Abstract:With the development of power electronics and energy storage technology, electrification of power equipment has become an important direction of global vehicle development. It has been successfully applied in the field of new energy vehicles, and has taken the lead in establishing a complete industrial foundation in China. At present, the global electric agricultural equipment is in its infancy, mostly based on theoretical research, and there is no mass-produced electric agricultural equipment product. The development of electric agricultural equipment has industrial advantages. The key components and software platforms of electric agricultural equipment were briefly analyzed, and a comprehensive review of the research status of electric tractors, electric micro-tillers, electric transplanters, electric orchard machines, and electric seeders at home and abroad were focused on. The performance comparison of electric agricultural equipment and traditional agricultural equipment was carried out, and the advantages and disadvantages of different agricultural equipment were obtained. It can provide support for the application scenario analysis of agricultural equipment. According to the agronomic characteristics of different agricultural equipment and the characteristics of key components of electrification, the application scenarios of different forms of agricultural equipment were expounded. Combined with the current development status of electric vehicles and the operation characteristics of agricultural equipment, the bottlenecks of the development of different electric agricultural equipment were analyzed, and the direction for the development of electric agricultural equipment was pointed out. It was hoped that the research result can serve as a valuable reference for the development of electric agricultural machinery in China.

Abstract:Agricultural robot plays an important role in accelerating transformation of agricultural modernization and achieving intelligent agriculture. Field positioning and navigation technology is the foundation for ensuring the safe and efficient completion of various agricultural tasks by robots, and combining high-precision positioning information of robots to achieve efficient planning of work paths is the technical core of field positioning and navigation. A path planning algorithm based on macro-micro integration was proposed. Firstly, the algorithm generated a global static work path based on the macro mapping information of the operation area. While implementing robot operations, various radar sensors were used to dynamically monitor the micro work environment and path information of the robot in real time. Finally, path tracking algorithms such as MPC were applied to real-time process local and global work environment information to achieve real-time job path optimization and correction to ensure the smooth progress of field work. Experimental verification showed that when the adjustable distances on both sides of the robot during field operations were 0.2m, 0.1m respectively, the algorithm can reduce the average crop compaction rate during the operation process from 3.4058% and 1.2763% to 0.6772% and 0.1889%. Meanwhile, the algorithm improved the precision of field operations under the requirements of precision agriculture and had important significance for achieving the goal of high yield, efficiency, and quality in agriculture.

Abstract:Identifying the trajectories of wheat harvester in the field is an important means to analyze the activities of agricultural machinery and improve the working efficiency. A machine learning based algorithm for recognizing the turning trajectories of wheat harvester was proposed. Identifing X-turn, abnormal working, and working trajectory through two-step K-means iterative clustering and three-step correction method: the first step (M1) was performed based on the three distance features between the trajectory segments and the cluster center of the trajectory segments. The second step (M2) was based on the direction change of the “turning” and “abnormal working” trajectories. The third correction step (M3) was based on the operating characteristics to specify the start and stop positions of the turning. In order to further classify U-turn trajectories from working trajectories, identifying X-turn, abnormal working, U-turn and working trajectories through SVM model and three-step correction method, firstly, the correction of U-turn boundary based on trajectory curvature (S1) was carried out. Secondly, based on the time difference between X-turn and U-turn, the misidentification as a U-turn was corrected (S2). Thirdly, the correction was based on the change of the angle before and after the U-turn (S3). The F1score of the four trajectories recognition results was 94%. The accuracy, recall, and F1 scores of data recognition results at different time intervals of 1~5s were all above 85%, indicating that the algorithm performed well on trajectory data at 1~5s intervals. When the time interval was extended to 10s and 15s,the U-turn trajectory would not be recognized, indicating that the algorithm cannot be applied to overly sparse trajectory data. The effective working trajectories were obtained after removing the X-turn trajectories, U-turn trajectories and abnormal working trajectories of the positioning track data in a field. The error of calculating the farmland area by the distance algorithm can be reduced by 12.76% compared with the calculation error of using the original data. The research result can provide a reference for fine management of farmland operations.

Abstract:Visual navigation has the advantages of low cost, wide applicability and high degree of intelligence, so it is widely used in orchard navigation tasks. Therefore, how to quickly and accurately identify the navigation path is a key step to achieve visual navigation. Aiming at the problems of multiple interference factors and complex image background in the application of visual navigation system in dragon orchard environment, a visual navigation path recognition method was proposed for dragon orchard based on improved DeepLabv3+ network. Firstly, the traditional DeepLabv3+ backbone feature extraction network was replaced by MobileNetV2 from Xception, and the atrous convolution in atrous spatial pyramid pooling (ASPP) was replaced with depthwise separable convolution（DSC）. While improving the model detection rate, the number and memory footprint of model parameters were greatly reduced. Secondly, coordinate attention (CA) was introduced at the feature extraction module, which was helpful for the model to locate and identify road areas. Then, experiments were conducted on a self-built dragon orchard road dataset containing three different road conditions. The results showed that compared with the traditional DeepLabv3+, the MIoU and MPA of the improved DeepLabv3+ were increased by 0.79 percentage points and 0.41 percentage points, respectively, reaching 95.80% and 97.86%. Frames per second (FPS) was increased to 57.89f/s, and the number of parameters and memory footprint were reduced by 92.92% and 97.00%, respectively, to 3.87×106 and 15.0MB. The recognition results of the improved model on the orchard road were verified on the test set, indicating that the model had good robustness and anti-interference. In addition, comparing the proposed model with Pspnet and U-net networks, the results showed that the improved models offered significant advantages in detection rate, amount of parameters, and model size, making them more suitable for deployment to embedded devices. According to the segmentation results of the model, the edge information on both sides of the road was extracted, the road boundary line was fitted by the least squares method, and finally the navigation path was extracted by the angle bisector line fitting algorithm. The navigation path recognition accuracy was tested in three different road environments, and the test results showed that the average pixel error was 22 pixels and the average distance error was 7.58cm. The road width of the orchard in this test was 3m, and the average distance error accounted for only 2.53%. Therefore, the research result can provide an effective reference for the visual navigation task of dragon orchard.

Abstract:A mobile inspection platform based on multi-dimensional perception was developed to enable intelligent inspection and monitoring of maize growth dynamics, drought stress and diseases in wide fields. Firstly, the chassis assembly’s steering system, drive system, and control system were developed, and the steering and driving functions of the inspection platform were implemented, using the Arduino UNO controller. Secondly, a multi-dimensional sensing system that consisted of a global navigation satellite system/inertial navigation system (GNSS/INS) integrated navigation system, light detection and ranging (LiDAR) and camera was constructed. The time synchronization scheme, data communication structure and information acquisition software of the sensing system were then designed to enable the patrol platform to perceive its environment and visualize. Finally, the chassis driving performance test and the perception system environment perception test were performed on the inspection platform in the maize field. According to the test results, the inspection platform’s average minimum turning radius for left turns was 2922mm, its average minimum turning radius for right turns was 2736mm, and its maximum climbing gradient was greater than 26.7%, the average straight-line speed under position PID control was 0.523m/s, with an error of 4.6% compared with the expected speed of 0.5m/s, the average deviation for 15m driving was 0.636m, and the average deviation rate was 4.24cm/m, all of which met the field driving performance requirements. Under the ROS system, the sensing system was capable of reliably gathering platform position information, high-precision 3D point cloud information, and color 2D image information, enabling the inspection platform to perceive the surroundings in many dimensions. The research result can be used to guide the intelligent creation of a maize field inspection platform.

Abstract:The lack of spatial positional information of picking targets and low target localization accuracy are one of the key problems that limit the picking effect of strawberry picking robots. To address these problems, a target localization and segmentation model was firstly designed based on color information and convolutional neural network for strawberry image and target point cloud segmentation;secondly, an image-based strawberry pickability and obscuration recognition model was implemented;finally, a strawberry spatial localization and pose estimation model was designed and a strawberry picking point localization method was implemented. Based on this method, the estimation error of intact strawberry position was 4.03%, the estimation error of obscured strawberry position was 9.06%, and the comprehensive error of picking position was 2.3mm. In the actual picking experiment, the picking success rate was 92.6%, the average calculation time of each strawberry was about 92ms, and the average execution time of single strawberry picking action was about 5.7s. The experimental results can provide effective target localization information for strawberry picking robots, which can effectively meet the needs of actual picking scenarios.

Abstract:In view of the problems in the current manufacturing methods of soft claws (such as soft offset printing, lost wax casting, etc.), such as complex forming process, unstable adhesion, easy tearing and so on. Stereo lithography apparatus (SLA) soft picking claw integrated structure was designed, which could realize the adaptive grasping of fruits and vegetables by positive and negative pressure drive. Firstly, based on the Yeoh model, the nonlinear mechanical properties of the bending deformation motion of the soft finger were studied, and the nonlinear relationship model between the internal pressure of the cavity and the bending angle of the finger was obtained. Then, the bending characteristics of the soft finger were analyzed by using Abaqus finite element software, and the influence law of the main structural parameters on the bending angle was obtained. And combined with the orthogonal experimental method, the optimum combination of the structural parameters was obtained as follows: the cavity height of the soft drive was 11mm, the channel height was 4mm, the cavity wall thickness was 1.6mm, the number of cavities was 7, the cavity spacing was 3mm, and the thickness of the bottom layer was 3mm. Finally, based on the optimal combination of structural parameters, a prototype of the soft picking claw was fabricated and installed on a test platform for fruit and vegetable grasping experiments, which verified the practicality of the SLA soft picking claw.

Abstract:Inspired by the tactile sensation of human hand nerves, an untethered intelligent soft gripper was proposed to adaptively and effectively pick fruits. This soft gripper applied a self-circulation air pump to supply air pressure and integrated a sensor system in which flexible membrane tactile sensors were embedded in the soft gripper fingers, resulting in adaptive grasping of multi-size and multi-type globose fruits. The working principle of the self-circulation air pump was investigated and the structure of the air pump was optimized. In addition, the model was built to predict the output air pressure of the self-circulation air pump and then conduct experiments to demonstrate the performance of the air pump. The adaptive soft gripper was then prototyped and a model was built to analyze the soft gripper grasping force. Static experiments were performed to obtain the flexural deformations and mechanical properties of the soft adaptive gripper under actuation. A control system and grasping rules were developed to perform adaptive grasping, an experimental platform in the laboratory was built to simulate conditions in an orchard, and a series of experiments were conducted to demonstrate the efficacy of adaptive grasping and the capability of picking and sorting globose fruits. The results showed that the soft intelligent gripper was capable of picking globose fruits efficiently with the help of tactile feedback and control systems, and the size range of fruit for picking was 48.5~97mm, the maximum mass can reach 350g, the average time for picking was 15s and the success rate of picking was 97.46%.

Abstract:In view of the problems existing in the actual operation process of the film hole forming device for rapeseed seeding, such as the excessive size of the hole, the irregular shape of the hole and the adhesion of the hole, etc., based on the principle of sliding cutting, a rolling film cutting and hole punching device was designed, the structural parameters of the perforating device and profiling mechanism were determined, and the kinematics model of the perforating device was established, the main factors affecting the film hole length and the range of its value were determined. Using DEM-MFBD coupling simulation, three factors and three levels regression orthogonal test, the simulation test was carried out with the forward speed of the whole machine, the length of the longitudinal blade and the height of the longitudinal blade as the test factors, the film hole length and the hole distance difference as the evaluation index. The results showed that the primary and secondary order of the influence of each factor on the length of the film hole was the length of the longitudinal blade, the forward speed of the whole machine and the height of the longitudinal blade. The primary and secondary order of the influence of each factor on the hole spacing difference was the height of the longitudinal blade, the forward speed of the whole machine and the length of the longitudinal blade. When the forward speed of the whole machine was 3.3km/h, the length of the longitudinal blade was 34mm, and the height of the longitudinal blade was 31mm, the film hole length was 44.78mm, and the hole distance difference was 0.64mm, the punching performance was better. The field experiment of rolling film cutting and punching device was carried out with the optimal parameter combination. The results showed that the average length of the film hole was 43.15mm, the coefficient of variation of the stability of the film hole length was 3.86%, the average hole distance difference was -1.32mm, and the film hole distance error was 4.22%, which met the requirements of film cutting and drilling for rapeseed sowing. The research result can provide reference for film cutting and punching device for rapeseed sowing.

Abstract:Variable rate fertilization technology is an important means of scientific fertilization, which can make fertilization more accurate and targeted, and effectively reduce farmland pollution. During high-speed rice transplanting and synchronous fertilization operations, the adjustment of fertilization amount is mainly controlled by pre calibration, which is time-consuming and unstable in accuracy. To quickly and accurately adjust the fertilization amount and achieve variable rate fertilization operations, an automatic control variable rate fertilizer application device was studied and designed for solid particle fertilizer. The overall structure and working principle of the variable rate fertilizer application device was expounded, and the key components were designed and tested. The on-line monitoring and intelligent control system of fertilizer amount were constructed with the single-chip microcomputer STM32 as the control core. By using the optimization method of experimental design, the performance and main influencing factors of the fertilizer flow online detection system were studied, and the best combination of factors was determined. Through the experiment, the change rule models between the detected flow of three kinds of main solid particle fertilizers and the voltage value of piezoelectric sheet, the change rule models between the actual flow of three kinds of main solid particle fertilizers and the rotation speed of the fertilizer discharge shaft, the change rule models between the rotation speed of the fertilizer discharge shaft and the working length of the electric push rod and the forward speed of the rice transplanter were respectively constructed, and the models were tested and analyzed. The results showed that when the forward speed of the rice transplanter was 1m/s and the rotation speeds of the fertilizer discharge shaft were 20r/min, 25r/min and 30r/min, the average detection accuracy of the fertilizer discharge of the three solid granular fertilizers were 94.45%, 93.85% and 93.15%, respectively. When the forward speed of the transplanting machine was 0.6~1.4m/s, the compound fertilizer application rates were 200kg/hm2, 250kg/hm2 and 300kg/hm2 and the urea application rates were 165kg/hm2,195kg/hm2 and 225kg/hm2,the average coefficient of variation of the fertilization rates of the three fertilizers were 3.02%, 3.15% and 2.82%, respectively. Field experiment showed that when the transplanting machine advanced at speed of 1~1.4m/s and the fertilization rates were 180kg/hm2 and 225kg/hm2, respectively, the average accuracy rates of compound fertilizer fertilization in Norway were 94.89% and 97.82%, respectively. Theoretical and experimental analysis showed that the variable rate fertilization device had stable regulation performance and can meet the operational requirements of rice side deep variable rate fertilization.

Abstract:A drone seeding device was designed to address the problem of scattered and disorderly seed falling caused by the widespread use of diffuse sowing in rice and rapeseed aerial seeding, which can simultaneously meet the agronomic requirements of rice and rapeseed aerial seeding. The design goal was to achieve dual use, lightweight, electric drive, and modularization of the strip aerial seeding system. The external groove wheel component with electric drive and adjustable working length was used as the seeding device, and the automatic folding seed guide tube driven by the linkage driven by the servo was used as the seeding component. Through bench tests, on-site mud box aerial seeding tests, and field tests, the structural and working parameters were determined and the operational effect was verified. The test showed that the designed seed metering system can meet the agronomic requirements of 6~7.5kg/hm2 oil rape, 15~45kg/hm2 hybrid rice and 60~105kg/hm2 conventional rice at maxium flight speed of 5m/s when in the range of rated speed and torque of seed metering motor, and the performance parameters such as the coefficient of variation of consistency of each row and the coefficient of variation of consistency of total discharge capacity were superior to the requirements of the industry standard. At height of 1m and speed of 4m/s, the average strip width of rapeseed and rice seeds in the mud box was 6.7cm and 3.8cm, respectively. The effect of seedling formation in the field was significant after 30d of sowing.

Abstract:Aiming at the problems of low operation efficiency and poor operation quality of semi-automatic transplanting machine, a stem-type automatic seedling picking device for vegetable transplanting robot was designed. The seedling pick-up device can realize efficient and high-quality automatic seedling pick-up and delivery operation through whole row of seedling pick-up, equidistant seedling separation and precise seedling delivery. The motion mechanics model of multi-stage scissors seedling mechanism and clamping seedling device was established. The model design and analysis calculation of the falling motion and pneumatic system of pot seedlings were carried out, and the seedling test device was built. Plug pepper seedlings were selected as the research object, and the seedling age, matrix moisture content and seedling frequency were taken as the experimental factors. A single factor experiment was designed to evaluate the success rate of seedling and matrix crushing rate. According to the test results, the Box-Behnken response surface analysis method was used to design the orthogonal experiment. The effects of the interaction between seedling age and substrate moisture content, seedling age and seedling picking frequency, substrate moisture content and seedling picking frequency on the seedling picking effect were explored, and the seedling picking parameters were optimized. The results showed that when the seedling age was 33d, the water content of pot seedling substrate was 46%, the frequency of seedling picking was 75 plants/min, the success rate of seedling picking was 97.36%, and the substrate crushing was 5.07%, which could meet the requirements of seedling picking and throwing of field automatic transplanting.

Abstract:The small space of kiwifruit and other pergola-structured orchards restricts the use of organic fertilizer applicators. The existing fertilization methods are labour-intensive and time-consuming, with low fertilizer efficiency. Therefore, a mechanized no-furrowing deep organic fertilizer application method was proposed, considering kiwifruit’s growth characteristics and shallow root distribution. The method consisted of three simultaneous and successive steps: fertilizer spreading, fertilizer-soil mixing and soil covering. No furrow was opened during fertilizer application, but the fertilizer was mixed into the soil and not exposed to the ground. Based on this method, a non-furrowing deep organic fertilizer applicator for kiwifruit orchards was developed with height of 1.5m, power configuration of 37.5kW, and fertilizer box volume of 1.2m3. The organic fertilizer applicator consisted of a low-side seat tractor, fertilizer trailer, fertilizer spreading mechanism, fertilizer mixing mechanism, anti-twisting mechanism, and reel mechanism. The fertilizer mixing mechanism rotated and cut the soil to realize the fertilizer-soil mixing and the soil covering of the mixed layer. The radius of the anti-twisting wheel was determined to balance the torque generated by the one-sided arrangement of the fertilizer mixing mechanism. The reel mechanism simultaneously retracted the fertilizer mixing mechanism and the anti-twisting mechanism, and the working angle of the fertilizer mixing mechanism can be adjusted according to the terrain. The optimal knife-to-machine speed ratio was determined as 32 based on the discrete element simulation analysis of the working process of fertilizer mixing mechanism. The results of field test showed that under the working condition of the maximum designed fertilizer applying depth of 150mm and the fertilization amount of 5.0kg/m, organic fertilizer was mixed into the soil, the relative error of the fertilizer applying depth was less than or equal to 7.73%, the fertilizer exposure rate was less than or equal to 5.56%, and the average power consumption of the fertilizer mixing mechanism was 4.7kW, which met the requirements of organic fertilizer application in kiwifruit orchards. The research result provided a novel method and efficient equipment for organic fertilizer application in kiwifruit and other pergola-structured orchards.

Abstract:Tea is a kind of global beverage, the mechanized tea refining operation is an important part of the entire tea mechanization process. Vibrating-sifting is an integral part of tea refining, which directly affects the purity of tea. In view of the unclear mechanism of vibrating-sifting in the process of tea vibrating-sifting, there are problems such as broken tea and hanging mesh, which affect the vibrating-sifting efficiency. The motion law of tea on the screen bed was researched, the dynamic model of the three-stage screening process of tea moving up and down along the screen bed was established, which was thrown up from the screen surface and colliding after falling to the screen hole, combined the motion law of tea on the screen bed, a simulation model of tea screen bed was established through EDEM, and the speed and force of tea on the screen bed were analyzed. The best parameters were crank radius, rotating speed of the crank and the inclined angle of the screen surface. Finally, through the orthogonal test of three factors and three levels, the software Design-Expert was used to analyze and optimize the experimental data and determine the optimal parameters. The significance order of each factor on the error screening rate and productivity was obtained. The order of significance for the error-sifting rate was the inclination angle of sifting bed, crank speed and crank radius, while the significant order of the productivity was the inclination angle of sifting bed, crank speed and crank radius. When the crank speed was 247.99r/min, the inclination angle of the sieve bed was 2.60°, the crank radius was 23.11mm, the error screening rate and productivity were 5.3% and 440kg/(m2·h), respectively, compared with the optimization result, the error was within the allowable range. The research result had a significant reference value for improving the screening efficiency of vibrating-shifting machine for tea.

Abstract:With the increase of mechanization of cotton planting and harvesting, it is particularly important to obtain accurate yield map and analyze field yield data, and it is an effective and feasible method to monitor the yield at the cotton conveying pipeline during the operation of cotton picker. The existing photoelectric beam cotton yield measurement sensor has problems such as mucus blocking detection channel and ambient light influence in operation. Facing the complex field working environment, linear or polynomial model is generally used for sensor calibration, and the accuracy and anti-interference performance are not ideal. In view of the above situation, the anti-interference in the structure and circuit design of the sensor was firstly improved. Then, in the process of sensor calibration, random forest regression (RFR) was used to train and test the experimental samples. After analyzing the performance of the model, a stochastic forest regression model based on sparrow search algorithm (SSA) was proposed. The mean square error was used as fitness value to optimize the model. After verification, the optimized model had better detection accuracy under the same verification set. The optimal detection model was obtained by optimizing the range of upper and lower bounds, balancing the running time and detection accuracy. The model performed well on the validation set with a coefficient of determination (R2) of 0.99 and a mean absolute percentage error (MAPE) of 6.34%. The bench test results showed that the maximum error was 9.21% and the average error was 8.33% at different wind speeds. The improved sensor and detection model had good performance and can accurately detect the cotton quality during the operation of the cotton picker.

Abstract:Aiming at the problems of insufficient safety and weak running stability of self-propelled Panax notoginseng combine harvester in field operation, the driving performance of the chassis under the condition of clay soil was studied. Firstly, theoretical analysis was carried out under the driving conditions of the harvester such as straight and steering, longitudinal and transverse climbing and crossing the obstacles, etc., and various factors influencing the driving performance were obtained. Secondly, multi-body dynamics simulation analysis was carried out by ADAMS ATV, and Euler angle and corresponding angular velocity curve were drawn under the driving conditions. A 1∶4 miniature platform model of the combine harvester chassis was designed by using the dimensional analysis method of similarity theory, and the driving performance of scaling model experiment was carried out under the three typical driving conditions of longitudinal climbing, crossing the ditch, and climbing over the ridge. The simulation results showed that the combine harvester chassis had smooth straight running and steering performance, which can pass through 30° longitudinal slope, 20° transverse slope, 600mm ditch and 300mm ridge smoothly. The model experiment results showed that the model can pass through 30° longitudinal slope, 150mm ditch and 75mm ridge smoothly. The trend of the pitch angle and corresponding angular velocity curve were consistent with the simulation under the above three conditions, and the amplitude change of the pitch angle curve obtained under the two conditions was equivalent. The experimentation error was mainly affected by the actual terrain flatness and soil evenness. It showed that the scaling model experiment can verify the correctness of the simulation results. Through the prediction of the model to the prototype and the simulation results, the design requirements of the trafficability for the harvester were satisfied. The research result can provide a theoretical foundation and reference for design of the root combine harvester chassis in the hilly and mountainous areas.

Abstract:King grass is a perennial hybrid pennisetum with cluster rooting and high stem. It has good feeding value, but it is difficult to harvest mechanization. Aiming at the problems of poor agronomic matching and insufficient supporting power in the operation of the existing king grass harvester, a self-propelled tracked chassis of the king grass harvester was designed. According to the work requirements of reducing stubble rolling and speed stable harvesting, the walking and stepless variable speed driving device were designed. The hydraulic power steering system and frame were designed and selected based on the terrain features of small plots and gentle slopes. The stability performance was analyzed, and the whole machine was tested. The results showed that the maximum driving speed of the chassis was 9.02km/h. Its minimum turning radius was 1349mm, and the maximum climb was 26°. The whole machine had no side slip and tipping phenomenon when driving on the contour line of the lateral slope of 15°~16°. The time of reliable parking along the uphill and downhill directions on the longitudinal slope of 10°~12° was more than 5min. When harvesting in the field with a slope of 8°~9°, the whole machine can achieve a stepless speed change of 0~4.19km/h, which had sufficient power, and ran smoothly. And the average stubble rolling rate of king grass was 7.43 %. The research result showed that the designed chassis can meet the mechanized harvesting requirements of king grass planting on small plots and gentle slopes.

Abstract:Conventional large-scale ground sprayers may hardly be deployed in hilly orchards due to the restriction of geo-characteristics, so that manual spray has to be a general approach for plant protection, which results in high labour intensity, low operational efficiency and severe chemical waste. A small self-propelled target-following sprayer for hilly orchards was proposed, which can work together with plant protection UAV. Two coupled nozzles were adopted for the target-following spray mechanism, and the ranges of their spray angle and height were calculated based on the developed physical motion model of droplets. Meanwhile, the method of target detection and tracking and the method of autonomous navigation were integrated into the proposed sprayer to realise autonomous operation. The verification experiment was performed in the apple orchard of Fuyu Forest Fruit Company in Hebei Province, and the results showed that the average coefficient of variation of droplet number deposited on the frontal surfaces of the leaves at different heights of canopies was 34.22%, while that at different sampling points at the same height was 34.56%. Compared with non-target spraying, the target-following spraying could effectively improve the uniformity of droplet distribution on the frontal surfaces of leaves both at different heights and at different internal and external positions of canopies. Besides, water consumption, ground loss and post-canopy drift loss could be reduced by 26.70%, 84.93% and 53.50%, respectively, which indicated that the proposed method can be a technical reference for the development and improvement of hilly orchard sprayers.

Abstract:In order to solve the problem of accurate measurement and control of the distance of the spray rod relative to the crop canopy when spraying solanaceous crops such as potatoes, a set of spray rod height control system of the potato sprayer was designed. The system adopted 2d laser radar scanning field potato plant canopy, according to the potato planting pattern of crop canopy crown unit segmentation, through the fusion of attitude sensor data on radar output data correction, and based on the median filtering algorithm, moving the least squares curve fitting method processing crown point cloud data, real-time calculate the vertical distance of crown relative information, and the fusion cylinder displacement sensor designed double threshold spray rod height control strategy, realize the precise control of spray rod relative potato crown height. The system was applied to the 3WP-1500 sprayer，and the system performance was tested by height detection precision test and height adjustment test. The test results showed the maximum relative error of the crop canopy height detected by LiDAR was 7.16%, and the average relative error was 3.95%. The height adjustment test showed that by determining the optimal adjustment threshold, the spray bar height adjustment error can be effectively reduced, and the system stability can be improved, and the standard deviation of the test height adjustment was 21.81mm, and the average relative error was 3.08%, and the system ran smoothly to meet the automatic control needs of the relative canopy distance of the spray bar.

Abstract:In an effort to investigate the influence of downwash airflow field on droplet motion characteristics of small unmanned aerial vehicles (UAVs) and enhance pesticide application efficiency, small UAVs were utilized as the basis and the Navier-Stokes (N-S) equations, realizable k-ε model, and Semi-Implicit Method for Pressure-Linked Equations algorithm were employed to detailed numerical simulations of the downwash airflow field and droplet dispersion motion during the spray process were carried out. The primary research focus encompassed the analysis of velocity characteristics of the UAV downwash airflow, droplet deposition characteristics, and the impact of operating height on pesticide deposition. By comparing simulated values with measured values, it was confirmed that the relative error between the simulation and the experiment remained within 20%, thereby verifying the feasibility of the downwash airflow field numerical model. Further simulation analysis results indicated that, under the influence of the UAV spray platform, the downwash airflow field attained its peak velocity at a distance of 1 m from the rotor. With the increase in operating height, the droplets gradually dispersed and diffused. The droplets were primarily distributed in two “airflow introduction zones” and two “airflow export zones”, which contributed to further optimization of spraying effects and enhanced efficiency of pesticide utilization. In light of the analysis results, it was substantiated that when the UAV flight operation attitude was kept flush with the ground and the operating height was adjusted to 0.8~1.0m, pesticide deposition could be significantly increased, ultimately improving the pesticide application efficacy of plant protection UAVs. The research result verified the feasibility of the numerical model of downwash airflow field, and provided a reference for the study of the drift and deposition of target fog droplets based on small plant protection UAVs.

Abstract:The spray process relies heavily on the droplet size distribution, which plays a crucial role in mass, momentum and energy transport. Currently, determining the droplet size distribution is a major scientific problem, which is represented by distribution functions classified into empirical and theoretical distribution methods. Empirical methods which derive droplet size distribution formulae from statistical analysis of experimental data lack practical physical significance and rely too heavily on empirical data. In contrast, theoretical approaches mainly use the maximum entropy approach, which originates from physical conservation laws but faces challenges in accurately predicting the droplet size distribution under complex conditions. To address these challenges, a maximum entropy model of droplet size distribution was proposed based on the maximum entropy principle, with an average diameter constraint condition used for constructing three and four-parameter maximum entropy models. The optimal model was selected based on the comparison of Akaike information criterion numbers, and the three-parameter maximum entropy model using the average diameter was found to be the best in predicting droplet number distribution. Air-blast nozzle atomization experimental data were used to optimize the proposed model, and the results showed that the correlation coefficient between predicted and experimental droplet number differential distribution values was above 0.96, with a mean square error lower than 0.135. Moreover, the three-parameter maximum entropy model accurately predicted the number and distribution of spray droplets. The proposed model was also tested against experimental data on atomized droplet size distribution from different nozzle types, yielding a good match with the experimental data. Finally, the selected model was applied to predict the particle size distribution of spray droplets from pressure nozzles manufactured by Pratt & Whitney Canada, demonstrating its accuracy in predicting the spray droplet size and quantity distribution despite the complexity of the working conditions. In conclusion, the research result can provide a significant contribution to accurately predicting droplet size distribution and quantity, and the proposed three-parameter maximum entropy model had great potential in improving spray droplet size and quantity distribution prediction accuracy.

Abstract:Due to the frequent changes of external operating conditions, the mixed-flow pump usually operates under non-design conditions. Therefore, it is of great significance to carry out multi-condition optimization research to improve the mixed-flow pump operating efficiency.On the basis of verifying the accuracy of numerical simulation by experiment, an optimization system consisted of the circulation method, experiment design, response surface model and the optimization algorithm was used to optimization design the mixed-flow pump with a specific speed of 511, and the influence of each design parameters on each optimization objective was analyzed. The blade loading, blade trailing edge lean angle and spanwise distribution of impeller exit circulation were selected as design parameters, the pump efficiency at 0.8 times and 1.2 times of design flow were selected as optimization objectives, the pump head at design flow were selected constraint. The results showed that the spanwise distribution of impeller exit circulation had a great influence on the performance of the mixed-flow pump and should be carefully considered in the optimization design. The blade loading at hub should be fore-loaded, the blade loading at shroud should be aft-loaded, and the spanwise distribution of impeller exit circulation should increase gradually from hub to shroud to further improve the optimized mixed-flow pump efficiency. The pump efficiency of the optimized mixed-flow pump at 0.8 times, 1.0 times and 1.2 times of design flow were 81.11%, 88.38% and 80.56%, respectively. The pump head of the optimized mixed-flow pump at design flow was 12.33m, compared with that of the original model, these efficiencies were improved by 0.63 percentage points, 3.18 percentage points and 6.72 percentage points, respectively, the pump head fluctuation was also less than 2% at the same time. Therefore, the optimization method based on full circulation control was effective and can provide a reference for the same type of rotating machinery.

Abstract:The shutdown transition process of the motor-pump device was studied by ANSYS Fluent software, which mainly explored the external characteristics and internal flow field in the shutdown process. It was found that the braking condition of the motor-pump accounted for the smallest ratio of the whole shutdown process, the runaway speed was about 84% of the design speed, and the runaway flow rate was 1.17 times of the design flow rate. The flow direction of gap backflow in the shutdown process was always from the impeller outlet to the impeller inlet, and its total flow rate showed a gradually decreasing trend. During the shutdown process, the axial force as a whole showed a decreasing trend;the radial force of the rotor as a whole showed a decreasing trend and then an increasing trend;the pressure pulsation at the impeller inlet and outlet was firstly decreased and then increased, and after reaching the maximum value in the braking condition, it was rapidly decreased in the turbine condition until it entered the runaway condition and stabilized. The impeller inlet pressure pulsation amplitude was the largest in the pumping unit, about twice that of the impeller outlet. Due to the influence of the gap backflow, there was a small vortex in the impeller inlet near the rim area, the range of which was firstly decreased in the pumping condition, then suddenly increased in the braking condition, and finally decreased again in the turbine and runaway conditions. The position of the impeller inlet vortex was gradually moved towards the impeller inlet during the shutdown process. The entropy production within the motor-pump unit was mainly concentrated in the downstream region, headed by the impeller. As the shutdown process progressed, the high entropy production area inside the pump unit was gradually moved towards the inlet guide vane, and the range of the high entropy production area was firstly decreased and then increased. The location and range of the high entropy production region in the pumping section of the motor-pump corresponded to the location and size of the vortex, which indicated that the vortex and the poor flow conditions such as deliquescence were the main reasons for the high entropy production within the pumping section of the motor-pump.

Abstract:As an important and unique ditch control project in the severe soil erosion area of Loess Plateau, check dams have the functions of flood detention, silt retention, and water storage. However, when using digital elevation model (DEM) for watershed water system analysis in this area, the extracted channel network will be disturbed by the check dam to have offset and dislocation problems. An automatic method for detecting the check dam and correcting the DEM was proposed to realize the effective extraction of the channel network. In this method, using the region growing and morphological algorithms to extract the center line of the channel to narrow the detection area of the check dam. Based on the improved line segment detector (LSD) algorithm and angle filtering, the candidate outline line of the check dam was determined, and then a cross model was constructed to locate the check dam. Finally, the grid elevation where the check dam was located was corrected to obtain a complete channel network. The F1 values of check dams detected by this method in two typical sample areas of the Loess Plateau were 86.67% and 86.95%, respectively. Compared with the two channel network extraction methods of filling and breaching, the method can effectively solve the problem of channel deviation. The experimental results showed that the results of the channel network extracted were more consistent with the real topography, which can provide useful supplements and support for the digital terrain analysis technology of the Loess Plateau check dam area.

Abstract:To identify the types of heavy metal stress on crops, corn leaves under heavy metal stress of copper (Cu) and plumbum (Pb) were selected as the research object. The hyperspectral data of corn leaves were obtained by ASD Field-Spectrometer. The original spectral data were processed by fractional differential (FD), and feature bands were extracted by competitive adaptive reweighted sampling method (CARS). Finally, multi-layer perceptron (MLP), K-nearest neighbor (KNN) and support vector machine (SVM) were used to distinguish the spectra of stressed leaves. The FD-CARS-MLP model constructed by the optimal MLP was selected to distinguish the spectral information of corn growth copper and plumbum pollution. The results showed that the FD-CARS-MLP model was better than the traditional methods in spectral discrimination of stressed leaves. The accuracy of the FD-CARS-MLP model could reach more than 98% in all test sets, and the accuracy of fractional differential discrimination of 0.1 and 0.2 orders could reach more than 99%. Corn leaves at the seedling stage and heading stage were selected for the feasibility test of the FD-CARS-MLP model. It was proved that the FD-CARS-MLP model had higher accuracy and more stability in identifying the spectral data of corn leaves under heavy metal stress, which could provide technology and methods for monitoring different heavy metal stresses of cereal crops.

Abstract:The first problem to be solved in potato cutting fast is the detection of potato seed bud eyes, an improved YOLO v5s-based potato seed bud eye detection method was proposed to improve seed potato eye detection performance. Firstly, by adding the CBAM attention mechanism, the feature learning and feature extraction of the potato bud eye images were strengthened. The influence of the potato surface background similar to the bud eyes on the detection results was weakened. Secondly, the weighted bidirectional feature pyramid BiFPN was introduced to increase the original information of bud eyes extracted by the backbone network and assign weights to feature maps of different scales, making multi-scale feature fusion more reasonable. Finally, it was replaced with an improved and efficient Decoupled Head to distinguish between regression and classification, speed up the convergence speed of the model, and further improve the performance of potato bud eye detection. The test results showed that the precision, recall rate, and average precision of the improved algorithm were 93.3%, 93.4% and 95.2%, respectively, which was 3.2 percentage points higher than that of the original algorithm in the mean average precision, and the precision and recall rate were improved by 0.9 and 1.7 percentage points. The comparative analysis of different algorithms showed that this algorithm had absolute advantages compared with Faster R-CNN, YOLO v3, YOLO v6,YOLOX and YOLO v7 algorithms. The mAP was increased by 8.4 percentage points, 3.1 percentage points, 9.0 percentage points，12.9 percentage points and 4.4 percentage points. In the actual detection application, the average recall rate of the improved algorithm was 91.5%, which was 17.5 percentage points higher than that of the original algorithm, and the missed detection rate was reduced. The method can provide technical support for the next step in the development of a sprout-eye identification device for the intelligent cutting of potato seed potatoes.

Abstract:Predicting soil organic matter (SOM) content based on images has the advantages of convenience and low cost. Interfered by objective factors such as soil type and moisture, there is still a gap between the detection accuracy of image prediction SOM content and traditional methods, which limits the promotion and popularization of related technologies. In order to improve the accuracy of image prediction of SOM content, a N_DenseNet multi-scale pooling module was added to DenseNet169 to improve the performance of the model by obtaining more dimensional features, and combined the development of SOM real-time detection APP on the Android side to realize the timely transmission of server and mobile phone data through intranet projection. Based on 350 soil samples from Youyi County, Heilongjiang Province, Changping District, Beijing City and Tai’an City, Shandong Province, high-definition images, R-band, red-edged band and near-infrared band images of in situ soil were obtained through mobile phones and multispectral drones to enrich data information, and image samples of soil samples under different moisture gradients were taken through indoor stress to alleviate the impact of moisture on the image. Compared with different deep learning models, the N_DenseNet trained based on multispectral image data performed the best, the overall performance was better than that of DenseNet169, the test set R2 was 0.833, RMSE was 3.943g/kg, and R2 was improved by 0.011 compared with the visible light data, which proved that the addition of R-band and red-edged and near-infrared images to the training process helped to improve the performance of the model, which proved the feasibility of the method. The mobile phone APP was connected to the background data to realize real-time data transmission, and realized the real-time detection of SOM content of soil samples in the field, and the model predicted R2 as 0.805 and the detection time was 2.8s after field verification, which met the needs of SOM content detection in the field and provided an idea for real-time detection of SOM content.

Abstract:In the case of dense forests, the accuracy of estimating underforest terrain using GEDI data and existing Tandem-X DEM digital terrain models has not been comprehensively evaluated. Aiming to focus on the dense forest situation as the main research object and using airborne data as real validation data. By extracting the longitude and latitude of the corresponding LiDAR spot, underforest terrain information, and data quality screening parameters of the GEDI L2A data product, to estimate underforest terrain data based on GEDI data. Compared with Tandem-X DEM data to estimate the underforest terrain under dense forest conditions, and further explore the effects of canopy height, forest coverage, and vegetation type on estimation accuracy. The R2 values of GEDI and Tandem-X DEM were 0.99 and 0.98, respectively. The RMSE, Average, and STD values of GEDI for estimating underforest terrain were 6.49m, -1.92m, and 4.42m, respectively. The RMSE, Average, and STD values of Tandem-X DEM for estimating underforest terrain were 18.15m, 14.63m, and 7.35m, respectively. In GEDI data, RMSE and Average were changed by 8.05m and 6.04m respectively in the case of mixed forest and sparse grassland, and in Tandem-X DEM data, RMSE and Average were changed by 21.63m and 26.43m respectively in the case of evergreen coniferous forest and farmland/natural vegetation. The experimental results indicated that there was a strong correlation between GEDI and Tandem-X DEM data and airborne validation data, and GEDI performed better evaluation criteria than Tandem-X DEM data. The surface vegetation types performed greater impact on the estimation of underforest terrain than canopy height and vegetation coverage.

Abstract:In order to detect the occurrence of rice koji disease quickly and accurately, and realize the early monitoring of rice koji disease in a large area, the airborne UHD185 hyperspectrometer was used to collect multiple sets of rice canopy hyperspectral image data with the disease area, and the image data was preprocessed to establish data sets. The classification training of healthy and diseased areas was carried out, and the recognition model of support vector machine (SVM) and principal component analysis (PCA) plus artificial neural network (ANN) was established to identify diseased rice, and the accuracy of the recognition model was verified by validating the samples. The support vector machine recognition model selected false color images under two sets of feature wavelengths. The first group of wavelength combination (TZH1) was 654nm, 838nm and 898nm, and the second wavelength combination (TZH2) was 630nm, 762nm and 806nm. The total commission error/omission error of the two sets of data reached 4.24% and 5.41%, respectively. Among them, the SVM model of the S-type kernel function had the best diagnostic performance, and the overall classification accuracy could reach 95.64% and the Kappa coefficient was 0.94, which basically achieved the purpose of accurately identifying rice disease areas. The recognition model of principal component analysis plus artificial neural network used the first three principal components, and the contribution rates were 93.67%, 2.80% and 1.24%, respectively, which were used as the optimal wavelength to establish the ANN recognition model. In the classification results, the nonlinear classification was better than the linear classification, the overall classification accuracy was 96.41% and the Kappa coefficient was 0.95. The results showed that through the diagnostic results of the support vector machine in the data of the two experimental groups, it can be seen that the classification accuracy of the recognition model using the support vector machine was stable overall, and there was no obvious difference in the diagnostic effect of the four kernel functions. In terms of overall classification accuracy, the nonlinear classification in the principal component analysis plus artificial neural network recognition model was 0.77 percentage points higher than that of the S-type kernel function classification of the support vector machine recognition model. Therefore, the nonlinear classification model in principal component analysis plus artificial neural network model was more suitable for early monitoring of rice koji disease.

Abstract:Tea picking point localization is one of the key technologies for selective tea picking. In the tea tree picking scenario, there are problems such as small scale of picking points, large background interference and complex lighting conditions, which lead to the problem of accurate segmentation of tea picking points. A semantic segmentation model based on multi-headed self-attentive mechanism combined with multi-scale feature fusion, RMHSA-NeXt, was constructed for the accurate segmentation of picking points in tea garden scenes. The attention module based on residuals and multi-headed self-attention mechanism was constructed to focus the model’s attention on the segmentation target and enhance the representation of important features. The features at different scales were fused by multi-scale structure (atrous spatial pyramid pooling, ASPP), in which strip pooling was used in the fusion process for the characteristics of picking points to reduce the useless. Finally, the information was decoded by convolution and upsampling, and the segmentation results were obtained. The experiment results showed that the model can segment the picking points effectively in the tea garden environment, and the pixel accuracy of the model reached 75.20%, the average region overlap was 70.78%, and the running speed reached 8.97f/s. The results showed that the model had the advantages of high accuracy, fast inference speed and small number of parameters, which can balance the accuracy and speed indexes well compared with other models. The research results can provide an effective and reliable reference for pinpointing tea picking points.

Abstract:Due to unknown flesh and bone boundaries in the hind legs of sheep carcasses, variable size and visibility constraints, the robot autonomous segmentation accuracy is low and easy to be blocked. An adaptive segmentation control method was proposed for the hind legs of sheep carcasses, and the segmentation test of sheep carcass hind legs was carried out to verify it. The method was centred on contact state perception and effectively extracted contact type features, contact abnormality features and contact direction features. LSTM-FCN deep spatio-temporal neural network was constructed to identify contact types, constructing deep self-coding network to estimate contact anomalies, and using principal component analysis to detect the main contact directions to achieve multimodal sensing of contact states. The robot imitated and learned human manipulation skills through dynamic motion primitives, and incorporated contact state sensing information to achieve adaptive adjustment of joint motion. The experimental results showed that the recognition accuracy of LSTM-FCN model on the validation set of sheep carcass hind leg segmentation was 98.44%, with a high recognition accuracy. The DAE model can better estimate the contact anomalies of the validation set samples and distinguish different contact states. Robot conducted practical segmentation tests based on adaptive segmentation control method. Compared with the control group, the maximum segmentation force was decreased by 29N and the maximum torque was decreased by 7N·m, proving the effectiveness of the method. The average maximum residual meat thickness was 3.6mm, the average segmentation residual rate was 4.9%, and the segmentation residual rate showed a negative correlation with the quality of sheep carcasses. It proved that the method had good generalization and accuracy. And the overall segmentation effect was good, meeting the requirements of sheep carcass hind leg segmentation.

Abstract:The small size of the chickens and the shading of the chickens from each other are factors that make it difficult to identify the daily behaviour of laying hens. To address this problem, a method of daily behavior identification of laying hens based on SEEC-YOLO v5s was proposed. By adding a SEAM attention module (separated and enhancement attention module) to the output part of the YOLO v5s model and introducing an EVCBlock module (explicit visual center) to the feature fusion part, the perceptual field of the model was expanded, the recognition ability of the model for occluded targets was improved, and the recognition accuracy of the model for the six behaviors of standing, feeding, drinking, exploring, feather pecking and grooming of laying hens was improved. A statistical method was proposed to calculate the duration of daily behavior of laying hens based on the ratio of video frames to video frame rate, and various behavioral changes of laying hens at different times of the day and throughout the day were analyzed. The improved model was encapsulated and packaged to develop an intelligent identification and automatic statistics system for the daily behavior of laying hens. The test results showed that the mAP of SEEC-YOLO v5s model for six behaviors recognition was 84.65%, which was 2.34 percentage points higher than that of YOLO v5s model, and compared with that of Faster R-CNN, YOLO X-s, YOLO v4-tiny and YOLO v7-tiny models, the mAP was improved by 4.30 percentage points, 3.06 percentage points, 7.11 percentage points and 2.99 percentage points, respectively. The method can provide effective support for daily behavior monitoring and health condition analysis of laying hens, and provide a reference for smart farming.

Abstract:Hyperspectral remote sensing crop growth monitoring technology, an essential instrument for developing contemporary precision agriculture, is characterized by non-destructiveness and real-time effectiveness. Taking leaf area index (LAI) of soybean at flowering stage under different levels of N application and mulching treatment as research object, the raw data for the hyperspectral reflectance of the soybean canopy were pretreated by using the 0~2 order differential transform processing (step 0.5). Based on five sets of pretreatment reflectance data, the optimum spectral index with a high correlation to the LAI of soybean at the blooming stage was the input data. And the support vector machine (SVM), random forest (RF), and BP neural network optimized by genetic algorithm (GA-BP) were used to construct the soybean LAI prediction model.The results showed that compared with the integer order and the raw hyperspectral reflectance, the spectral indices built from the fractional order differential preprocessed hyperspectral reflectance correlated better with the soybean LAI.The corresponding bands of different orders of optimal spectral indices concentrated in the red-edge band. The correlation between the spectral index and soybean LAI was increased and then decreased as the differential order was increased, and the accuracy of the prediction model showed the same pattern.When the input data were the same for all three machine learning techniques, the model created by RF had the highest accuracy. A thorough analysis determined that the soybean LAI prediction model built by using RF had the highest accuracy of prediction when the input variable was the 1.5-order differential optimal spectral index. The R2 of the model validation set was 0.880, the RMSE was 0.3200cm2/cm2, the NRMSE was 10.354% and the MRE was 9.572%. The research result can help advance the development of precision agricultural production by offering theoretical references for enhancing the inversion accuracy of soybean LAI hyperspectral prediction models.

Abstract:Vegetation moisture content is an important characterization of the sensitivity of farmland ecosystem. The spectral reflectance of two vegetation covers, alfalfa and corn were extracted, based on the UAV multispectral image data, and on the basis of which the red-edge band was introduced to calculate the improved spectral indices in order to increase the efficiency and accuracy of the inversion of vegetation water content by near-earth remote sensing. A back-propagation neural network (BPNN) was created after the five spectral bands and 25 indices were filtered by using the variable importance in projection (VIP), gray relational analysis (GRA), and Pearson’s correlation analysis. To find the optimum inversion model for vegetation water content under various crop covers, back-propagation neural network, partial least squares regression (PLSR), support vector regression (SVR), and random forest (RF) were used. The findings indicated that, among the three screening algorithms, the accuracy of the models following GRA and VIP was significantly higher than that of Pearson’s correlation analysis, and the inversion results were less volatile. Among the four machine learning algorithms, the SVR algorithm had a stronger nonlinear problem resolution ability and model robustness than BPNN, PLSR, and RF algorithms. In the nonlinear problem, the SVR algorithm outperformed the BPNN, PLSR, and RF algorithms in terms of analytical ability and model robustness. The validation set coefficient of determination R2 reached above 0.77 and its results can offer more accurate feedback on vegetation water content. The GRA-SVR based inversion model for vegetation water content had the highest accuracy in the two sample sites. The GRA-SVR validation set R2 of alfalfa cover reached 0.889, RMSE of 0.798%, and MAE of 0.533%;the inversion result validation set R2 of corn cover was 0.848, RMSE of 0.668%, and MAE of 0.542%. The research results can provide a theoretical basis for rapid and accurate inversion of vegetation water content.

Abstract:Aiming to predict apple production in Loess Plateau in a timely and accurate manner, based on historical weather records in a total of 86 counties in the apple producing areas of the Loess Plateau, the data of different meteorological feature variables (e.g., temperature, precipitation, and radiation in different apple growing months), spatial feature variables (e.g., latitude, longitude and elevation of meteorological stations), and meteorological disaster feature variables (e.g., time of freezing damage at flowering stage, time of continuous rain, and standardized precipitation evapotranspiration index (SPEI)) were extracted at first. The influential feature factors were determined according to Spearman correlation analysis. Nextly, the prediction models for apple relative meteorological yield were established based on different algorithms (i.e., gradient boosting decision tree, GBDT;support vector machine, SVM;Bayesian regularization back propagation artificial neural network, BRBP;multiple linear regression, MLR). At the same time, the optimal combination of model input feature variables was determined for each of the established yield prediction models. Finally, based on the optimal combinations of input feature variables in different apple growth periods and in different months of apple growing seasons, the prediction leading time were analyzed for different simulation models for apple relative meteorological yield. The results were as follows: the influential meteorological feature variables were the highest temperature, lowest temperature, air relative humidity, precipitation and solar radiation. The best model input variable combination was selected as the influential meteorological, spatial and disaster feature variables. Based on the best combination of model input variables, the GBDT and BRBP models had better prediction accuracy (r was 0.77, RMSE was 0.44;r was 0.70, RMSE was 0.44), while the MLR model performed the worst (r was 0.63, RMSE was 0.49). In different growth periods of apples, the GBDT and BRBP models could obtain relatively high apple yield prediction accuracy in each growth period, while the SVM and MLR models could obtain relatively ideal simulation results in apple fruit expansion period. In each month of the apple growing season, the GBDT, SVM, BRBP and MLR models could realize early prediction of apple relative meteorological yield about one to two months before apple maturity. The research result can provide a scientific foundation and technical reference for apple yield prediction on the Loess Plateau.

Abstract:In order to explore the effects of different soil and water conservation tillage techniques on nitrogen utilization and greenhouse gas emissions in sloping farmland of black soil in Northeast China, a field experiment was conducted. Seven tillage treatments were set up: transverse slope planting ( TP ), ridge to the district field ( RF ), subsoiling tillage ( SF ), transverse slope planting + subsoiling tillage (TP-S), ridge to the district field + subsoiling tillage (RF-S), transverse slope planting + ridge to the district field (TP-R), and down-slope cultivation (CK). Explore the effects of soil and water conservation tillage techniques on soil nutrient status, greenhouse gas emission, nitrogen absorption and utilization, and yield of black soil slope farmland in Northeast China. The results showed that during the whole growth period of maize, soil and water conservation tillage treatment significantly increased maize yield, organ nitrogen transport rate and nitrogen use efficiency, and some soil and water conservation tillage measures could also significantly reduce N2O and CO2 emissions. Among them, at the maturity stage of maize, the plant yield of soil and water conservation tillage measures were increased by 3.39%~26.43% compared with that of CK treatment, and TP-S treatment had the highest improvement effect. For nitrogen use efficiency, soil and water conservation tillage technology were increased by 25.23% (RF treatment ) ~ 76.98% (TP-S treatment) compared with CK treatment, and the improvement effect was significant. For CO2 emissions, except for SF treatment, the remaining soil and water conservation tillage treatments were significantly lower than that of CK treatment. However, for N2O emissions, TP treatment, TP-S treatment and TP-R treatment were significantly lower than that of CK, while SF treatment, RF treatment and RF-S treatment were significantly higher than that of CK treatment. Therefore, it was suggested that the local maize planting should adopt the soil and water conservation tillage technology of transverse slope planting+subsoiling tillage.

Abstract:To investigate the effect of applying biochar and straw on the improvement of agricultural soils before and after freezing in seasonal permafrost areas, black soil was selected as the research object, and four different regulation measures (BL: blank control;CLS: application of biochar;JLS: application of straw;CJLS: combined application) were set up based on field experiments to analyze the stability of soil aggregates, pore size distribution characteristics and soil moisture characteristic curves. The results showed that the variation of soil physical properties such as soil water holding capacity and hydraulic conductivity was investigated based on the variation of soil infiltration and saturated hydraulic conductivity (Ksat). The results showed that the application of biochar and straw effectively suppressed the adverse effects of freeze-thaw cycles on soil structure, and effectively maintained the stability of soil aggregates. The application of exogenous biomass materials improved soil pore distribution, increased the proportion of intermediate pore size (0.3~100μm) in the early freezing period, increased the proportion of intermediate pore size 19.05%~35.04%, increased the proportion of soil voids (greater than 100μm) by 4.33%~16.22%, and decreased the proportion of very small pore size (0~0.3μm) by 9.09%~18.18% under the combined effect of exogenous biomass materials and freeze-thaw alternation, and with the CJLS treatment performing the best. In the pre-freezing period, the application of biochar and straw increased the cumulative soil infiltration at 60min under tension -5cm by 73.68%, 60.52% and 151.10%, while in the thawing period, the positive effect exerted by freeze-thaw aging of exogenous biomass materials was diminished, and the cumulative soil infiltration was increased by a maximum of 112.28% only. Meanwhile, the application of biochar and straw increased the saturated soil water content before and after freeze-thaw and enhanced the soil water holding capacity, which contributed to the improvement of soil drought resistance in spring. The plant available water content was increased under the dual effect of exogenous biomass materials and freeze-thaw cycles, while Ksat was increased both before and after freeze-thaw, but the difference between the CLS and JLS treatments and BL were gradually slowed down during the thawing period. The results of the study had important implications for the rational application of biochar and straw resources, revealing the response mechanisms of improved soil physical properties.

Abstract:In order to reveal the effects of aged biochar on greenhouse gas emissions and maize growth under different irrigation amounts and water salinities, two biochar addition rates: 0t/hm2（B0）, 60t/hm2（B1）, two irrigation water salinities: 0.71g/L（S0） and 4.0g/L（S1）, and two irrigation levels of full irrigation (W1) and deficit irrigation （W2，1/2 W1) were set and the field experiment of spring corn was conducted at the National Field Scientific Observation and Research Station on Efficient Water Use of Oasis Agriculture in Wuwei of Gansu Province from April to September 2022. The results showed that the cumulative CO2 emission of deficit irrigation was decreased by 24.13%~52.68% compared with that of full irrigation. However, there was no significant difference in N2O emissions between the two irrigation levels. The cumulative emissions of CO2 and N2O were increased by 9.06%~24.79% and 9.95%~18.03% respectively, due to brackish water irrigation. Under the four irrigation treatments, aged biochar decreased the accumulated emissions of CO2 and N2O by 7.33%~18.78% and 21.14%~29.76%, respectively. There was no significant impact on CH4 emissions among different treatments. Deficit and brackish water irrigation inhibited the growth of crops. Aged biochar significantly increased the biomass of spring maize by 7.86%~25.82%, but its effect on the yield of maize under W1S0, W1S1 and W2S0 treatments was not significant, and the yield of maize under W2S1 treatment was significantly reduced. At the same irrigation level, brackish water irrigation increased the global warming potential, while using aged biochar and deficit irrigation reduced it. Brackish water and deficit irrigation significantly reduced soil carbon budget by 17.70%~65.36% and 37.30%~71.96%, respectively. Under different irrigation treatments, aged biochar significantly increased the carbon sink of farmland with a value of 15.86%~33.52%. In general, the application of aged biochar under W1S0, W1S1 and W2S0 treatments had slightly increased the maize yield, significantly reduced the global warming potential and promoted the net carbon income of farmland. Therefore, under these three irrigation methods, the application of aged biochar would increase the local economic and environmental benefits.

Abstract:Scientific and efficient irrigation management is the key to improving water productivity. To address the problem that there are significant differences in soil water and heat conditions before and after mulching, and it is still unclear how irrigation parameters affect crop growth and water consumption, field experiments were conducted at the National Field Scientific Observation and Research Station on Efficient Water Use of Oasis Agriculture in Wuwei of Gansu Province, using seed maize as the research object. Three dripper discharges D1 (2.0L/h), D2 (2.5L/h), D3 (3.0L/h), and five irrigation intervals P1 (6d), P2 (8d), P3 (10d), P4 (12d), P5 (14d) were designed to investigate the effects of different irrigation management schemes on the growth, yield, transpiration, and water consumption of seed maize. The results showed that higher dripper discharges and shorter irrigation intervals were more favorable for the growth and leaf area index of seed maize. Under D1 treatments, plant height, and leaf area index were increased firstly and then decreased with the increase of irrigation interval, and reached the maximum value at P2. The grain yield showed a significant quadratic relationship with the irrigation interval, and the maximum values under D1 and D3 appeared in P2. When the irrigation interval was 6~10d, the highest yield was obtained under the D2 dripper discharge. The daily-scale and hourly-scale liquid flow rates of seed maize single-plant were increased with the increase of dripper discharge and decreased with the increase of irrigation interval. The peak value of the hourly liquid flow rate in D3 treatment was increased by 85.88%~127.02% and 117.80%~151.89% compared with that in D2 and D1, respectively. When the irrigation interval was increased from P1 to P5, the peak value of the hourly liquid flow rate was decreased by 23.56%~31.48%. The transpiration of seed maize was increased with the increase of dripper discharge. Within a certain range (P1 to P4), increasing the irrigation interval significantly reduced the water consumption of seed maize throughout the growth period. At irrigation intervals not greater than P3, D2 had the lowest water consumption, which was 1.86%~4.14% lower than that of D1, and higher water productivity was obtained. The highest water productivity was obtained at P2 treatment for both D1 and D2 dripper discharge. The results showed that the dripper discharge of 2.5L/h (D2) and the irrigation interval of 8~10d were the optimal irrigation scheme for seed maize under film in dry areas for efficient water conservation, while a higher yield and water productivity could be achieved with an irrigation interval of 6~8d.

Abstract:Bone char, as an important product for the resource utilization of bone meal of different animal sources, contains rich pore structure and active sites that can be used as catalyst carriers or adsorbents in water, soil and other environments. In order to characterize the pore structure of different animal-derived bone char particles in three-dimensional and in situ visualization, porcine and bovine bone char particles with pyrolysis temperatures of 400℃, 500℃, 600℃, 700℃ and 800℃ were used. The Micro-CT scanning conditions of bone char particles were tube voltage 50kV, tube current 200mA, image resolution 10μm;Micro-CT image binarization method was Adaptive mean-C. The results showed that the Micro-CT image filtering method combining UM and ACE was optimal. The optimal parameters of the UM method were Radius value set to 5, Amount value set to 90%, Threshold value set to 0, and ACE using adaptive model. The tomographic structure images and volume model of bone char particles at different pyrolysis temperatures were similar to the electron microscopy results. The quantitative characterization results of specific surface area, porosity and pore size distribution of bone char particles were in good agreement with those determined by laboratory BET method. The results can provide a non-destructive and visualized means for in situ 3D analysis of bone char pore structure.

Abstract:In order to solve the problems of the existing wireless detection system cannot accurately and effectively reflect the environmental changes of the three-dimensional space in the greenhouse, as well as the large positioning error of the sensor node and the high hardware cost, an intelligent greenhouse three-dimensional temperature and humidity detection system was designed based on ultra wide band (UWB) positioning. The system located each sensor node through a self-designed STM32F system board integrating UWB positioning module, and it was equipped with AHT25 high-precision sensor to collect environmental data. The UWB main base station used the 4G network communication module to send the sensor data and location information to the host computer, and visualized the three-dimensional temperature and humidity field of the greenhouse according to HTML5 technology on the Web side to complete the remote detection of the three-dimensional temperature and humidity of the greenhouse. The system positioning test proved that the accuracy of each sensor node was mainly concentrated in the range of 10~30cm, the measurement position error of some nodes is greater than 50cm, the maximum packet loss rate of each node was 2.5%, and the average packet loss rate was 1.9%, which met the basic needs of greenhouse measurement and is of great significance for detecting greenhouse thermal defect areas and studying the suitable environment for plant growth.

Abstract:Water-soluble β-glucan with different concentrations was added to the simulated grape juice and then inoculated with S. cerevisiae Aroma White strain to start alcohol fermentation. The antioxidant activities of the tested strain were detected on the 2nd, 3rd, 4th and 7th day of fermentation process respectively, and the volatile aroma compounds in the final simulated wine samples were determined by headspace solid phase microextraction combined with gas chromatography-mass spectrometry (HS-SPME-GC-MS) to clarify the correlation between the antioxidant activities and aroma compounds. The results showed that the activities of adenosine triphosphatase (ATPase), superoxide dismutase (SOD) and the content of glutathione (GSH) in yeast strain cells treated with water-soluble β-glucan were significantly higher than those in control group （P<0.05), while the contents of reactive oxygen species (ROS) and malondialdehyde (MDA) were significantly lower than those in control group （P<0.05). Water-soluble β-glucan with concentration of 300mg/L significantly increased the content of volatile compounds，especially linalool，phenylethanol，ethyl octanoate and ethyl decanoate with floral and fruity. These compounds were positively correlated with the activities of ATPase and SOD as well as the content of GSH in S. cerevisiae cells. To sum up，water-soluble β-glucan with concentration of 300mg/L can significantly enhance the antioxidant capacity of S. cerevisiae Aroma White strain cells，remarkably increase the contents of esters and higher alcohols in simulated wine samples，and it had the application potential to improve the fermented aroma of the wines.

Abstract:Developing electricized and intelligent modern agricultural equipment system and improving the modernization level of agricultural equipment is one of the important measures to vigorously promote the comprehensive revitalization of rural areas and quickly realize the “modernization of agriculture and rural areas”. Because of its potential advantages of green, intelligent and efficient, electric tractors have become one of the effective means to support the green and sustainable development of agriculture and ensure the safety of food production. Among them, as the core and key of electrification technology of tractor, driving motor is the power source of electric tractor, which directly determines the operation efficiency and tillage quality of the whole machine. The closed-loop control system of permanent magnet motor needs accurate rotor position information, while the reliability of traditional mechanical sensors is difficult to guarantee under bad working conditions. In order to improve the system reliability, the sensorless control method was studied. In the traditional sensorless control based on sliding mode observer, the problem of poor parameter robustness and stability seriously affected the control effect of sensorless operation. Therefore, a sensorless control method based on fuzzy sliding mode observer and fuzzy phase-locked loop was proposed to effectively solve this problem. Firstly, fuzzy controller was used to process the control parameters of the sliding mode observer and phase-locked loop, and these parameters were adjusted in real time according to the actual working conditions of the motor. Secondly, the recursive least squares adaptive linear harmonic extractor was used to effectively filter the high harmonic component of the back electromotive force, which avoided the influence of the high frequency component on the observed results. The experimental results showed that the proposed control method can improve the accuracy of speed and position estimation.

Abstract:The existing tractor traction performance prediction model does not include the influence of front and rear wheel different performance, load transfer and uncoordinated front and rear axle movements on the travel reduction ratio and motion resistance ratio, resulting in the low accuracy of field traction performance prediction for four-wheel drive tractors. Starting from the driving characteristics and load characteristics of agricultural tires, and soil-tire driving model and front and rear tire load model were established, including axle load transfer respectively by introducing wheel numeric and mobility number;on the basis of traction force analysis, the prediction models of travel reduction ratio and motion resistance ratio for the whole machine were established respectively considering the influence of actual front and rear axle motion uncoordination on the overall chassis operation. The traction performance prediction model for four-wheel drive tractor containing tire specifications, soil characteristics, uncoordinated front and rear axle motion characteristics of the whole machine, and transmission efficiency was derived. The prediction algorithm and process were designed based on the two-dimensional iterative method to solve the problem arising from the multivariate and nonlinearity of the model;an example analysis and application were carried out by using the research method;a field traction test was designed and carried out to verify the validity of the prediction model, and the results showed that the errors of the simulated values in the maximum traction force and the traction force corresponding to the characteristic slip rate were 1.41% and 1.74%, respectively, and the error in rolling resistance was 0.64%, which was a large improvement in accuracy over the control group and a small overall error.

Abstract:The dynamic performance and inertia coupling strength of a 3T1R decoupling parallel mechanism (PM) were analyzed. Firstly, the kinematics of the mechanism was established, and the forward kinematics and inverse kinematics were given to obtain the Jacobian matrix of the moving platform, where the velocity and acceleration of each link and the moving platform were derived. Based on the Newton-Euler method, the inverse dynamics model of the mechanism was established considering the gravity of the components and the external load. The driving forces of the PM were solved, which was then verified by ADAMS dynamics simulation. At the same time, the influences of acceleration and the attitude angle of the moving platform on the driving force of the branch chain were analyzed based on the established dynamics model and the analysis results can provide theoretical basis for trajectory planning of the mechanism. Finally, an inertia coupling evaluation index was proposed based on the inertia matrix in the joint space, which represented the coupling strength of the driving branches when the PM worked at different poses in the workspace. Then the distribution law of the index in workspace was studied and compared with that of the Quadrupteron mechanism before decoupling. The results showed that the decoupling of the PM not only reduced the coupling strength between the branches, but also made the distribution of the coupling strength in the workspace more consistent, which improved the isotropy of the dynamic performance of the mechanism.