size of cottonseed in the vertical length direction. The polyhedral model of cottonseed was rapidly constructed by using the mesh modelling function in Solidworks 2022.The correlation was established between the volume relative error, simulation time, and the number of facets of the model. It was determined that the optimal number of facets for the polyhedral cottonseed model was 1.098, with corresponding simulation time of 171 minutes and volume relative error of 0.46%. The stacking angle test was employed to calibrate the interspecific parameters of cottonseed. The results indicated that the collision recovery coefficient of cottonseed-cottonseed, the static friction coefficient of cottonseed-cottonseed, and the rolling friction coefficient of cottonseed-cottonseed had a significant effect on the stacking angle. The optimal parameter combinations were determined to be 0.106, 0.248, and 0.105, respectively. Relative error of 0.28% was observed between the simulation stacking angle and the actual stacking angle when using these parameter values, demonstrating the accuracy of the cottonseed parameters. The parameters of the Tavares model were calibrated and verified by the indexes of cottonseed crushing force and crushing energy through the single particle compression test. The results showed that the relative errors of cottonseed crushing force and crushing energy were 2.37% and 2.87%, respectively. This indicated that the constructed cottonseed model and the calibrated parameters of the Tavares model were able to effectively characterize the compression and crushing process of cottonseed.

Abstract:Due to the lack of accurate discrete element model parameters in the design and optimization process of the key agricultural machinery components of the plough layer residual film recycling machine, the research on the interaction mechanism between the plough layer residual film recycling machine and the residual film was restricted to a certain extent. Taking the residual film in the tillage layer of cotton field as the research object, the residual film in the tillage layer content and ultimate tensile force were measured, and the residual film content and ultimate tensile force of different tilting depth and thickness were obtained. According to the measurement results, the Hertz-Mindlin with Bonding contact model was selected for discrete element model parameter calibration by using EDEM software. The normal stiffness per unit area, tangential stiffness per unit area, critical normal stress, critical tangential stress，bond radius and contact radius were selected as test factors. Through Plackett-Burman test, the steepest climbing test and Box-Behnken test were used in turn. Finally, the significant parameters of the optimal Bonding model were determined as normal stiffness per unit area, critical normal stress and bond radius, whose values were 2.36×105N/m3, 6.47×104Pa and 0.004mm, respectively. The simulation test was carried out to verify the parameters, and the error value was 5.88%, which met the requirements. Through the comparative analysis of the residual film in the tillage layer state and tensile curve of the physical testing and simulation test in the stretching process, the rationality of the residual film model in the tillage layer was proved, which provided theoretical support for the later residual film in the tillage layer recycling machine to simulate the mechanical recovery of residual film in the tillage layer and analyze the separation mechanism of film and soil.

Abstract:Aiming at the demand for precision seeding of wheat and the uniformity seed spacing, a precision seed metering device with staggered hook-tooth was designed. The hook tooth were used to grab seeds orderly, and the staggered arrangement of the teeth allows the falling seeds to form a staggered and ordered surface, reducing collision and overlap between seeds. The key structural parameters and curve profile of the hole were determined by analyzing the seed postures during seed-filling process. The effects of the rotational speed of the seeding wheel and the angle of the seed-filling area on the operation performance were analyzed by using discrete element method EDEM. The simulation results showed that the speed of the seeding wheel had a significant effect on the seeds postures. The increase in the angle of the seed-filling area was conducive to improving the seed-filling performance. However, too large an angle of the filling area may cause the successfully filled seeds to fall out of the hole. On this basis of that, laboratory tests were conducted. Qualified rate, single rate, and multiple rate were used as performance indicators. Rotational speed of seeding wheel, height of seed-filling area and speed ratio of brush/seeding wheel were the test factors. The regression equations of the qualified rate, single rate, and multiple rate were obtained. Through the regression equations, it was found that the optimum performance of the seed metering device was achieved. A validation tests were carried out using the optimal parameters at the seeding wheel speed of 18r/min, the height of seed-filling area of 73mm, and the speed ratio of brush/seeding wheel of 2.5. The qualification rate was 90.36%, and the single rate was 87.26%, which verified the reliability of the regression model. Comparison tests were carried out by selecting the seed metering device commonly used in wheat planter. A high speed camera was used to capture the process of seed dropping. The metering device designed realized the staggered and orderly seeds falling, which improved the uniformity of seed distribution. The research result can provide theoretical support for the development of wheat precision seeding technology.

Abstract:Considering the practical problems that the inadequate seed supply and discharge capacity at high speed operations in the existing mechanical centrifugal metering device for rapeseed which led to mismatch between the amount of seed row and the operation speed, and poor seed row performance, a type of metering device with passive seed filling and “round hole+gradient hole column” combined type holes for seed discharge was designed. The mechanical model of seed supply and discharge process was constructed and the key structural parameters affecting seed discharge performance were analyzed and determined. The experiment on the influence of the seed supply adjustable height on the adjustment range of the seed supply rate was simulated by using EDEM discrete element simulation software, which revealed the seed supply rate was adjustable in the range of 64.95~357.54g/min when the seed supply adjustable height was in the range of 3~8mm. The influence of the hem height and inclination angle of the seed limiting sleeve on the initial seed quantity, critical speed and the maximum height of the annular seed layer was analyzed by using two factors and three levels orthogonal test. The optimum structural parameters of seed limiting sleeve were determined by bench test. Combined with high-speed photography, the relationship between the rotating speed of the moving cone and the seed metering amount under five kinds of hole structures was compared, and the optimal hole structure was determined as “round hole+gradient hole column” combined type holes. The bench verification test results of the metering performance of the metering device with better parameter combination at different rotating speeds showed that when the speed of the metering device was 115~205r/min, the seeding amount rate was 60.96~355.76g/min, the apiece row consistency variation coefficient was less than 5.2%, the single row stability variation coefficient was less than 1.3%, and the damage rate was less than 0.5%, which can meet the working speed of 6~12km/h. The field experiment showed that when the planting speed was 7.89km/h and 11.98km/h, the apiece row plant distribution consistency of rapeseed was less than 11%, and the planting density was 43~58 plants/m2, which could meet the seeding performance requirements of rapeseed.

Abstract:The existing darkroom mode of the automatic tray winch has encountered problems such as a decline in qualification rate and working stability, as well as damage to seedlings and rice trays during transportation. To address these issues, an improved hook rod lifting device with axial auxiliary push function for rice trays was designed. A simulation single factor test platform was constructed to determine the main parameters and their value range that affected its performance. The structure parameters of key components were determined, and a multi-case target evaluation result function was established by using Matlab interpolation method for standardization of evaluation index. The quadratic regression orthogonal rotating center combination test method was adopted, with tray spacing, transverse conveying speed, and lifting angle as test factors, while tray lifting qualification rate, leaf damage rate, hard disk damage rate, and number of injured seedlings served as performance evaluation indexes. Parameter combination optimization tests were conducted on the Rocky-DEM simulation test platform that had been constructed. Design-Expert 11 software was used for multi-objective parameter combination optimization. Field verification tests under conditions of 190mm tray spacing, 0.20m/s transverse conveying speed, and 72° lifting angle achieved a tray lifting qualification rate of 97.5%, leaf damage rate of 0.52%, hard disk damage rate of 0.45%, number of injured seedlings at 9 units, with an error of less than 1% compared with the theoretically optimized value. In comparison with the non-optimized lifting and stacking device, the pass rate for tray lifting was increased by 5.6 percentage points, leaf damage rate was decreased by 3.5 percentage points, reducing seedling damage, and hard tray damage rate was decreased by 3.8 percentage points.

Abstract:Aiming at the problems of high resistance and high injury rate in mechanized harvesting of Panax notoginseng in hilly and mountainous areas, taking different shovel types of excavator as the research object, comparative experimental research on the operation mechanism and parameter optimization of excavator shovel of Panax notoginseng harvester was carried out. Through theoretical analysis and mechanical calculation, the main structural parameters of each shovel type excavator were determined, and the three-dimensional model of the excavator was established by Solidworks. The simulation and comparison tests of displacement and flow direction of four shovel types of excavator were carried out to track the spatial motion trajectory of rhizome particles and soil particles, and the triaxial displacement segments were obtained. It was showed that the special curvature change of the bionic curve of the boar head had a significant impact on the soil breaking resistance, and the bionic shovel surface reduced the excavation resistance by changing the flow direction of soil particles. By using the Design-Expert 13 software, taking the shovel length, width and blade inclination as the test factors, and taking the excavation resistance as the test index, the combination test was carried out. The optimal combination of operation parameters was obtained as the shovel length of 354mm, the shovel width of 40mm, and the blade inclination of 70°. Under this combination, the average excavation resistance was 439.75N. The high-speed photography experiment was carried out to obtain the movement trajectory of Panax notoginseng rhizome and soil. The results showed that the movement trend of Panax notoginseng rhizome and soil particles was consistent with the simulation experiment, which verified the reliability of the discrete element model. The orthogonal test was conducted with the penetration angle, blade spacing and excavation speed as the test factors and the excavation resistance as the test index. The results showed that the primary and secondary order of the influence of different working parameters on the excavation resistance was the penetration angle, blade spacing and excavation speed. The optimal combination of working parameters was the penetration angle of 15°, blade spacing of 80mm and excavation speed of 0.4m/s. Through comprehensive evaluation of the performance of reducing viscosity and resistance, the bionic excavator had the best operation effect.

Abstract:Aiming at the problems such as big damage to pepper and unclean removal of pepper pedicel that exist in the mechanised pedicel removal of pepper, the design of a double roller dried pepper pedicel removal device was introduced, taking into consideration of the physical and mechanical properties of pepper and the requirements of mechanized processing. The key components of the device included three parts: drum, pedicel removal roller and screen mesh, using rotating drum and rotating roller to achieve pepper pedicel removal. The process and working principle of the pedicel removal mechanism were thoroughly analyzed to determine the critical conditions for pedicel removal of pepper. Through parameter design and mechanical analysis, optimal specifications were determined. The roller’s diameter was set at 1.8m, with a length of 6.0m and a rotational speed of 28.83r/min. The pedicel removal roller, on the other hand, had a diameter of 36.0mm, a clearance of 0.5mm, and a rotational speed of 134.21r/min. A square screen with high screening efficiency was selected based on its screen surface utilization factor, which was determined to be 60.5%. Through the Design-Expert 12 software for experimental design and data processing, the influence of drum speed and pedicel removal roller speed on the damage rate and cleaning rate was clarified, and the optimal parameter combinations were determined to provide theoretical basis for the research and development of pepper mechanized processing technology and equipment. Subsequent pepper pedicel removal tests were conducted, and the results demonstrated that under the test conditions, the device achieved a damage rate of 0.6% and a cleaning rate of 98.8%. These results met industry standards and fulfilled the requirements for pepper processing.

Abstract:The feeding sagittal angle of the forage harvester has a significant impact on the torque load and cutting performance of the cutting roller. By detecting the relationship between the transmission belt pressure and the feeding sagittal angle of the crop, kinematic and dynamic analysis was conducted on the cutting characteristics of the harvester and the crop cutting process. The relationship between the cutting quality, energy consumption, and the feeding sagittal angle of the forage crop was revealed. A mathematical model for the feeding sagittal angle of the harvester’s cutting roller was established and optimized. The experimental results showed that when the crop feeding amount in the optimized parameter group was 5.5kg/s, the cutting roller speed was 1600r/min, and the initial pressure of the transmission belt was 1.2MPa, the mathematical model accuracy of the cutting crop feeding sagittal angle under this operation parameter group was the highest, with an accuracy of 79.52% and a model deviation of 0.491MPa (0.957rad), which can effectively represent the cutting load of the hay cutter. At the same time, the cutting ability of the hay cutter was improved while ensuring the quality of crop cutting. Under the optimized cutting feeding sagittal angle model, the cutting disturbance response time was shortened by about 26.41%, and the unit operation energy consumption was reduced by about 11.36%. The establishment of the sagittal angle model for feeding the chopped crops enriched the modeling method for the cutting operation of the hay cutter, providing a reference for the design of the load control system of the hay cutter for similar forage crops.

Abstract:The change of overflow boundary morphology with erosion time is important to objectively and realistically reflect the erosion characteristics of double suction centrifugal pumps and erosion morphology. The Euler-Lagrange method and the dynamic boundary method of geometric reconstruction of the erosion wall were used to calculate the solid-liquid two-phase flow of the double suction centrifugal pump of Jingtaichuan Pumping Station in Gansu Province under the average sand content and the value of grain size of the Yellow River. The progressive erosion characteristics of the pump vanes were predicted, and the effects of the vane erosion mechanism and the change of wall geometry on the pump performance were analyzed combining the experimental data. The results showed that taking the maximum value of the impact angle function corresponded to the impact angle α0 as a threshold to distinguish erosion patterns. If impact angle was less than the threshold, the erosion pattern was liking a rounded crater, if it was bigger than the threshold, the erosion pattern was liking a groove. The erosion rate was high in areas where the impact angle was in the range 50°~75° and the impact velocity was high, which both led to high level of blade erosion. The prediction period was divided into three phases based on the characteristics of the rate of change of the hydraulic performance loss. The erosion rate had the largest growth rate in the early stages, but the values were orders of magnitude smaller than those in the middle and late stages, so that the head loss rate, efficiency loss rate and mass loss rate of blade in the first 1000 hours of erosion was less than in other stages. All these parameters showed a trend of slow growth in the early stage, fast growth in the middle stage and slow growth in later stage, and maximum growth rate in the middle stage, values of 1.51×10-3, 1.97×10-3, and 4.12×10-3 respectively. Erosion caused the fastest performance degradation of double suction centrifugal pump in the 1000 hours to 6000 hours erosion length range.

Abstract:Protecting high-quality concentrated and contiguous farmland is crucial for ensuring national food security. Taking Xingguo County in Jiangxi Province as a case study, an index system for evaluating comprehensive farmland quality was systematically constructed based on natural quality, site conditions, and ecological conditions. Utilizing the technique for order preference by similarity to an ideal solution (TOPSIS) comprehensive evaluation method, farmland quality was assessed. Farmland spatial aggregation characteristics were analyzed through farmland contiguity and spatial connectivity patterns, leading to the demarcation of permanent basic farmland based on comprehensive quality and connectivity. Results indicated that Xingguo County’s farmland quality was divided into 4 grades, with areas of 6204.95hm2, 16031.72hm2, 19321.79hm2, and 3573.76hm2, representing 13.75%, 35.52%, 42.81%, and 7.92% of the total farmland area respectively, with the majority falling in the medium quality range at 78.33%. In terms of farmland connectivity, the area was classified into 5 contiguity levels, with areas of 24731.44hm2, 6199.73hm2, 3131.54hm2, 7397.71hm2, and 3671.80hm2, accounting for 54.80%, 13.73%, 6.94%, 16.39%, and 8.14% of total farmland area, revealing varying degrees of fragmentation. Farmland of grade three or above and contiguity level four or above was designated as permanent basic farmland, covering 37029.62hm2, or 82.05% of the total area. This delineation achieved the goal of “overall stability, optimized layout, and improved quality” compared with previous permanent basic farmland designations.

Abstract:In order to quantify the difference in mechanical tillage efficiency of plots with different areas and shapes, and scientifically evaluate the potential of land improvement in mechanical tillage efficiency, a set of standard plots was designed, and a 69.83kW Dongfanghong LX950 tractor supporting rotary tiller was selected for tillage, with a working width of 200cm. The mechanical rotary tillage efficiency of plots with different area gradients and shapes was measured by field experiment. On this basis, the coupling relationship and function model of the plot area, shape and mechanical rotary tillage efficiency were established based on regional scale, and the regional mechanical rotary tillage efficiency was modified. The results showed that the regularity of plot shape affected the mechanical rotary tillage efficiency. The more regular the shape was, the higher the mechanical rotary tillage efficiency was. In the same area gradient, the mechanical rotary tillage efficiency of rectangular plots was the highest, followed by trapezoidal plots, and triangle plots were the lowest. With the increase of the land area, the mechanical rotary tillage efficiency was increased continuously. When the land area reached a certain degree, the mechanical rotary tillage efficiency tended to be stable. When the land area exceeded 7000m2, the mechanical rotary tillage efficiency basically remained unchanged. The mechanical rotary tillage efficiency of rectangular and trapezoidal plots was higher, maintaining at 0.25s/m2;the mechanical rotary tillage efficiency of other shaped plots was lower, maintaining at about 0.30s/m2. The mechanical rotary tillage efficiency at the unit level in the case area was 0.27s/m2. Among them, the number of low-efficiency land units was the least, accounting for 7.10%. The number of efficient land units was the largest, accounting for 58.24%. The mechanical rotary tillage efficiency in the western part of the case area was greater than that in the eastern part of the case area, mainly because the land fragmentation degree in the eastern part was higher than that in the western part, the land unit area was smaller and the shape was not regular, which greatly reduced the regional mechanical rotary tillage efficiency value. The experimental results showed that plots with an area greater than 7000m2 were suitable for large-scale mechanized farming, and both the plot area and the regularity of the shape affected the efficiency of mechanical rotary farming. While expanding the plot area, the land regulation project also needed to ensure the regularity of the plot shape.

Abstract:As one of the economic forest species in Jiangxi Province, Camellia oleifera is also a characteristic advantageous industry in Jiangxi Province, and it is of great significance to accurately obtain its spatial distribution in terms of yield estimation, production management and policy formulation. In response to the lack of optical images due to the cloudy and rainy climate in the south, as well as the problem of fragmented terrain in hilly and mountainous areas, Yuanzhou District, Yichun City, Jiangxi Province, was taken as the study area. Using time-series Sentinel satellite imagery and SRTM DEM data as data sources, a total of 125 feature variables were constructed and selected, including spectral features, vegetation-water indices, red edge indices, radar features, terrain features and texture features. Among them, the texture features were calculated by comparing 15 different scale windows by using the cumulative difference method to calculate the best texture features for Sentinel-1 and Sentinel-2 images. Based on ReliefF feature preference algorithm and random forest classification algorithm, eight feature combination schemes were designed to carry out experiments to explore the impact of different feature types on the extraction accuracy of Camellia oleifera. The results showed that the optimal texture feature window for both Sentinel-1 and Sentinel-2 calculated experimentally by using the cumulative difference method was 35×35, and the optimal texture feature combinations were mean, variance and contrast. Building upon spectral features and vegetation-water indices, the incorporation of different features for Camellia oleifera classification demonstrated varying degrees of effectiveness. The favorability ranking of different feature types for Camellia oleifera extraction from large to small was as follows: S2 texture features, S1 texture features, terrain features, radar features and red edge index. Compared with single-spectrum and index features, the inclusion of texture features significantly enhanced classification accuracy. The synergistic classification results of multiple features surpass those of single-feature classification, with the highest precision achieved through Camellia oleifera extraction based on feature selection. The ReliefF algorithm feature optimized scheme had the highest accuracy with overall accuracy of 88.29% and Kappa coefficient of 0.81. This study utilized time-series Sentinel satellite imagery and DEM terrain data to develop a large-scale remote sensing extraction method for Camellia oleifera in the cloudy and rainy southern hilly mountainous region. This method can serve as a reference for the investigation and monitoring of Camellia oleifera resources in the hilly areas of southern China.

Abstract:In order to reduce the influence of canopy geometry and other factors on the sensor detected solar-induced chlorophyll fluorescence (SIF), the response characteristics of red SIF (RSIF) fluorescence under stripe rust stress were discussed. Simple linear regression (SLR) and non-linear regression (NLR) models for remote sensing monitoring of wheat stripe rust were constructed with RSIF as the independent variable. The results showed that the leaf scale RSIF had a significant advantage in the remote sensing monitoring of wheat stripe rust, with a 13.2% higher correlation with the severity level (SL) of wheat stripe rust compared with far-red SIF(FRSIF). Compared with FRSIF, the R2 between predicted DSL and measured DSL was increased by 9.8% and 38.9%, and the RMSE was decreased by 23.1% and 36.4%, respectively, using the linear regression model and non-linear regression model constructed with the blade scale RSIF as the independent variable. In addition, downscaling can improve the accuracy of RSIF monitoring of wheat stripe rust. The R2 between leaf scale RSIF and DSL was increased by 126.3% compared with the canopy scale. The R2 between DSL and measured DSL predicted by SLR and NLR models using leaf scale RSIF as independent variable was increased by 114.3% and 233.3%, respectively, compared with the canopy scale, and RMSE was decreased by 16.7% and 15.4%, respectively. The research results were of great significance for improving the remote sensing monitoring accuracy of wheat stripe rust, and also had certain reference value for remote sensing monitoring of other stresses.

Abstract:Rapid and accurate estimation of wheat yield can improve the efficiency of breeding. Yield data of wheat lines and hyperspectral data during grain filling period were collected. Firstly, the feature wavelengths were selected as model input variables by using recursive feature elimination method. Then three linear algorithms (ridge regression, partial least squares regression, multiple linear regression) and six nonlinear algorithms (random forest, gradient boosting regression, eXtreme gradient boosting, Gaussian process regression, support vector regression, K-nearest neighbor) were employed to establish single algorithm yield estimation models for precision comparison. Finally, the Stacking algorithm was adopted to develop multi-model ensemble combinations, aiming to identify the optimal ensemble model. The results showed that the accuracy of yield estimation models, based on different algorithms, varied significantly, and that the nonlinear models were better than the linear models. The yield estimation model based on GBR performed best in the single models, with R2 of 0.72, RMSE of 534.49kg/hm2 and NRMSE of 11.10% in the training set, R2 of 0.60, RMSE of 628.73kg/hm2, and NRMSE of 13.88% in the testing set. The performance of the ensemble models based on Stacking algorithm was closely related to the selection of primary and secondary models. The model with KNN, RR, SVR as primary models and GBR as the secondary model effectively improved the yield estimation accuracy. Compared with the single model GBR, the training set R2 was increased by 1.39% and the testing set R2 was increased by 3.33%. The research result can provide an application reference for yield estimation of wheat lines based on hyperspectral technology.

Abstract:In response to the low accuracy of jujube variety recognition in current natural scenarios, a jujube variety recognition model was proposed based on attention mechanism and multi-semantic feature enhancement (ICBAM_MSFE_Res50). On the basis of ResNet-50, the attention mechanism ICBAM (improved convolutional block attention module) was introduced. ICBAM improved the convolutional block attention module (CBAM) by using one-dimensional convolution and multi-scale hole convolution, eliminating information loss during feature map dimensionality reduction, reducing the computational and parameter complexity of the model, and improving the model’s ability to extract fine-grained features in jujube fruit regions. At the same time, a multi-semantic feature enhancement (MSFE) module was proposed, which extracted more local salient features of jujube fruit through jujube fruit region localization algorithm, and used saliency feature suppression algorithm to force the model to learn secondary features of jujube fruit, thereby achieving the learning of multiple semantic features of jujube fruit. The experimental results showed that the accuracy of the model on the dataset of 20 types of jujube varieties was 92.20%, which was 4.26 percentage points higher than that of ResNet-50. Compared with the AlexNet, VGG-16, ResNet-18, and InceptionV3 models, the accuracy was improved by 15.84, 9.22, 6.86, and 3.55 percentage points, respectively. Compared with other jujube variety recognition methods, this method still performed the best in the recognition of 20 types of jujube, which can provide reference for research on jujube variety recognition in natural scenarios.

Abstract:To achieve rapid and accurate identification of fusarium head blight on wheat spikelets in complex field background, a wheat fusarium head blight image dataset comprising a total of 640 images across three growth stages: flowering, grain filling, and ripening of winter wheat was constructed. Additionally, a wheat spikelet fusarium head blight recognition method based on an improved YOLO v8s model was proposed. Firstly, using the omni-dimensional dynamic convolution (ODConv) to replace the standard convolution in the backbone network enhanced the network’s extraction of features from target regions and suppressed interference from cluttered background information. Secondly, an improved Efficient RepGFPN feature fusion network was utilized in the neck network to integrate low-level features with high-level semantic information, enabling the model to extract richer feature information. Lastly, the enhanced intersection over union (EIoU) loss function was employed instead of the complete intersection over union (CIoU) loss function to accelerate model convergence speed and further improve model accuracy, thus achieving rapid and accurate identification of fusarium head blight on wheat spikelets. Model validation on a self-built dataset revealed that the improved model (OCE-YOLO v8s) achieved a detection accuracy of 98.3% for fusarium head blight on wheat spikelets, which was an improvement of 2 percentage points compared with the original model. Compared with Faster R-CNN, CenterNet, YOLO v5s, YOLO v6s, and YOLO v7 models, the OCE-YOLO v8s model achieved improvements of 36 percentages, 25.7 percentages, 2.1 percentages, 2.6 percentages, and 3.9 percentages, respectively. The OCE-YOLO v8s model effectively met the requirements for precise detection of fusarium head blight on wheat spikelets and could provide valuable insights for real-time monitoring of crop diseases and pests in complex backgrounds of field environments.

Abstract:Due to the rich genetic diversity of maize germplasm resources, the size, morphological structure and color of tassels were quite different. The resolution of maize tassel image collected by UAV equipped with visible light sensor was lower than that of ground acquisition, and some tassels in the image were too small, which were highly similar to the background, occluded and interlaced. The above factors led to low accuracy of tassel detection. Therefore, a tassel detection method for maize germplasm resources based on improved YOLO v7-tiny model was proposed. This method enhanced the model’s ability to extract tassel features by introducing SPD-Conv module and VanillaBlock module into YOLO v7-tiny, and adding ECA-Net module. Tested on the self-built tassel dataset of maize germplasm resources, the mean average precision of the improved YOLO v7-tiny was 94.6%, which was 1.5 percentage points higher than that of YOLO v7-tiny, and 1.0 percentage points and 3.1 percentage points higher than that of the lightweight models YOLO v5s and YOLO v8s, respectively. This method significantly reduced the occurrence of missing tassels and false detection of background as tassels in the image, and effectively reduced the misdetection of a single tassel as multiple tassels and the number of tassels in interlaced state. The model size of the improved YOLO v7-tiny was 17.8MB, and the inference speed was 231f/s. The proposed method can improve the accuracy of tassel detection under the premise of ensuring the lightweight of the model, and can provide technical support for the real-time and accurate detection of tassel of maize germplasm resources.

Abstract:Image-based on-site detection technology for rice blast relies on prior knowledge which is affected by computational power and field network conditions, rendering adaptive real-time detection impossible. To tackle these challenges, a Mask R-CNN (Mask region-based convolutional neural network) model for rapid, high-throughput, and adaptive identification of rice blast was proposed. This model can be accelerated by using field programmable gate array (FPGA). Firstly, the backbone network was replaced with MobileNetV2, leveraging its inverted residual module to decrease computations and enhance the model’s parallel processing capabilities. Following that, a feature pyramid network module was incorporated to facilitate multi-scale feature fusion for rice blast, enabling the model to possess multi-scale adaptive processing abilities. Finally, the fully convolutional network(FCN) branch outputed the instance segmentation of rice blast lesions, utilizing the Softmax function to accurately localize and classify rice blast diseases. The validation results of the model using test datasets for rice blast disease demonstrated significant capabilities: when the input was a full HD image, the average inference time of the model was reduced to 85ms, which was 86.2% and 63.0% faster than the GPU server and the same level GPU edge computing platform, respectively. When the intersection over union ratio was 0.6, the accuracy can reach 98.0%, and the disease spot capture ability was improved by 21.2% on average. The Mask R-CNN adaptive fast identification model proposedcan realize the rapid field detection of rice blast disease under severe network conditions, and had better anti-noise ability and robust performance, which provided an efficient real-time system-on-chip scheme for real-time detection, inspection and mitigation of rice disease.

Abstract:Traditional manual methods for identifying dead ducks within large-scale stacked cage poultry houses have proven to be inefficient, labor-intensive, and costly. Focusing on stacked cage housing for meat ducks, a deep learning-based method was proposed for dead duck recognition. To collect the necessary dataset, a specialized autonomous inspection system tailored for meat duck housing within three-dimensional stacked environments was initially designed. To address the issue of severe wire mesh obstruction within the cage housing, machine vision techniques were employed to repair the cage mesh and enhance images by using OpenCV. A dead duck recognition model was constructed based on Mask R-CNN, and further optimized with the Swin Transformer to overcome the limitation of Mask R-CNN’s global information integration. The accuracy of dead duck recognition among the SOLO v2, Mask R-CNN, and Mask R-CNN+Swin Transformer models was compared and analyzed. Experimental results demonstrated that under the condition of mAP value of 90%, the Mask R-CNN+Swin Transformer model achieved an overall dead duck recognition rate of 95.8% within the duck cages, outperforming other mainstream object detection algorithms on the autonomous inspection equipment.

Abstract:Aiming at the low identification accuracy of cows with solid color or less pattern in pattern-based individual identification of cows, an individual identification method was proposed based on cow gait features. Firstly, the backbone network of DeepLabv3+ semantic segmentation algorithm was replaced by MobileNetv2 network. The channel and space based CBAM attention mechanism was introduced into this segmentation algorithm. The improved model was used to segment the silhouette of the cow. Then the 3D convolutional neural network (3D CNN) and the bidirectional long short-term memory network (BiLSTM) were constructed as the 3D CNN-BiLSTM network. The adaptive temporal feature aggregation module (ATFA) was further integrated into the above network to generate the 3D CNN-BiLSTM-ATFA cow individual identification model. Finally, individual identification experiments were conducted on a total of 1242 video datasets from 30 cows. The results showed that the MPA, MIOU and Accuracy of the improved DeepLabv3+ algorithm were 99.02%, 97.18% and 99.71%, respectively. Individual recognition was optimal when r3d_18 was used as the backbone network of 3D CNN-BiLSTM-ATFA. The average accuracy, sensitivity and precision of individual identification based on cow gait were 94.58%, 93.47% and 95.94%, respectively. Individual identification experiments with weighted feature fusion for torso and legs showed that identification accuracy can be further improved. Lameness in dairy cows had a significant effect on gait identification, the individual identification accuracies were 89.39% and 92.61% for cows that changed from healthy to lame and cows that remained lame during the experiment, respectively. The results can provide technical reference for intelligent individual identification of dairy cows.

Abstract:The daily behaviors of sheep, such as standing, walking, eating, drinking and sitting, are closely related to their health. Efficient and accurate recognition of sheep behaviors is crucial for disease and health detection. To address the current problem of the limited behavior of sheep caused by contact devices such as sensors and lower accuracy caused by diverse behaviors, complex scenarios, and occlusions in group farming, the method for sheep behavior recognition based on improved YOLO v8s was proposed. Firstly, the SPPCSPC was introduced to improve the feature extraction ability and the detection accuracy of the model. Secondly, the P2 detection was used to enhance ability of the model to identify and locate the small targets. Finally, multi-scale lightweight modules PConv and EMSConv were introduced and the number of parameters and calculation of the model were reduced and the lightweight was realized while ensuring the recognition of effects. The results showed that the average accuracy of the model proposed for standing, walking, eating, drinking, and sitting was 84.62%, 92.58%, 87.54%, 98.13% and 87.18%, respectively. And the overall average accuracy was 90.01%. Compared with Faster R-CNN, YOLO v5s, YOLO v7, and YOLO v8s model, the average accuracy was 12.03 percentage points, 3.95 percentage points, 1.46 percentage points, and 2.19 percentage points higher, respectively. The results can provide technical support for sheep health management and disease warning.

Abstract:In response to the need for smart agriculture to accurately discriminate the degree of crop water demand, taking growing peppers as the experimental samples, different degrees of water stress treatments such as water immersion and drought to the leaves of peppers were applied to analyze the hyperspectral response characteristics of pepper leaves under different degrees of water stress. The samples were divided into four water stress groups, including severe drought, mild drought, mild water-soaked, and severe water-soaked, and one experimental control group, with a total of five data groups of 20 chili peppers in each group, and the chlorophyll fluorescence parameters and hyperspectral data of chili peppers’ leaves in each group were collected separately when the appearance of leaves in each group appeared to be obviously different. The effects of three different preprocessing methods, namely, multiplicative scatter correction (MSC), SG convolutional smoothing filter and standard normal variate transform (SNV), on the elimination of background information interference were compared. The SPA algorithm and CARS algorithm were used to extract the characteristic wavelengths sensitive to water stress. Support vector machine (SVM), BP neural network, radial basis function (RBF) and random forest (RF) modeling were established for predicting different levels of water stress. The results illustrated that SG-SPA-RFB was the optimal combination for predicting the degree of water stress with 99.02% accuracy in the training set and 94.00% accuracy in the test set. The research result can provide a convenient and reliable non-destructive method for determining the water stress status of pepper plants.

Abstract:Virtual water flow is an important manifestation of spatial redistribution of water resources. As a carrier of virtual water, grain trade and transportation involve virtual water flow. China’s population and resources show obvious spatial dislocation with food production, and the spatial matching degree of virtual water flow pattern, distribution of land and water resources, and economic and social development is poor. Based on exploring the resource and environment effect of virtual water flow in the current stage, adding the impact effect analysis on economy and society can further optimize resource allocation and promote regional sustainable development. The spatiotemporal evolution pattern of virtual water flow in 31 provinces of China from 1997 to 2021 was analyzed. Spacetime clustering analysis and coupled analysis of virtual water-economic and social data clarified the existence of significant spatial clustering of grain virtual water transportation. The amount of virtual water transportation in the main input and output provinces of food virtual water showed a significant positive correlation with the economic and social development level. Based on this background, totally nine major economic and social impact factors were selected and their spatial differences in influencing food virtual water flow were explored. The types of impact effect on food virtual water flow in 31 provinces of China were classified into industrial correlation type, social correlation type, and resource correlation type. Relevant regulatory strategies were proposed according to the economic development level, natural resource endowment, and industrial structure development of each province to weaken or even avoid the negative impact of virtual water flow on the regional economic and social development and natural environment. The results showed that the overall virtual water flow presented a trend of flowing from “deficient” northern regions to “rich” southern regions, and from economic backward areas to economic developed areas. The economic and social system to some extent influenced the virtual water flow. According to the impact effect of economic and social factors on food virtual water flow in different provinces, the industrial, social, and resource correlation types of food virtual water flow in each province were classified into industrial correlation type, social correlation type, and resource correlation type. In summary, promoting regional coordinated development and optimizing industrial structure would be an important solution to address the negative feedback effects of food virtual water flow in China caused by environmental and economic factors.

Abstract:Soil salinization and arable land degradation have seriously restricted the sustainable development of oasis irrigation agriculture in Xinjiang. Exploring the distribution characteristics of soil physicochemical properties and salt ions is a prerequisite and foundation for saline-alkaline land improvement and comprehensive utilization as well as high-quality development of oasis agriculture. The distribution characteristics of nutrient contents, and salt content and its ions in the 0~500cm soil profile were quantitatively analyzed in the Manas River Irrigation Area in northern Xinjiang, and in the Aksu River Irrigation Area and Kashgar River Irrigation Area (Aksu-Kashgar River Irrigation Area) in southern Xinjiang. The results showed that the contents of soil organic matter, alkali-hydrolysable nitrogen, available phosphorus, available potassium, and total nitrogen in the study areas all showed a gradually decreasing trend with the increase of soil depth. In the Manas River Irrigation Area, the average values of soil organic matter, alkaline dissolved nitrogen, available phosphorus, total nitrogen content and pH value were higher in the 0~30cm and 0~100cm soil layers, but the average values of organic matter, alkaline dissolved nitrogen, available potassium and total nitrogen content in the 100~500cm soil layer were lower in the Manas River Irrigation Area. The average values of total salinity and electrical conductivity in the 0~30cm soil layer were 21.14% and 8.53% higher in the Aksu-Kashgar River Irrigation Area, but the average values of total salinity and electrical conductivity in the 60~100cm soil layer were 17.55% and 16.50% lower in the Aksu-Kashgar River Irrigation Area, respectively. In the study areas, the dominant cations were Na+ and Ca2+, while the dominant anions were SO2-4 and Cl-. In the Manas River Irrigation Area, Na+ was the highest salt ions, while SO2-4 was the highest in the Aksu-Kashgar River Irrigation Areas. In the Manas River Irrigation Area, 53.85% of the 0~30cm soil layer was classified as saline soil, 50.00% of the 30~60cm soil layer was classified as moderately and severely saline soils, and 25.00% of the 60~100cm soil layer was classified as severely saline soil. The dominant saline soil was sulfate, followed by chloride-sulfate. In the Aksu-Kashgar River Irrigation Area, 78.26% of the 0~30cm soil layer was classified as saline soil, with the highest proportion being severely saline soil. Additionally, 60.86% of the 30~60cm layer was classified as mildly or severely saline soils, and 39.13% of the 60~100cm soil layer was classified as non-saline soil. The dominant saline soils were chloride-sulfate, followed by sulfate. The results can provide a scientific foundation for the comprehensive utilization of saline-alkaline land and precise fertilization in the oasis irrigation area of Xinjiang.

Abstract:The Gansu section of the Yellow River Basin, a crucial ecological security barrier in China, plays a vital role in water conservation and northern desertification control. Understanding the vegetation changes and driving forces is crucial for the ecological protection and high-quality development of the basin. Based on the MODIS NDVI dataset from 2001 to 2022, the vegetation change trend and driving force in the Gansu section of the Yellow River Basin over the past 20 years were analyzed by using pixel dichotomy, coefficient of variation, trend analysis, Hurst index and geographic detector. The results showed that the FVC in Gansu section of the Yellow River Basin showed an increasing trend during 2001—2022, indicating a positive vegetation growth. In the past 20 years, nearly 1/2 of the Gansu section of the Yellow River Basin had relatively high-high volatility, which was mainly distributed in the northern Loess Plateau region, while the medium volatility and relatively low-low volatility were mainly distributed in the Gannan Plateau region and the Longnan Mountain region. The vegetation in most areas of the Gansu section of the Yellow River Basin showed an improvement trend, while a few areas showed degradation, indicating a potential anti-sustainable trend in future vegetation changes. The analysis of driving forces indicated that precipitation, vegetation type and soil type were the predominant factors influencing vegetation cover in Gansu section of the Yellow River Basin.

Abstract:The planting experiment was conducted from 2020 to 2021 in the lysimeters at Jiangning District of Nanjing City to investigate the effects of farmland water level and nitrogen fertilization regulations on winter wheat growth, yield, grain quality, water and nitrogen utilization, nitrogen and phosphorus load after the winter wheat suffered from waterlogging during the jointing and booting stage. Winter wheat variety “Yangmai 25” was chosen as the experimental material, after the winter wheat suffered from waterlogging during the jointing and booting stage, three high, middle and low farmland water level (the farmland water level down to -40cm, -60cm, -80cm in 3 days after the 1 day waterlogging) and three low, middle and high nitrogen application rates (160kg/hm2, 190kg/hm2, 220kg/hm2), as well as a control group with non-waterlogging and the nitrogen of 190kg/hm2 were set in the experiment. The results indicated that the winter wheat plant height, aboveground dry matter, yield, water use efficiency,grain crude protein content and grain lysine content were gradually increased with the decrease of farmland water level and nitrogen application rate. Partial factor productivity of nitrogen was increased gradually with the decrease of farmland water level, while decreased with the increase of nitrogen application rate. The total nitrogen, total phosphorus, and other pollutant loads were increased gradually with the decrease of farmland water level. Compared with the control treatment, the increase of nitrogen fertilizer (220kg/hm2) could alleviate the effect of waterlogging stress, and promote the increase of winter wheat aboveground dry matter and yield by 4.76%~23.81% and 2.75%~9.19%, respectively. The reduction of nitrogen fertilizer (160kg/hm2) made the winter wheat yield at the medium and high farmland water level decreased by 2.20 % and 14.00%, respectively, and the corresponding water use efficiency was decreased by 4.55% and 9.74%, respectively. The low farmland water level could decrease the yield reduction effect due to the nitrogen reduction, which increased the yield and nitrogen partial factor productivity of winter wheat by 3.98% and 23.49%, respectively.The higher the farmland water level was, the greater the comprehensive waterlogging degree during the water control period was, and the lower the yield was.In addition, short-term waterlogging had a positive effect on increasing the crude protein content of grains, as the crude protein content of each treatment was increased by 11.50%~20.21% compared with that of the control treatment. Aiming at high yield, high efficiency, pollution reduction and quality improvement,it was recommended that the farmland water level should be lowered to -80cm in 3 days after the winter wheat suffered from waterlogging with 5cm water layer for 1day at winter wheat jointing and booting stage, and the nitrogen application rate should be 220kg/hm2. The research results can provide a theoretical basis for the winter wheat planting and waterlogging disasters restoration in southern China and similar waterloggingprone agricultural areas.

Abstract:To elucidate the effects of water-saving irrigation and nitrogen reduction combined with biochar application on the photosynthetic characteristics and water and nitrogen utilization of rice, a combination of field trials test and microzone test was used. 15N tracing technology was applied, with B0N1 (no biochar application+conventional nitrogen application level) as the control. B1N2 (10% nitrogen reduction+12.5t/hm2 biochar), B2N2 (10% nitrogen reduction+25t/hm2 biochar), B1N3 (20% nitrogen reduction+12.5t/hm2 biochar), B2N3 (20% nitrogen reduction+25t/hm2 biochar), B1N4 (30% nitrogen reduction+12.5t/hm2 biochar) and B2N4 (30% nitrogen reduction+25t/hm2 biochar) were set up, the photosynthetic characteristic parameters of rice plant leaves, as well as dry matter accumulation and water consumption processes, were observed, and the relationship between photosynthetic characteristic parameters and WUE, NUE, amount of dry matter, and yield was established. The results showed that under water-saving irrigation, appropriate nitrogen reduction combined with biochar application can increase leaf area index (LAI), chlorophyll content (SPAD), net photosynthetic rate (Pn), stomatal conductance (Gs), and transpiration rate (Tr), while reducing stomatal limitation values (Ls). Excessive reduction of nitrogen fertilizer or application of biochar would increase Ls, decrease LAI, SPAD, Pn, Gs, and Tr Compared with B0N1 treatment, B1N2 treatment increased the total accumulation of dry matter in plants by 14.79%, while B2N4 treatment decreased that by 16.02%. The NUE, yield, and WUE of rice treated with B1N2 were significantly higher than those treated with B0N1（P<0.05), with increases of 12.92%, 9.95%, and 1258%, respectively. The NUE, yield, and WUE of rice treated with B2N4 was significantly lower than that of B0N1（P<0.05), with decreases of 22.87%, 18.20%, and 5.66%, respectively;WUE and photosynthetic characteristic parameters (except LAI-tillering stage, SPAD, Tr-grouting stage) were significantly or extremely significantly positively correlated, and were significantly negatively correlated with Ls （P<0.01). NUE, amount of dry matter, yield, and photosynthetic characteristic parameters (except LAI-tillering stage, SPAD-tillering stage, Tr-grouting stage) were significantly or extremely significantly positively correlated, and significantly or extremely significantly negatively correlated with Ls. Overall, B1N2 treatment was the most optimal, which meant reducing the application of 10% nitrogen fertilizer and apply 12.5t/hm2 of biochar under water-saving irrigation was beneficial for improving water and nitrogen utilization efficiency and yield. The research results can provide theoretical basis and technical support for the application of water-saving irrigation with nitrogen reduction and biochar application in cold and black soil paddy fields.

Abstract:The temperature fluctuation during the low-temperature storage process of citrus is a key factor that triggers quality and safety risks for the fruit and increases refrigeration energy consumption. Simultaneously, quality and energy consumption are crucial evaluation indicators for assessing the efficiency of citrus cold storage. Achieving dynamic predictions for both aspects can provide reliable support for scientifically anticipating and precisely optimizing citrus cold storage efficiency. In light of this, a citrus cold storage efficiency time-series prediction model was proposed based on PatchTST. Firstly, a basic PatchTST model was constructed based on the self-attention mechanism and the channel independent (CI) prediction method. Secondly, by integrating the basic PatchTST model with the covariate feature extraction module from the TiDE model, feature extraction for all sequences in the multivariate time series dataset was achieved, effectively improving the model’s prediction accuracy. Finally, quantitative analysis of the correlation between cold storage refrigeration parameters, energy consumption, and citrus temperature was conducted by using the Pearson correlation analysis method. This analysis helped determine the input parameters for the TiDE-PatchTST model. The model was then trained and tested with 5000 sets of experimental data, and its accuracy and superiority were compared and validated against other models like basic PatchTST and Informer. The results showed that the predicted cold storage energy consumption values of the TiDE-PatchTST model had average absolute errors (MAE) and root mean square errors (RMSE) of 3.645W·h and 10.421W·h, respectively. The MAE and RMSE for citrus temperature predictions were 0.034℃ and 0.042℃, respectively. Compared with Transformer model, the MAE and RMSE in energy consumption predictions were decreased by up to 41.43% and 39.27%, and in citrus temperature predictions, they were decreased by up to 46.03% and 28.81%. The research result can provide strong support for the dynamic perception and optimization control of temperature fluctuations and energy consumption during the citrus cold storage process.

Abstract:Research on pitaya detection methods is the basis and prerequisite for realizing intelligent picking. Existing pitaya detection methods only target a single performance indicator, which is difficult to meet the needs of real agricultural scenarios. Therefore, an accurate and efficient dual-index detection method for pitaya quality and maturity was proposed. Firstly, the adaptive discriminator enhanced style generation adversarial network algorithm was used to expand the pitaya image and establish a pitaya dataset. The image was enhanced by gamma transform to highlight the characteristics of pitaya and reduce the impact of lighting environment. Secondly, the YOLO v7-RA model was proposed, by designing ELAN_R3 to replace the efficient layer aggregation network (ELAN) module to reduce the extraction of repetitive features by the backbone network. This enhanced the model’s attention to fine-grained features and improved the accuracy of dual-index detection. The mixture of selfattention and convolution (ACmix)was applied to enhance the model’s ability to extract and integrate feature information, and reduce the interference of cluttered background information. Finally, the detection level of the YOLO v7-RA model was verified through experiments. Experimental results showed that the precision rate of the method was 97.4%, the recall rate was 97.7%, the mAP0.5 was 96.2%, and FSP was 74f/s. A balance between detection accuracy and detection speed was achieved. Even under occlusion, the YOLO v7-RA model detection accuracy still reached 91.4%. The model had good generalization ability to provide strong technical support for the development of intelligent pitaya picking.

Abstract:To address the complicated installation and maintenance of the steering angle sensor and inaccurate angle estimation in traditional agricultural machinery navigation systems, the ARMAX-KF method and vehicle speed compensation were proposed to estimate the steering angle of tractors without steering angle sensors. Initially, the Hammerstein nonlinear system was used to model the tractor’s steering system, followed by identification using the RLS method as an ARMAX model. Subsequently, a method was proposed to obtain the velocity of the rear axle center point through speed lever arm compensation. Finally, ARMAX-KF was designed to estimate the steering angle, utilizing the correcting characteristics of the Kalman filter, using the tractor’s kinematic steering angle as the observation value to correct the integrated angle velocity predicted by the ARMAX model, thus estimating the steering angle of the tractor. The method of measuring speed for speed lever arm compensation achieved the average absolute error of the compensated kinematics steering angle of 1.110°, with a standard deviation of 1.727°, reducing the error by 61.13% and 31.55% compared with the values obtained before compensation. In the dynamic angle test, the standard deviation of the angle velocity predicted by the ARMAX model was 2.439(°)/s, reducing the error by 56.58% compared with the method using a fixed transmission ratio. The absolute average error of the steering angle estimation based on ARMAX-KF was 0.649°, with a standard deviation of 0.371°, reducing the error by 56.9% and 78.82%, respectively, compared with the methods using a fixed transmission ratio and the Kalman filter. In the straight-line navigation tracking test, the steering angle estimation standard deviation based on ARMAX-KF was 0.649°. The proposed method improved the accuracy of angle estimation and enhanced the quality of agricultural machinery navigation.

Abstract:The direction vector of the first rotating axis near the fixed platform remains constant in most existing 2R parallel mechanisms, while the direction vector of the second rotating axis near the moving platform only changes around the first rotating axis, and it remains constant relative to the moving platform. Based on spherical 4R mechanism, a class of 1Rv parallel mechanism with variable/invariable rotation axis was synthesized, and the change of the rotation axis of 1Rv mechanism was derived according to the finite screw theory. Based on bifurcation 1Rv parallel mechanisms with variable/invariable rotation axis, a 2Rv parallel mechanisms with two variable/invariable rotational axes was proposed by using the finite screw theory. The assembly condition and drive configuration of this kind of 2Rv parallel mechanism were analyzed. The novel bifurcated 2Rv parallel mechanism consists of four motion modes: invariable-invariable axes motion mode (in this mode, the first axis of rotation vector was invariable, and the second axis of rotation was also invariable relative to the moving platform), invariable-variable axes motion mode (in this mode, the first axis of rotation vector was invariable, and the second axis of rotation was variable relative to the moving platform), variable-invariable axes motion mode (in this mode, the first axis of rotation vector was variable, and the second axis of rotation was invariable relative to the moving platform), and variable-variable axes motion mode (in this mode, the first axis of rotation vector was variable, and the second axis of rotation was also variable relative to the moving platform). The existing 2R parallel mechanism with two invariable rotation axes was extended to the bifurcation 2Rv generalized parallel mechanism with two variable/invariable rotation axes (variable and invariable rotation axes).

Abstract:Error modeling and analysis are important prerequisites for ensuring the operational accuracy of robots, and many modeling methods were proposed by scholars in the past. However, few literatures directly verified the correctness of the established error model through either theoretical derivation or experimental means. To this end, an error model verification method of parallel robots based on parameter identification was proposed, which aimed to directly verify the rationality of the established error model through experiments. Firstly, the parameter identification model was established to acquire the actual structural parameters of the parallel robot, to establish the actual kinematics model. On this basis, the error model of the actual parallel robot was established by using the partial differential theory, the error parameters in the actual error model were quantitatively analyzed, and the influence of each error parameter on the pose error of the end-effector was obtained. Then, the influence of each error parameter on the pose of the end-effector was mapped to the joint input to drive the parallel robot to execute the error model verification experiment. Finally, the 3-PUU parallel robot was taken as the object for error analysis and experimental verification. The position data collected by the laser tracker were compared with the results of error model analysis. The maximum deviation between the two was 0.50mm, with the average deviation maintained within 0.31mm, which intuitively indicated the rationality and correctness of the established error model.

Abstract:With the development of automation technology, industrial robots are widely used in various fields, in order to improve the overall performance of industrial robots and reduce their static and dynamic performance errors, a topology optimization method that integrally took into account the flexibilities of connecting rods and joints of industrial robots was proposed. Combining robot dynamics and topology optimization, the multi-objective topology optimization function model of industrial robot forearm was established by linear weighted sum method based on the solid isotropic material with penalization (SIMP), based on the theory of flexible multi-body dynamics, the simulation model of robot rigid-flexible coupling dynamics with joints flexibilities and connecting rods flexibilities was established by using the finite element software and multi-body dynamics software, and the load spectra of the robot forearm was obtained in the extreme working conditions. Finally, the weight coefficients of each sub-objective in the optimization objective function were determined by using hierarchical analysis and the function was solved. The optimization result showed that the stiffnesses and natural frequencies of the optimized robot forearm were improved, the robot forearm was lightened by 18.71% from 20.233kg to 16.477kg. The robot model was reconstructed by virtual prototype technology and its whole was analyzed, and the result showed that the maximum deformation displacement of the robot was decreased from 0.208mm to 0.188mm under the maximum load, and the static deformation error was reduced by 9.62%, the dynamic localization error was decreased from 0.777mm to 0.687mm, and the localization accuracy was improved by 11.58%. The above topology optimization method provided an effective theoretical reference for improving the overall static-dynamic performance of industrial robots.