Abstract:Stalks are not only the main by-products of crops, but also abundant biomass resources in the world. Stalk cutting is a direct interaction process between rigid body (machinery parts) and flexible body (stalks), which plays a pivotal role in the treatment of stalks. The cutting performance of machinery, the biological characteristics of the stalks, and the coupling between them all affect the cutting process. Stalk cutting is closely related to the efficient and low damage harvest of crops and the comprehensive utilization of stalk resources. The research on the cutting interaction process between cutting blades and stalks is an important aspect of the deep integration of agronomy and agricultural machinery, which has practical significance for agricultural production and ecological development. To this end, a comprehensive review and analysis of research progress at home and abroad around the related issues of stalk cutting was made, specifically as follows: focusing on mechanical parameters and constitutive model of stalks, the relationship between biological characteristics and mechanical parameters, testing methods and equipments, establishment and application of constitutive model of stalks were described; combined with the structural parameters and types, wear resistance and self-sharpening performance, the structural types and material properties of the cutting blades were introduced; according to the cutting interaction process between cutting blades and stalks, the cutting principle and the cutting technical objectives of high efficiency, low power consumption and low damage were systematically introduced; from the specific parameters and target values of experimental research to the different types of simulation research, the application of the two research methods in stalk cutting was summarized. On this basis, combined with the existing problems, the future development directions of stalk cutting field were discussed. It contributed to comprehensively understanding the phenomenon of stalk cutting and providing a reference for the in-depth research of stalk cutting in agricultural production. 

Abstract:To obtain the physical parameters and contact parameters of discrete element modelling (DEM) for simulating soil and interaction between soil and soilengaging components in mulberry field, a method combining the experiments and the DEM simulations for calibrating the contact parameters of soil particles in mulberry field was proposed. Firstly, the particle size distribution and physical parameters of soil, such as the angle of repose of soil, sliding friction angle of soil and 65Mn steel, shear stress of soil, cohesion and angle of internal friction of the soils sampled at different depths in the mulberry field, were measured respectively with the powder instrument, inclinometer and equal strain direct shear apparatus. Then an unequal-diameter sphere particles model was built according to the measured particle size distribution by using the EDEM. On this basis, the coefficient of the static friction, coefficient of rolling friction and coefficient of restitution between soil particles and between soil and 65Mn steel were used as test factors, and the angle of repose of soil and the sliding friction angle between the soil and 65Mn steel were used as target values to construct a central combined experimental design (CCD) scheme (three factors and five levels). Subsequently, by analyzing with the Design-Expert software, the simulated optimum values of the coefficient of static friction, coefficient of rolling friction and coefficient of restitution between soils were calibrated to be 0.89, 0.45 and 0.43, respectively; the simulated optimum values of coefficient of static friction, coefficient of rolling friction and coefficient of restitution between soil and 65Mn steel were calibrated to be 1.15, 0.05 and 0.4, respectively. The simulation experiments for verifying the values of the angle of repose of mulberry field soil and the sliding friction angle between the mulberry field soil and 65Mn steel were performed with the simulated parameters such as the coefficient of static friction, coefficient of rolling friction as well as coefficient of restitution. The simulating results showed that the relative errors between the simulating values and the experimental values for the angle of repose and for the sliding friction angle between the mulberry field soil and 65Mn steel were 1.69% and 2.88%, respectively. On the basis of these simulating results, using the soil shear stress gained from the experiments as the judgment standard, the bond parameters of the Hertz-Mindlin with the Bonding contact model for describing the particles of soils were calibrated with the trialanderror method with measured soil internal friction angle as the target value. The normal bond stiffness and the tangential bond stiffness were calibrated to be 1×108N/m3 and 5×107N/m3, both of the critical normal stress and the critical tangential stress was calibrated to be 10 kPa, and the contact radius was 1.1 time of radius of the particles, the simulating angle of the internal friction was 30.24°, the relative error between the simulation value and the experimental value of the angle of internal friction was 5.53%.

Abstract:Aiming at the problems of high resistance, high energy consumption of traction in subsoiling, low operation efficiency, cultivated layer restoration and soil remediation, an active lubricating drag reduction scheme for curved subsoiler for low water content and high compactness farming soil was proposed. Firstly, on the basis of the three-dimensional model of the curved subsoiler obtained by three-dimensional scanning, the interaction between the subsoiler and the soil particles during the operation of the curved subsoiler was analyzed by the discrete element method, and the maximum friction contact surface of the subsoiler body was determined as the active lubrication drag reducing surface. Secondly, the idea of active lubricating drag reduction was proposed, and the body fluid distribution and body surface texture of earthworm were used for reference. To form the active lubrication and drag reducing curved subsoiler, the surface configuration in the form of groove, the structure of the lubricating surface such as throttle hole and the pumping system of the lubricating medium were designed on the maximum friction contact surface of the subsoiler surface and the subsoiler tip, respectively. Finally, the operation speed and lubricating fluid flow rate were used as the test factors, and the horizontal operation resistance was the main indicator. Field tests were carried out under two soil conditions. The test results showed that in the brown soil environment, when the operating speed was 3km/h and the lubricating fluid flow rate was 12L/min, the drag reduction rate could reach 13.48%. In the saline-alkali environment, when the operating speed was 1.87km/h and the lubricating fluid flow rate was 12L/min, the drag reduction rate could reach 19.87%. It was preliminarily proved that the active lubricating drag reduction operation mode had a good drag reduction effect in the soil with low moisture content and high compactness.

Abstract:The spatial distribution quality of straw in the soil has a significant effect on the decomposition rate of straw and the distribution of soil nutrients. To explore different rotary tillage operation parameters on the quality of the straw spatial distribution, simulation model of rotary tillage straw returning rotary tillage operation simulation process was built based on the discrete element method, and combined with field experiment with different speeds and qualities of straw spatial distribution of the knife roller speed comparison to test and verify. The simulation and field experiment areas were divided vertically and horizontally, the number of straw in each area was calculated, and the spatial distribution quality of straw under different rotary tillage operation parameters was evaluated using the coefficient of variation of straw ratio as the index. The results showed that in vertical stratification, the variation coefficient of straw ratio in each layer was increased with the increase of cutter roll rotation speed, and the minimum value was at 240r/min, and the simulation value and test value were 60.09% and 80.65%, respectively. However，with the increase of advancing speed, the coefficient of variation was decreased at first and then increased, and the coefficient of variation was the smallest at 0.50m/s, and the simulation value and experimental value were 61.00% and 79.90%, respectively. In the horizontal division treatment, the increase of cutter roll speed had no obvious regular effect on the variation coefficient of straw proportion in each layer, but the increase of advance speed could reduce the variation coefficient in the longitudinal division area, and the minimum value was 11.36% and 20.12% at 0.75m/s. The variation trend of simulation value and field value was basically consistent. The maximum mean difference between simulation and test values of vertical distribution and horizontal distribution was 22.13% and 12.23%, respectively, and the error was within the acceptable range. The discrete element simulation can simulate the spatial distribution of straw under different rotary tillage operation parameters, which can provide support for the rapid prediction and evaluation of straw returning operation quality. It can also provide a theoretical basis for the selection of operation parameters of rotary tiller.

Abstract:As the channels for straw usage are steadily increasing, sub-plot level straw information abundance is increasingly a topic drawing the attention. However, a systematic investigation on sub-plot straw information abundance is lacking. A field study was conducted in a post-harvest wheat field to explain sub-plot straw information abundance. Multiple indicators were proposed, including straw mass distribution, fragmented straw clumping index, side-shadow of standing straw, fragmentation index of straw pieces, lodging index of standing straw etc., and specific digitization techniques were designed, i.e. cell sampling + weighing method, machine vision with uniform layer straw on white plate method, background plate side vison method, sieving method, supplemented with manual counting for calibration. In addition to the acquired indices, normalized parameters of straw information were analyzed and the correlations among indicators were visually explained. Results showed that techniques investigated in this work were effective for collecting straw information for improved info-abundance. Information technologies proposed were powerful when applied for interpretating the underlying causes for each indicators, e.g. straw distribution within the cutter width, influence from wheat row width, straw exits performance, straw damage status by the harvesters etc. Correlation analysis also indicated relationships among observed indicators, e.g. similarity between straw mass distribution and fragmented straw distribution was 0.89, yet the value was 0.43 with standing straw distribution. Fragmented straw has 0.64 in similar with clumping index, while that for standing straw and side-shadow was 0.48. Achieved results by this work could be useful for future when the subplot straw information abundance and its utilization were to be systematically investigated.

Abstract:Due to the nonconformity of coating formula and low-level automation of seedcoating machines, seed-coating technologies in China exhibit low coating-success rates, detection accuracy and efficiency. To solve these problems, an recognition and detection system for pelleted coating seeds was designed to recognize the coating seeds with spherical shape. Firstly, a vision shooting platform was built, and the captured image was transferred to the recognition control system for image pre-processing. Secondly, according to the characteristics of different types of coated seeds after image processing, a recognition and detection algorithm was proposed. According to the difference of image area ratio between damaged coated seeds and other coated seeds, the recognition of damaged coated seeds was realized by advanced morphological processing. The identification of multiple seeds and qualified seeds was realized according to the difference of the pixel values of multiple seeds and qualified seeds. Finally, the total number of seeds, the number of qualified seeds, the number of multiple seeds and the number of damaged seeds were detected, and the qualified rate of coating was calculated. The experiment was carried out on red clover seeds. The results showed that the time of single image acquisition, processing and recognition was about three seconds. The accuracy of using advanced morphological treatment to identify damaged coated seeds was 98.8%. When the test samples was 200, the success rate of the total number recognition algorithm was 99.1%, the relative error rate of qualified coating seeds and multiple coating seeds was 1.18% and 3.36% respectively. All these results suggested that the developed recognition and detection system realized the functions of shooting, image processing, detection and recognition, as well as output and storage of results. Therefore, the developed recognition and detection system can be used to fulfil non-destructive testing of coating seeds.

Abstract:In order to further improve the planting efficiency of the high-speed automatic transplanter for vegetable plug seedlings, a three-arm gear train planting mechanism was proposed, and an approximate multi-pose motion synthesis method based on genetic algorithm (GA) was introduced. Firstly, with the key pose (position and orientation) data on the ideal planting trajectory as constraints, the approximate multi-pose motion synthesis optimization model of gear train planting mechanism simplified model (planar RR mechanism) was established by the condition of invariable link length, and the optimal structural parameters of the mechanism were obtained by using Matlab GA toolbox. Then, the total transmission ratio of the gear train was calculated and distributed by the motion parameters of revolute joints of the planar RR mechanism, so as to realize the design of gear train planting mechanism. Finally, the structure design, simulation analysis and test verification of the three-arm gear train planting mechanism were carried out. The results showed that the actual motion trajectory and posture of the mechanism were basically consistent with the theoretical design. When the planting frequency was 120 plants/(min·row) and the theoretical planting spacing was 300mm, the planting success rate was 96.7%, the actual average planting spacing was 298mm and the average hole width was 70mm, which met the requirements of high-speed transplanting. The correctness of the proposed method and practicability of the three-arm gear train planting mechanism were verified.

Abstract:Considering the practical problem that the high-speed planting operation leads to the seed feeding capability insufficient and the difficulty of precise adjustment of feeding quantity of the existing centrifugal metering device, a prototype of spinning disc high-speed metering device for rapeseed with a spiral was developed. Based on the mechanical and physical properties and parameters and the demand of planting quantity for rapeseed, a spiral seed feeding device was proposed and the mechanical model of seed feeding were established and the main structural parameters were analyzed and determined. A three-factor quadratic orthogonal combination test was performed to establish a mathematical model between seed feeding rate, stability variation coefficient of seed feeding rate, damage rate and speed, blade width, and lead. The analysis result showed that the sequence of factors affecting the seed feeding rate were lead, blade width and speed, and the the sequence of factors affecting the stability variation coefficient of seed feeding rate and the damage rate were both speed, lead and blade width. The optimal parameter combination was established when the speed was 81r/min, the blade width was 4mm and the lead was 15mm. The bench verification test results showed that the seed feeding rate was 92.7g/min, the stability coefficient of variation was 0.32%, and the damage rate was 0.29% of the seed feeding device under the optimum combination of parameters. The seed feeding rate was varied from 36.55g/min to 190.94g/min, the stability variation coefficient of the seed rate was less than 1.29%, and the damage rate was less than 0.5%. The field experiment showed that when the planting speed of the seeder was 10km/h, the uniformity variation coefficient of rapeseed seeding was 9.4%, and the planting density was 48~60 plants/m2, which could complete high-speed seeding. It could meet the performance requirements of rapeseed and would provide a reference for the structure improvement of spinning disc metering device for rapeseed.

Abstract:Row-oriented spraying technology can improve the utilization rate of pesticides, protect the environment and reduce pesticide residues. A vision based row-oriented spray control system for field cabbage was established. The improved ExG algorithm was used to extract color information, and the method of OTSU and morphological opening and closing operation were used to segment crops and background. A method of cabbage crop row localization and multi row adaptive ROI extraction was proposed. In the ROI of strip segmentation, the feature point set was collected based on the limited threshold vertical projection, and the crop row centerline was obtained by linear fitting of the feature point set by the least square method. The offset information of crop rows was obtained based on the geometric relationship of the centerline. A row offset compensation model was established based on the kinematic characteristics of the row mechanism, and row-oriented spray control system based on PID trajectory tracking algorithm was designed. Laboratory tests showed that the accuracy of crop row recognition was 95.75%, and the average algorithm time-consuming was 77ms. Field tests showed that under different periods of illumination, the recognition algorithm had the best test results in the time periods of 09:00—11:00 and 14:00—16:00, and the average recognition deviation was kept below 2.32cm. In the weed press test, the average accuracy rate of the recognition algorithm was 95.56%, indicated that the algorithm had strong robustness.in the comparison test with other recognition algorithms, the algorithm proposed had the shortest average time consumption and the highest recognition accuracy rate, and it could be used for real-time operations.in the field row-oriented spray control system tests, the system row-oriented accuracy rate reached 93.33%, and the control algorithm could control row-oriented deviation within 1.54cm, which could meet the requirements of practical field applications.

Abstract:In order to solve the problem that the accuracy of plant protection machinery in the process of target application in large fields is reduced due to the drift of fog droplet deposition, a research on the method of fog droplet directional deposition control was carried out based on the robot of target application in large fields. Based on the target spraying droplet population, the working principle and operational characteristics of target application were explained, and the causes of droplet population deposition shift were analyzed. Combined with the prescription chart target position information to control the nozzle for fast action, the target spraying simulation test was carried out on a leveled field with target spraying accuracy and deposition offset distance as test factors, and the group meeting the operational requirements was selected for the verification of target application in a large field. The test results showed that the average accuracy of target application on level ground at 0.5m/s, 1.0m/s, 1.5m/sand 2.0m/s was 99.8%, 98.4%, 95.9% and 76.5%, respectively, and the deposition offset distance was 3.8cm,5.4cm, 7.5cm and 10.0cm, respectively. The fieldtotarget application accuracies of the operating speeds were 98.7%, 96.7%, and 95.3% at speeds of 0.5m/s, 1.0m/s and 1.5m/s, respectively. The results showed that the nozzle control method based on GNSS/IMU Kalman filter for droplet deposition position prediction can meet the demand of target application in large fields. This target application simulation method allowed accurate recording of target application accuracy and droplet deposition offset distances in a near natural environment.

Abstract:Aiming at the problems of low accuracy of variable spray in wide field and insufficient consideration of the influence of agricultural machinery speed changes on the spray effect, a variable spray system with adaptive following speed was designed based on 3WF-1000 sprayer, which could improve the efficiency of accurate application of pesticides and fertilizers. The system consisted of four parts: early warning monitoring, speed monitoring, core control and execution agencies. Based on the multi-sensor real-time monitoring of speed, flow rat, pressure and liquid level height, the system used Bisector fuzzy control algorithm to optimize the system control. It realized the dynamic control of the change angle of the proportional valve and achieved the goal of accurate control of outlet pipe flow. In order to verify the superiority of the system algorithm in accurate variable application, the PID, Bisector fuzzy and Centroid fuzzy control modes were modeled and simulated under the Matlab platform. Through comparison，it could be seen that the Bisector fuzzy control was superior to the other two control modes in terms of adjusting time, overshoot and steady-state error. Field trials were carried out, such as nontravel fixed speed, fixed speed following, dynamic speed following and spray volume per unit area. In three speed states, the variable spray system adjustment times to stable operation were 13.4s, 27.6s and 17s, respectively, the maximum absolute error ratios of unit area spray volumes were 1.20%, 2.27% and 2.87%, respectively, which showed that the control system could meet the accuracy requirements of field accuracy application.

Abstract:In order to solve the problems of low intelligence and inability to monitor the operation of the machine in the process of carrot combine harvest in China, an intelligent monitor system which can be mounted on the carrot combine harvest machine was developed. The intelligent monitor system mainly included the carrot combine harvest adaptive speed regulation module, the carrot jam monitor module, the carrot fruit counting module, the man-machine interaction module and the position information module. The monitor system was mainly controlled by STM32F103 single chip microcomputer, information was transmitted by CAN bus, and a variety of sensor fusion technologies were applied to realize information collection and regulation of carrot joint harvest operation. Carrot harvest adaptive tape speed regulating module based on the fuzzy PID control algorithm collected machine operating speed, clamping angle conveyor belt speed and gripping delivery mechanism through the sensors, by using pulse width modulation control solenoid valve opening clamping conveyor belt speed adjustment to realize the carrot harvest machines adaptive adjusting job status. The experimental results showed that the model had good robustness and low overshoot. Field experiments showed that the monitor accuracy of each module was not less than 96%, the error value of the adaptive belt speed regulation module was not more than 0.1m/s, the belt speed response time was no more than 0.8s, and the adjustment time was no more than 1.6s. The intelligent monitor system can meet the requirements of field operation and realize the real-time monitor of carrot combine harvest and automatic control of clamping conveyor belt speed, which was beneficial to the development of carrot intelligent combine harvest.

Abstract:In order to solve the low loading efficiency caused by uneven loading during the cooperative operation of combine-harvester and transport vehicle, a three-dimensional point cloud based dynamic uniform loading method was proposed. The three-dimensional point cloud of the stowage in the grain box obtained by the camera was used as a state feedback to build up the uniformity evaluation method. With the purpose of accomplishment of the most uniform loading state, the position of unloading and loading point was adjusted in real time to keep the granary in the uniform loading state. According to the established grain pile model and camera occlusion model, the height of grain bulks within camera blind area was estimated with the minimum expected error to complete the defective point cloud which caused by mutual occlusion between heaps. Three loading states of light load, medium load and heavy load which may occur during loading were measured based on the constructed experiment platform. The average error of the blind area height estimation was less than 5cm. The simulation results showed that the proposed method can load the granary from arbitrary initial loading state to uniform loading state in a finite loading cycle. The maximum height variance of the load was 1cm2 under the condition that the initial loading state of the grain box was empty and loading with average height increment of 2cm. The height variance of loading in stable state was positively correlated with the single loading amount according to single factor analysis. Compared with the traditional fixed points loading method, the loading capacity of dynamic uniform loading method was twice higher when they both stopped the loading process at the maximum loading height of 0.35m.

Abstract:Potato harvest mechanization is the weakest link in the whole mechanized production of potato, in which stem and leaf chopping is an important part of potato mechanized harvest. Stem and leaf chopping has the advantages of accelerating the separation of potato seedlings and tubers, promoting potato epidermal hardening, reducing disease transmission, reducing epidermal damage and missing harvest during excavation, which can effectively reduce the loss rate of mechanical harvest per unit area. At the same time, potato stem and leaf crushing can effectively avoid potato seedling winding operation machinery, reduce the working load of harvester, improve the reliability of machinery, improve the operation speed of harvesting machinery and increase soil fertility. Aiming at the problems of low qualified rate of breaking length, high potato carrying rate and slow working speed of the active potato stem and leaf chopper, a full ridge imitation form of stem and leaf chopping knife roller was designed, the working process of the cutter was analyzed, the mathematical models of cutter movement, cutter stalk collision and stalk picking were established, and the main parameters and value range affecting the working performance of the device were determined, and the overall structure of the full ridge imitation and the design of stem and leaf chopping knife roll were completed. Using the three factor five level quadratic regression orthogonal rotation center combination test method, the field test was carried out with the operation speed, knife roller speed and the height of knife roller from the ground as the test factors, and the qualified rate of broken length and potato belt rate as the evaluation indexes. The test data processing and parameter combination optimization were carried out with the software Design-Expert 8.0.6.1. The results showed that all factors had a significant impact on the qualified rate of broken length. The primary and secondary relationships from large to small were the rotation speed of knife roller, operation speed and the height of knife roller from the ground. Each factor had a significant impact on the potato carrying rate, and the primary and secondary relationships from large to small were the height of knife roller from the ground, the rotation speed of knife roller and the operation speed. When the rotating speed of the knife roller was 1450r/min, the operating speed was 3.5~6.7km/h and the distance from the knife roller to the ground was 285~317mm, the qualified rate of breaking length was more than 90% and the potato carrying rate was no more than 0.3%. The research results can provide design theory and feasible technical support for improving the operation quality and efficiency of potato stem and leaf chopping machine.

Abstract:In order to improve the navigation efficiency of the kiwifruit harvesting robot, the rapidly expanding random tree (Rapidly-exploring random trees, RRT) algorithm was researched, and an improved method (Straight-RRT) for real-time guided random tree expansion based on sampling state was proposed. Firstly, aiming at the problem of blind search in the traditional RRT algorithm, an evaluation index and a threshold were introduced to divide the sampling state, and the selection method of sampling nodes was determined according to the sampling state, so as to guide the expansion of the random tree in real time. Secondly, in order to enhance the adaptability of the algorithm to different environments, a dynamic threshold was introduced, which was adaptively adjusted according to the complexity of the environment, and at the same time, the nearest node selection mechanism was optimized to make the random tree avoid irregular obstacles faster. Finally, the path was optimized, the redundant points of the path were removed, and the Bezier curve was used to smooth the path to reduce the complexity of the path. Combined with the environment of the kiwifruit orchard, the effective harvesting area of the kiwifruit harvesting robot was analyzed, and a navigation method to achieve full coverage of the kiwifruit orchard was proposed. Based on the kiwifruit orchard environment, the path planning experiment was carried out, and the algorithm was compared with other improved algorithms. The path planning experiment results showed that the improved algorithm had better adaptability and planning efficiency in the kiwifruit orchard environment, which can provide a basis for improving the navigation efficiency of the kiwifruit harvesting robot.

Abstract:With the development of single seed sowing and precision seeding technology in China,higher requirements are put forward for the quality of single seed. In response to the current demand for fine seed selection in agricultural production,an online single seed detection and sorting device was designed. The device consisted of a feeding device, a detection unit, a sorting unit and a control system. The feeding device was composed of two sets of linear vibration devices, which can realize the flattening of the grains through two-stage vibration, and cooperate with the conveyor belt to complete the single granulation of the grains. The detection unit obtained the seed image in real time by the high-speed industrial camera, and transmitted it to the upper computer for detection and analysis. The sorting unit was made of sorting components and air compressor, which was used to remove the identified damaged or moldy grains. The control system controlled the sorting unit to complete the sorting according to the detection result and the position of the seed in the image. Furthermore, images of 3600 corn seeds (1200 normal seeds,1200 moldy seeds, and 1200 damaged seeds) were collected by using the built device, and image processing algorithms were used to obtain the 18 color and 12 morphological characteristics of a single seed, and the partial least squares discrimination analysis method (PLSDA) was used for discriminant analysis, and the detection models of moldy and damaged seeds were constructed respectively. Then the online verification experiment was carried out by using the built device and model. The results showed that the sorting rate of the device was greater than 300 seeds/min; the sorting accuracy of the mildew model was higher than 95%, and the sorting accuracy of the damaged model was higher than 89%. The device can realize the full automation of corn seeds from feeding to sorting, and can detect and sort moldy and damaged corn seeds in real time.

Abstract:Xinjiang Uygur Autonomous Region is the major region for high quality dried processing jujube fruit. The period of jujube harvesting was usually after the frost to enhance the quality of jujube fruit. At this time, a large number of mature jujube fruit were fallen off from trees naturally. The method of jujube fruit harvester was usually picked-up from the ground. The jujube harvesting machine of air suction type was the most commonly used one among the jujube harvesters. Its working principle was to suck up the jujube fruit through negative pressure airflow. But when sucking jujube fruit, the impurities materials such as jujube leaf, jujube crane were sucking attached. Thus the cleaning device was employed to remove the impurities materials. Traditional cleaning devices would cause some problems such as high impurity, damage, and lost jujube fruit. Inertial airflow cleaning system was designed that utilized the difference between the mechanical properties and fluid properties of jujube fruit and impurity. The cleaning system was mainly composed of the feeding inlet, guide surface, closed-air aspirator, baffle, filtering impurities device etc. Then the structure size of the cleaning system was designed by combining the principle of cleaning and the principle of conservation of fluid flow rate. In addition, the guide surface shape was analyzed through characteristics of material motion trajectory and numerical analysis. Furthermore, Fluent software was used to simulate the flow field track in the clearing system. This can assess the flow field characteristics intuitively and determine whether swirling flow occurred that available for separating impurities with expectations. To analyze the influence regularity of single test factors on the evaluation index, the factors of airflow velocity and baffle opening size values were selected as the test factors. Meanwhile, the impurity rate, loss rate, and damage rate were chosen as evaluation indexes. The single factor tests were employed to determine the parameter interval. The results showed that airflow velocity and baffle opening size values were 32~38m/s and -1~4cm, respectively. Moreover, the central combination experimental method of Design-Expert 10.0.3 software was used to analyze the influence of the interaction factors on the evaluation index. After that, the analysis module were used to determine the optimal parameter values. The results showed that the impurity rate, loss rate, and damage rate were 1.32%, 3.25%, and 0.57%, respectively, when the airflow velocity was 32.0m/s and the baffles opening size was 3.4cm. Furthermore, the field tests were carried out at the combination of optimal parameters to verify the accuracy of the optimization parameters and the performance of the cleaning system. The verified results showed that the impurity rate, loss rate, and damage rate were 1.38%, 3.37%, 0.60%, respectively, which were increased by 0.06 percentage points, 0.12 percentage points and 0.03 percentage points respectively compared with the optimized parameters. The cleaning system can meet the requirements of jujube mixed cleaning operation and the research result can provide reference for the development of air suction type jujube harvester.

Abstract:The tip leakage flow of axial-flow pump has an important influence on the internal and external characteristics of the pump. The relationship between the blade loading distribution patterns and tip leakage flow was established from the perspective of controlling the blade load. Based on the three-dimensional inverse problem design method, the impeller models of axial-flow pump with three typical blade loading distribution patterns of front load, middle load and after load at the blade tip were obtained. The influence of the blade loading distribution on tip leakage flow and induced leakage vortex flow of the axial-flow pump was studied based on three-dimensional turbulence simulation technique. It was found that compared with the front-loading type impeller and the middle-loading type impeller, the after-loading type impeller can effectively eliminate the low-pressure area near the blade inlet, which was beneficial to the cavitation performance of the impeller. The external performance simulation results showed that the pump performance was improved at the small flow rate and the hump phenomenon of the discharge-head characteristic curve was effectively suppressed, which was consistent with the experiment results. At the same time, the impeller with after load had better pressure pulsation performance under the condition of small flow rate.

Abstract:In order to reveal the external characteristics and pressure pulsation characteristics of various critical operating points in the forward full-feature partition of the tubular pump device, taking a bulb tubular pump device with specific speed of 1179 as the research object, the forward full-feature test of the pump device involving the critical operating point was conducted. The external characteristic parameters and pressure fluctuations of a total of 64 flow operating points were collected in the experiment, and the external characteristics and pressure pulsation characteristics of various critical operating points in the forward full characteristic partition of bulb tubular pump device were emphatically analyzed. The test results showed that the turning-off point was the dividing point between the countercurrent braking condition and the pump condition. The head of the pump device near the turning-off point was 6.41m, which was 3.27 times of the head of the design point. The axial power was 15.39kW, which was 2.67 times of the design point power. The dimensionless peak value of pressure fluctuation at the inlet of the impeller was 1.26, the middle of the impeller was 0.99, the outlet of the impeller was 0.84, and the outlet of the guide vane was 0.23, which were 2.3, 2.8, 4.9 and 23 times of the design point, respectively. The zero head point was the boundary point between the pump condition and the positive current braking condition. The flow rate near the zero head point of the pump device was 1.42Qd, which was 1.42 times of the design point flow. The axial power was 2.41kW, which was 0.42 times of the design point power. The dimensionless pressure fluctuation peak value of the impeller inlet was 0.21, the middle part of the impeller was 0.21, the impeller outlet was 0.15, and the guide vane outlet was 0.01, which were 0.38, 0.6, 0.88 and 1 times of the design point, respectively. The zero torque point was the boundary point between the positive flow braking condition and the turbine condition. The flow at the near zero torque point of the pump device was 1.63Qd, which was 1.63 times of the flow at the design point, and the head was -1.36m, which was -0.69 times of the head at the design point. The peak value of pressure fluctuation at the inlet of the impeller was 0.37, the middle of the impeller was 031, the outlet of the impeller was 0.20, and the outlet of the guide vane was 0.04, which were 0.67, 0.89, 1.18 and 4 times of the design point.

Abstract:Draft tube vortex is a sign of flow instability of Francis turbine, and even leads to unit fatigue failure in serious cases. In order to accurately capture the transient turbulence characteristics of fluid flow in draft tube under different working conditions, the newly developed Liutex vortex identification method was used to capture and compare the draft tube vortex rope based on slip grid technology and SST k-ω turbulence model. The influence of upstream disturbance on formation, development, rupture of vortex rope and low-frequency pressure pulsation of wall was analyzed emphatically. The results showed that the accuracy of result was verified through a comparison with literature and experimental results; The shape of vortex rope in draft tube was different under different upstream disturbances. In the best efficiency point, a stable spindle-shaped vortex structure was formed, which had little effect on the flow field. When the flow rate was reduced to 81% of the best efficiency point, the spiral vortex rope was formed, and the eccentric rotational motion of the vortex rope had a great disturbance effect on main flow. For example, due to the squeezing effect of vortex, an obvious highspeed zone appeared between the vortex structure and wall surface, and the amplitude of pressure coefficient was increased by 1.36~4 times. The pressure fluctuation in draft tube presented typical characteristics of low frequency and high amplitude. As the opening continued to be decreased, the volume of vortex rope was increased greatly, forming a large cavity vortex zone, and occupying a wide range and hit against the draft tube wall directly. With the decrease of the guide vane opening, the chaotic vortex and unstable flow field structure generated in draft tube were increased. When the opening dropped to the lowest, the original shape of vortex rope was damaged, and the broken little vortex filled the whole draft tube.

Abstract:Timely and accurate grasp of crop plant nitrogen content (PNC) information is helpful to monitor crop growth and realize the scientific management of farmland nitrogen fertilization. Based on this, taking unmanned aerial vehicle (UAV) as the platform to obtain digital images of potato budding, tuber formation, tuber growth, starch accumulation, and maturity period, and the PNC, plant height, and the three-dimensional coordinates of the ground control point (GCP) were measured. Secondly, the digital orthophoto map (DOM) and digital surface model (DSM) of the test area were generated by combining the digital images of UAV in each growth period with GCP. Then, the correlation analysis between the Hdsm and the constructed image variables of each growth period with the PNC measured on the ground were carried out, and the image variables with good correlation were selected as the input parameter of the potato PNC estimation models with the Hdsm. Finally, based on the image variables and image variables combined with Hdsm, three methods of multiple linear regression (MLR), error back propagation (BP) neural network, and Lasso regression were used to construct the PNC estimation models of potato at each growth stage. The results showed that the Hdsm extracted based on DSM had a high degree of fit with the measured H(R2 was 0.860, RMSE was 2.663cm, and NRMSE was 10.234%). Adding Hdsm in each growth period can improve the accuracy and stability of estimating potato PNC. The effect of PNC estimation model constructed by MLR method in each growth period was better than that of BP neural network and Lasso regression. Therefore, the research result can provide a technical reference for the efficient and non-destructive monitoring of potato PNC status.

Abstract:Aiming at the problem that the physical model has poor antinoise ability and is easy to overfit, a PROSAIL model was proposed by combining VMG (VARI (visible atmospherically resistant index), MGRVI (modified green red vegetation index) and GRRI (green red ratio index)) to retrieve the leaf area index (LAI) of winter wheat. The experiment was conducted based on unmanned aerial vehicles (UAV). Shanyang District in the southeast of Jiaozuo City, Henan Province was selected as the experimental area, and LAI data of winter wheat during two growth periods were measured. Firstly, an RGB vegetation index model was constructed, and the optimal VMG model was selected to invert LAI of winter wheat. Then, the sensitivity of PROSAIL parameters was analyzed to obtain the optimal parameter value and invert winter wheat LAI. Finally, the two models were combined using the very fast simulated annealing (VFSA) algorithm to obtain the optimal LAI of winter wheat. The results showed that VFSA can effectively combine PROSAIL model and VMG model to improve the inversion accuracy, and it was better than thta of VMG model and PROSAIL model. The coefficient of determination (R2) was higher than 0.8, and the root mean square error (RMSE) was lower than 0.4m2/m2. To sum up, the ground coverage was increased during the growth of winter wheat, and the method presented had strong radiative transmission mechanism, providing an effective inversion method for LAI inversion.

Abstract:The information of vertical distribution of chlorophyll content in different leaf positions of crops was obtained scientifically and efficiently to facilitate monitoring of crop growth conditions and field management. Based on the hyperspectral reflectance and chlorophyll content of different leaf positions of winter wheat obtained during the heading period, the correlation analysis of raw spectra, first-order differential spectra, second-order differential spectra, vegetation indices, continuous wavelet coefficients and chlorophyll content were performed to screen the spectral feature parameters with strong correlation. Then partial least squares regression, support vector machine, random forest and back propagation neural network algorithms were employed to construct chlorophyll content estimation models for the upper 1, upper 2, upper 3 and upper 4 leaves of winter wheat, and the best models for chlorophyll content estimation at different leaf positions were screened based on the accuracy assessment results. The results showed that the chlorophyll content estimation models constructed using wavelet coefficients combined with partial least squares were the most accurate for the upper 1, upper 2 and upper 3 leaves, with modeling and validation R2 of 0.82 and 0.75, 0.80 and 0.77, 0.71 and 0.62, respectively; the chlorophyll content estimation models constructed using vegetation indices combined with support vector machine were the best for the upper 4 leaves, with modeling and validation R2 of 0.74 and 0.79, respectively. The research result could provide theoretical and technical support for accurate monitoring of the vertical variation characteristics of crop nutrient content based on remote sensing technology.

Abstract:Taking Landsat remote sensing image data of 2008 and 2018 in Danjiang River Basin

Abstract:In order to facilitate the comparison of wheat model algorithms and multi-algorithm integrated simulation, a wheat model algorithm integration platform (WMAIP) was established by referring to the main module algorithms of domestic and foreign mainstream crop models (CERES-Wheat, APSIM-Wheat, WheatSM, WOFOST, SWAT, etc.). The method for simulation phenology integrated two algorithms with “wheat clock” model method and thermal time method based on WheatSM and APSIM-Wheat phenology module, respectively. The method for simulation biomass integrated three algorithms with carbon assimilation (CA), canopy photosynthesis (CP) and radiation use efficiency (RUE) based on WOFOST, APSIM-Wheat and WheatSM biomass modules, respectively. The method for simulation yield formation integrated three algorithms with harvest index (HI), grain filling (GF) and biomass remobilization (BR) based on SWAT, APSIM-Wheat and WheatSM grain yield modules, respectively. Six representative simulation models were constructed based on the model platform. The parameters of model were calibrated and verified by using field observation data of sowing date experiment from 2017 to 2019 in Wuqiao, Hebei Province and literature data of sowing date coupling irrigation experiment from 2011 to 2014. Finally, different algorithms for specific modules were compared. The results showed that simulated values of different algorithms in each module could be used to represent measured values with a reasonable error range. Therefore, the NRMSE values of phenology, above ground biomass, yield and soil water storage were ranged from 0.56% to 4.00%, from 16.13% to 18.72%, from 12.48% to 18.95% and from 10.78% to 11.63%, respectively. The effect simulated by multi-model platform was better than that of the single model. In the phenology module, the simulation of duration from sowing to jointing by thermal time method was better, while the simulation of durations from sowing to anthesis and from sowing to maturity were poorer than that by “wheat clock” model method. In the biomass module, the three algorithms were the best models to predict the biomass of wheat, but the simulated biomass of CP method was higher under high radiation conditions. In the yield formation module, the variation trend of yield simulated by the three algorithms was consistent, but the simulation results by BR method was rather better than that by others. In general, the WMAIP platform integrated multiple algorithms for specific modules to simulate soil water storage and biological indicators of winter wheat well. The application potential of this platform was great in comparison and improvement of wheat model algorithms, integrated simulation and assessment of climate change impacts.

Abstract:Early prediction for the growth and development of plants was an important component of the intelligent breeding process. However, it is difficult to accurately predict and simulate plant phenotypes. A prediction model of plant growth and development was proposed based on spatiotemporal long short-term memory (ST-LSTM) to predict future growth and development of plant. Firstly, the plant masks were recognized and extracted by the pre-trained Mask R-CNN model and the background of the plant image was removed by morphological operations. Then, the plant growth and development prediction data set was constructed. After that, utilizing the spatial and temporal dependence of plant growth and development, the image sequence of plants future growth and development was predicted by the prediction model for plant growth and development using the spatial and temporal depth characteristics integrated from the image sequence of early plant growth and development. The results showed that the image sequence predicted by the proposed model had high consistency and similarity with the actual image sequence of growth and development. At the first prediction time node, the structural similarity index measure was 0.8741, the mean square error was 17.10, and the peak signal to noise ratio was 30.83. The prediction determination coefficient (R2) of canopy leaf area, crown width, and leaf number were 0.9619, 0.9087 and 0.9158, respectively. Finally, the research realized the prediction of growth and development based on the image sequence of plant growth and development, which would effectively reduce the time, land and labor cost of repeated experiments in the field, and provided a reference for improving breeding efficiency.

Abstract:Automatic and accurate segmenting a single poplar leaf is very necessary for non-contact extraction of plant leaf phenotype. However, a single leaf segmentation is a challenging task, especially for the complexity of field poplar seedling planting environment. An automatic leaf segmentation method combined SegNet with 3D point cloud clustering was proposed. In the proposed approach, to obtain accurate sample images, the Kinect V2 camera was firstly calibrated. Subsequently, the RGB and depth data were aligned, the background was filtered, and the RGB and deep fusion data of poplar seedling were collected. Then, for RGB and deep fusion data, a large number of samples were labelled and SegNet was utilized to segment poplar seedling leaf and trunk candidate regions. Finally, in order to better segment single poplar leaves, 3D point cloud of leaf regions were reconstructed by using the RGB-D fusion data of poplar leaf regions separated by SegNet, and kd-tree based on geometric distance was introduced to classify single leaves. The performance of the proposed method was verified by various comparative experiments for poplar seedlings in different growth environments. SegNet and FCN were used to segment the leaf region and stem region of poplar seedlings respectively. The results showed that the precision of SegNet for leaf and stem detection were 94.4% and 97.5% respectively, and the intersection over union (IoU) were 75.9% and 67.9% respectively, which was better than that of FCN. In order to find the suitable segmentation threshold for a single poplar leaf segmentation, the comparison experiments of different threshold segmentation using kd-tree for single and multiple poplar seedling leaf areas were performed. The experiment results validated that the proposed method can segment poplar leaves not only for a single poplar seedling, but also for multiple poplar seedlings.

Abstract:In order to improve the precision of the detection and recognition of the potato seedling leaf bud and improve the efficiency of the automatic seedling production system, an improved recognition network based on the YOLO v4 network was proposed. The residual block in the feature extraction part CSPDarknet53 was replaced with Res2Net, and the depthwise separable convolution was used to reduce the computation. In this way, the receptive field of the convolutional neural network can be enlarged, the finer feature information of leaf bud can be got, and the missed detection rate of potato leaf bud can be reduced. Furthermore, a spatial feature pyramid (D-SPP module) based on dilated convolution was designed and embedded in the output of the three feature layers of the feature extraction part to improve the recognition and localization precision of potato leaf bud target. The ablation experiment was used to verify the effectiveness of the improved strategies. The experiment results showed that the recognition precision, the recall rate, the comprehensive evaluation index F1 value and the average precision of the improved network were 95.72%, 94.91%, 95% and 96.03% respectively. Comparing with the common networks such as Faster R-CNN, YOLO v3 and YOLO v4, the improved network had the better recognition performances, thus the production efficiency automatic seedling production system can be enhanced.

Abstract:In view of the existing field soil conductivity detection devices are mainly handheld, there are problems such as low detection efficiency and poor real-time performance. Based on the principle of current-voltage four-terminal method, a vehicle-mounted field soil conductivity on-line detection system was designed. The system mainly consisted of constant current signal source circuit, signal processing circuit, Arduino controller, GPS positioning module and vehicle-mounted sensor, which could detect field soil conductivity online. The performance of the system was verified by laboratory and field tests. The test results showed that the system had good stability, and the dynamic response time was about 540ms, and the maximum deviation of temperature drift caused by startup warm-up was 3.70%. The correlation coefficient R2 of the system detection accuracy was 0.7342 without considering the influence of temperature difference, and the detection accuracy R2 was 0.8645~0.9156 after eliminating the influence of temperature difference, which was higher than that of the commercial handheld conductivity detector with R2 of 0.6095. The influence of tractor vibration, sensor insertion depth, operation speed and soil solidity on the detection accuracy of the system was explored, the vibration state was relatively static, and the maximum error of detection data was 10.37%, and the error was mainly concentrated within the range of 0~10μS/cm. When the operating speed was no more than 5.0 km/h and the sensor insertion depth was greater than or equal to 10cm, the system could stably detect the field soil conductivity online, and the soil solidity of the test plot should not be too small to ensure that the sensor electrodewas in full contact with the soil. The system can provide technical support for the follow up research of real-time variable fertilization technology based on the online detection of soil conductivity.

Abstract:To address the problem of low recognition accuracy for identifying different apple leaf diseases, an apple leaf disease identification model was proposed based on snapshot ensemble. Firstly, the original dataset was augmented by various digital image processing methods. Then, an Inception-ResNet V2 was chosen as base model. The convolutional block attention module (CBAM) was introduced to enhance the feature extraction capability for apple leaf diseases. And focal loss was used to alleviate the imbalance of samples in each category. Finally, the model was integrated through snapshot ensemble to obtain the final identification model for different degrees of diseases on apple leaves. The image was input to the final model for identification. Compared with the original single Inception-ResNet V2, the recognition accuracy of the improved model was increased from 88.32% to 90.82%. Experimental results showed that the ensemble model had a high accuracy rate, which provided an idea and explored a approach for diseases of different degrees on apple leaves.

Abstract:Black spot is one of the fungal diseases of Korla pear. It is of great significance to realize early diagnosis of black spot disease before the symptoms are evident, as it can prevent the spread of the disease and reduce the economic loss. Hyperspectral imaging technology was combined with Stacking ensemble learning algorithm to construct early and rapid diagnosis model of Korla pear black spot. Hyperspectral images of healthy, incubation period, mildly diseased and severely diseased Korla pear were obtained, and the average spectra in the region of interest were extracted. After pretreated by standard normal variable transformation, the first derivative, second derivative and their combinations, principal component analysis was implemented to reduce the data dimension. Then, the Stacking ensemble learning prediction model for black spot disease was constructed with K-nearest neighbor method (KNN), least squares-support vector machine (LS-SVM) and random forest (RF) algorithm as the base learner and LS-SVM as the meta-learner. The results showed that with the deepening of the disease degree, the reflectance spectra showed a downward trend, significant difference was observed, which provided a theoretical basis for the establishment of classification models. The total classification accuracy of healthy and different disease degrees of Korla pear was 98.28%, and the classification accuracy for incubation period pear was 100%. Compared with the results using single classifier, the classification accuracy for all pear and incubation period pear was increased by 5.18 and 23.08 percentage points, respectively. The results showed that Stacking ensemble learning had strong feature learning ability, and its combination with hyperspectral imaging technology can realize the recognition of incubation period of black spot in Korla pear. The results can provide a method for the early diagnosis and real-time monitoring of black spot of Korla pear.

Abstract:Each apple is unique but can be classified into an “apple type” via features such as color, contour, texture, and other physical characteristics. Many apple growers classify apple types manually, often at great expense due to misclassification errors, low efficiency, inconsistent results, and high labor costs. Therefore, a real-time apple type detection and classification system is needed to prevent these complications, which typically happen in the period between sourcing and sales. To automate apple type classification, EBm-Net, an automatic identification and classification model was proposed based on a dual-branch structure network. The model fully extracted the contour, color, and texture characteristics of an apple’s surface by fusing channel attention and spatial attention mechanisms; this was done to further increase the feature difference between apple types by using a distance metric. The effectiveness of the EBm-Net apple type classification method was validated by analyzing its feature map and category probability statistics map. Experimental results showed that the classification accuracy of the EBm-Net model applied to Red Fuji, Jonagold, Qin Guan, Xiao Guoguang, Golden Crown, Granny Smith, and Gala apples was 96.25%, 96.25%, 100%, 92.50%, 98.75%, 100% and 93.75%, respectively; the overall classification accuracy of the seven apple types was as high as 96.78%. Therefore, it was feasible to use visual images combined with deep learning to classify and recognize apple type, which provided a method for real-time autonomous apple type classification.

Abstract:In order to solve the problems with manual assessment of individual sheep’s pain, which includes the requirement for a high level of human experience on the subject matter, a lack of pain recognition accuracy, and extended delay for the treatment for sheep, spatial transformer visual geometry group network (STVGGNet) was proposed as an improved model to the current mainstream deep learning model visual geometry group network (VGGNet). The STVGGNet algorithm introduced the spatial transformer networks and increased the area of analysis and in return improved the level of recognition of a sheep’s facial expression with regards of pain. Additional 887 images were added to the pre-existing dataset of sheep’s facial expression images. However, because the new image dataset remained low in quantity, the model also utilized ImageNet for transfer learning and fine-tuning classification between painful and non-painful sheep’s facial expressions. The experimental results showed that the best performance accuracy of STVGGNet in training stood at 99.95% with the best validation results upwards of 99.06% vs the VGGNet model which yielded 99.80% and 95.07% respectively. Therefore, with STVGGNet’s improved accuracy and strong robustness to classify pain within a sheep’s facial expression, it provided technical support for the intelligent development of sheep disease detection in animal husbandry.

Abstract:In aquaculture, dissolved oxygen is a key water quality factor to ensure the survival of aquaculture organisms. In order to ensure that there is sufficient dissolved oxygen in the water body, aquaculture plants generally adopt a regular oxygen production method. Although this ensures sufficient dissolved oxygen, it causes a large energy consumption. In response to this problem, a dissolved oxygen regulation method was proposed based on modeling prediction and relational rule database, which mainly included three parts. Firstly, an adaptive enhanced particle swarm optimization-extreme learning machine model (AdaBoost-PSO-ELM) was constructed to achieve accurate prediction of dissolved oxygen. Then, the curved surface fitting method was used to quantify the relationship between the initial concentration of dissolved oxygen, the aeration flow rate and the opening time of the aerator, and a relation rule database was built to provide a basis for controlling the aerator. Finally, based on the predicted value of dissolved oxygen and combined with current dissolved oxygen content, the computer monitoring platform called the relation rule database to reasonably control the opening time of the aerator. The dissolved oxygen prediction results showed that the MSE, MAE and RMSE of the AdaBoost-PSO-ELM model reached 0.0055mg2/L2, 0.0531mg/L and 0.0745mg/L, respectively. Compared with particle swarm optimization extreme learning machine (PSO-ELM), extreme learning machine (ELM), BP neural network (BPNN) and wavelet neural network (WNN), the prediction performance of AdaBoost-PSO-ELM was significantly improved. The results of aeration experiments showed that the priori equation based on cubic polynomial can accurately quantify the nonlinear relationship between the initial concentration of dissolved oxygen, the aeration flow rate and the opening time of the aerator, and the R2 of fitting was above 0.99. At the same time, the rule database constructed based on the quantitative results can reasonably control the opening time of the aerator, which was of great significance for saving energy and promoting sustainable aquaculture, and it had great application prospects in the future.

Abstract:Aiming to solve the problems of centralized data storage, easy data tampering, and data trust in the existing agricultural product quality traceability system, as well as to ensure the quality and safety of agricultural products, protect the rights and interests of consumers and improve the brand competitiveness of production enterprises. Based on the analysis of the agricultural product industry chain business process and the key technologies of the blockchain, the trusted traceability block structure of agricultural products was designed to ensure that the traceability data of agricultural products were unforgeable, safe and reliable; the “On-Chain+Off-Chain” agricultural product was proposed. The collaborative management and storage strategy of quality and safe traceability information solved the problems of high data storage pressure, low query efficiency and data explosion of each node in the agricultural product traceability blockchain network. Kafka consensus mechanism was used to achieve consensus operations with multi-agent participation and provide real time data with high throughput, high-volume and low-latency processing capabilities; and agricultural product traceability smart contract rule sets and contract trigger conditions were developed to ensure the reliability of agricultural product data and the credibility of the traceability platform. A trusted traceability system was developed for agricultural product quality and safety based on the Hyperledger Fabric blockchain platform, and the verification and analysis were conducted on the traceability of the quality and safety of black tea products. The results showed that the developed agricultural product quality and safety credible traceability system can solve the problems of data security and authenticity of traceability information among multiple nodes in the production, processing and circulation of the agricultural product industry chain, which achieved good application results. 

Abstract:With the global warming, droughts occurred more frequently than before. Droughts have occurred in all major farming areas in China, and spread all over the country. The wheat production in the western Loess Plateau is obviously affected by drought, flood and water conditions, so it is necessary to study its response characteristics under the background of meteorology and agricultural droughts in order to put forward effective measures to prevent agricultural production from being negatively affected by drought. The DSSAT-CERES-Wheat model was combined to simulate spring wheat at seven sites in the western Loess Plateau. The growth factors and yield data from 1961 to 2018 were collected and their temporal and spatial changes were analyzed. The standardized precipitation evapotranspiration index (SPEI) and soil moisture deficit index (SMDI) at the 0~10cm depth and 10~40cm depth at the time scales of 1~6 months were estimated, and the temporal and spatial changes of meteorological and agricultural drought were explored. The effects of drought severity on the growth process and yield of spring wheat were further studied. The results showed that taking Linxia Station in Gansu as an example, the dry and wet status of SPEI and SMDI at the time scale of 1~6 months were generally consistent, SPEI generally showed alternatively wet and dry conditions, and the changes in SMDI at the 0~10cm depth and SMDI at the 10~40cm depth were basically the same, showing a wetter trend. The DSSAT-CERES-Wheat model was effective in simulating the key growth period and yield of spring wheat in the western Loess Plateau (0.65≤R2≤0.84). The maximum leaf area index of spring wheat from 1961 to 2018 and the above-ground biomass had no obvious change trends, but the yield had an increasing trend after 2005. Drought in the jointing and filling stages had greater impacts on the growth process and yield of spring wheat. The relationship between SMDI and spring wheat growth and yield elements was closer than SPEI, indicating that agricultural drought had a greater impact on winter wheat growth and yield. The 2 month SMDI at the 0~10cm depth was a key time scale for showing the effects of drought on spring wheat growth and yields. The research result provided a useful reference for the prevention of meteorological and agricultural drought in spring wheat production on the western Loess Plateau.

Abstract:To explore the impact of nitrogen application rates on CO2 sequestration and emissions in peanut field under regulated deficit irrigation in Liaoning Province, a split plot experiment was conducted in lysimeter in 2018 and 2019 to study the effects of different irrigation regimes (full irrigation during the whole growth period (F) and regulated deficit irrigation at the flowering and pod filling stages (F)) and nitrogen application rates (0kg/hm2(N0)、50kg/hm2(N50)、100kg/hm2(N100)、150kg/hm2(N150)) on dry matter accumulation, carbon sequestration and peanut yield an so on in farmland. The results showed that the dry matter accumulation, carbon sequestration, and peanut yield under the D treatment were 7.59%, 15.08% and 7.16% (two years average) higher than that under the F treatment, respectively. Under the two irrigation regimes, the dry matter accumulation, carbon sequestration and peanut yield were increased first and then decreased with the increase of nitrogen application rates, and reached the maximum value at the 100kg/hm2 nitrogen level. From the seedling to the pod filling stage, CO2 emissions in peanut field was increased first and then decreased, and reached the maximum value at the flowering stage. Compared with the F treatment, D treatment significantly reduced the average CO2 emissions in the soil at the flowering, pod setting, and pod filling stages, and decreased the cumulative CO2 emissions during the entire growth period. Under the two irrigation regimes, soil CO2 emissions were increased with the increase of nitrogen application rates. Under the same nitrogen application level, D treatment significantly reduced the cumulative CO2 emissions during the whole growth period compared with F treatment. The cumulative CO2 emissions in the DN100 treatment were 7.51% (two years average) lower than that of the FN100 treatment. Under different irrigation and nitrogen treatments, the DN100 treatment had the largest carbon sequestration and peanut yield, and relatively lower CO2 emissions, which was the best combination for carbon sequestration and CO2 emission reduction in peanut farmland ecosystem.

Abstract:To realize simulation of the two-layer progressive causal relationship of soil salt-water and crop production benefits under the influence of multiple factors, based on deep learning theory and technology, the progressive salt-water embedding neural network (PSWE) model was constructed. In PSWE model, the time serialized data encoder framed by hierarchical long short-term memory (HLSTM) and decoder framed batch-normalized multi-layer perceptron (BMLP) were coupled, and the coupling between Dropout and Adam algorithm was optimized as an improved algorithm for convergence regression. The validity of PSWE model was evaluated, and the dynamic changes of soil water-salt of different irrigation amounts under multi-factors cooperative straw deep burial were simulated, and the production benefit of summer maize was predicted. The results showed that PSWE model had multivariable overall synergy, self-learning habit and high accuracy. PSWE model could effectively describe the law of soil salt-water migration under straw deep burial in Hetao Irrigation District, the internal dependence relationship between summer maize production benefit and various variables. The root mean square error of the PSWE model was 0.031, the mean absolute error was 0.569, and the determination coefficient was 0.987. Through the model simulation, along with the summer maize growth period, the moisture content of treatment of single irrigation 60mm was reduced continuously in the tillage layer (0~40cm), and affected the summer maize for normal growth, while the change of treatment of 135mm was larger. In the mature stage, they produced salt accumulation in the straw inter-layer, and the salt accumulation rate was 49.2% and 11.2%. The water content in the tillage layer of single irrigation 90mm and 120mm were kept between 16% and 24%. At the end of the growth period, the water content in the soil layer over 40cm was kept stable. The straw inter-layer showed a trend of desalting, and the desalting rate was 6.1% and 5.9%, respectively. It was suggested that the single irrigation amount should be 89.3~96.8mm,and the theoretical salt content of cultivated layer was 1.38~1.55g/kg. In conclusion, under multi-factors cooperative straw deep burial, appropriate irrigation amount could achieve the goal of salt suppression effect and improvement of water use efficiency. The PSWE model could effectively simulate soil salt-water migration. The simulation of soil water-salt migration and crop productivity benefit by PSWE model was applicable, which provided a reference for deep learning theory and technology in soil salt-water migration.

Abstract:The key to effectively control salinization is to deeply understand the spatial-temporal distribution of water and salt at the regional scale after large-scale water-saving and salt control in arid oasis. Yanqi Basin in northwest Xinjiang of China was selected as the research area. LandSat remote sensing and field sampling water and salt data were used to analyze the water and salt migration and accumulation process after large-scale water-saving under changing environment. A total of five indicators, including surface salinity, soil salinity, groundwater salinity, surface irrigation, and groundwater depth were selected in 2000, 2005, 2015 and 2020. The weights of each indicator factor were determined by extension analytic hierarchy process. With the help of ENVI and ArcGIS software, spatial distribution raster maps of each indicator factor were obtained, and each raster map was standardized, the space was nested and superimposed according to the weight of each index factor, so as to get the spatial-temporal distribution of water and salt in the regional scale. The results showed that the salinization in the study area experienced three stages: expansion in 2000—2005, shrinkage in 2005—2015 and stabilization in 2015—2020. The areas with serious salinization were mainly along Bostan Lake, especially Bohu County in the west and Heshuo County in the north of Bostan Lake. Overall, the area of mild saline land was the largest, accounting for 6.17%~11.39%, followed by moderate saline land, accounting for 3.08%~10.65%, the area of heavy saline land was the smallest, accounting for 0.56%~1.10% during 2000—2020; the weight of each factor affecting the spatial-temporal differentiation process of water and salt on a regional scale was ranked as follows: groundwater depth (0.325), groundwater salinity (0.282), soil salinity (0.198), surface salinity (0.184), and surface irrigation (0.031). Groundwater depth and groundwater salinity were the main driving factors affecting the spatial-temporal differentiation of water and salt at regional scales. From the interpreted salinization development trend, the salinization in the study area was alleviated and tended to be stable. 

Abstract:Nutrient solution configuration and irrigation control technology have become the key technology in soilless culture. However, the current nutrient solution management control has problems such as cumbersome configuration process, inaccurate nutrient solution ratio and single irrigation strategy, which hinder the development of soilless culture technology. Aiming at the above-mentioned various problems, nutrient solution management and control system was designed based on the Mariotte device. The system was mainly composed of nutrient solution preparation system and nutrient solution irrigation control system. Through EC (conductivity) value and pH value sensors, the components of the nutrient solution were obtained in real time, the mathematical model of liquid preparation and prescription was established, and the target nutrient solution was automatically and quickly configured according to the nutritional needs of different crops. According to the needs of different users, combined with the meteorological parameters and application environment of the crop cultivation area, three different irrigation strategies were formulated. Experiments showed that the system can quickly and accurately realize the management and control of the nutrient solution, laying a part of the foundation for the production of soilless cultivation of crops. 

Abstract:The methods of extracting lipophilic proteins from different sources were introduced, and the structures and properties of these lipophilic proteins were compared. By means of SDS-PAGE, solubility, surface hydrophobicity, thermogravimetric analysis, and determination of protein secondary and tertiary structures, the composition of the extracted protein was analyzed from structural characteristics and physicochemical properties to determine whether it was lipophilic proteins, and the similarities and differences between these proteins. The contents of protein and lipid in the four extracts were 84%~86% and 13%~13.5%, respectively. The results of SDS-PAGE, Tunku Abdul Rahman spectra and infrared spectra showed that the structure of soybean lipophilic proteins, black bean lipophilic proteins and mung bean lipophilic proteins were similar, but pea lipophilic proteins were slightly different, the main difference was that the secondary structure of pea protein was less stable than that of other lipophilic proteins. The results of differential scanning calorimetry (DSC) showed that the four protein extracts were still mixture, and the denaturation temperature was ranged from 60℃ to 74℃, the results also proved that all the four proteins were lipotropic components. The extraction of these protein components not only broadened the source of lipophilic protein, but also provided a theoretical basis for the analysis and modification of protein components of four legumes, and it can provide a way for the processing and application of four legumes.

Abstract:Aiming at the problems of low cleaning rate of eggshell membranes, high energy loss and complex structure of device for collecting the eggshell membranes and eggshells in the current research, multi-aspirator device was designed. The influence of particle collision on the movement of eggshells and eggshell membranes particles was analyzed. Under the condition of different number of inlet baffles in the cleaning chamber, the motion trajectories of eggshells and eggshell membranes and the characteristics of flow field inside the multi-aspirator device were analyzed by using the method of CFD-DEM coupling. The analysis results showed that with the increase of air inlet baffles’ numbers, the airflow velocity of cross flow area in the cleaning chamber was increased, and the loss rate of eggshell membranes was decreased. There was no vortex in the outlet of the cleaning chamber, so as that the cleaning rate of eggshell membranes decreases caused by airflow hindering the falling eggshells was avoided. The orthogonal test method of two factors and three levels was applied, feeding flow rate and airflow velocity at the interface of the sucking fan were taken as test factors, and the loss rate and the cleaning rate of eggshell membranes were taken as evaluation indexes. The parameters of influencing factors were optimized and verified by tests. The results of tests were that when the feeding flow rate was 200g/s, the airflow velocity at the interface of the sucking fan was 5.5m/s, the loss rate of eggshell membranes was 9.4%, the cleaning rate of eggshell membranes was 96.3%, and the power of sucking fan was 330W. The results could provide foundations for the application of multi-aspirator device for collecting eggshells and eggshell membranes.

Abstract:Soybean protein isolate (SPI), gum Arabic (GA) and carrageenan (CA) were used as raw materials to prepare different polysaccharide and SPI complexes by physical mixture. The effects of different amounts (the mass ratios of SPI and GA/CA were 20, 15, 10 and 5) on the structure, properties and emulsion stability of protein-polysaccharide complexes were investigated. Finally, the interaction mechanism between protein and different polysaccharide complexes was confirmed. The structural characteristics of different complexes were analyzed by infrared spectrum, fluorescence spectrum and electron microscope. The physical and chemical properties of different complexes were determined by emulsifying property, particle size, Zeta-potential and surface hydrophobicity, and their stability properties were clarified by discussing the emulsifying activity, emulsifying stability, apparent viscosity and emulsifying index of different complex emulsions. The results showed that the two polysaccharides could form complexes with SPI under acidic conditions, and when the mass ratio of SPI to CA was 20, the highest Zeta-potential of the complex was (20.47±0.82)mV and the minimum average particle size was (1.37±0.01)μm, the emulsifying activity index was (106.46±4.75)m2/g, and the emulsifying stability index was (145.33±8.53)min, the stability of the composite emulsion was better under the condition. The addition of CA significantly reduced the endogenous fluorescence intensity of SPI and changed the secondary structure of SPI. SPI combined with CA to form a stable complex, which provided some theoretical support for the application of protein-polysaccharide composite emulsion in the transport of active substances.

Abstract:In order to improve the accuracy and automation level of the air conditioning and fresh-keeping process for fresh agricultural products, a set of remote monitoring and gases distribution for fresh-keeping system was designed, which realized the control of raw gas flow, concentration and mixing ratio based on embedded PID control algorithm and mass flow controller. 4G network was used as the bridge of remote data transmission, 4G DTU was selected as the network data transceiver, and the remote monitoring function of mobile clients was realized by application on the phone. The gas parameter sensing layer, data network transmission layer and control operation application layer were set up in the system structure. Based on the requirements of remote monitoring of gas distribution system, Modbus RTU program was embedded in PLC to ensure the communication between DTU and PLC. TCP protocol was used as a bridge in order to enable data transmission between it and the cloud server. At the same time, remote monitoring and system maintenance on mobile client was realized based on application software in the phone. According to the test of system stability, accuracy and communication performance, the error of distribution volume fraction was 0.22% when the distribution was stable, the average error was reduced by 91.67% compared with the traditional method. Moreover, the gas distribution speed was increased by about 50%, which greatly improved the working conditions and production efficiency of the gas distribution site. It provided technical support for further perfecting the automatic gas distribution production line of fresh agricultural products.

Abstract:Aiming at the current attitude-adjustable hilly tractors can only achieve static leveling, height difference leveling, and low leveling accuracy, an attitude adjustment device of the twisting and swinging wheel hilly tractor was designed. This device realized the tractors adaptation to complex roads by adjusting the relative rotation of the front and rear vehicle bodies. Firstly, according to the requirements of the special pavement in hilly areas, the stability of tractors on slopes was studied, and a twisting and swinging attitude adjustment device was designed. Then, dynamic simulation of twisting and swinging attitude adjustment device was carried out, model of the tractor was established, and dynamic simulation analysis of multiple working conditions was performed. The dynamic simulation showed that the device maximum rotation angle was 15.2°, the maximum inclination angle for the tractor to maintain stability on longitudinal slopes was 23.2°, and the maximum inclination angle for stable driving on transverse slopes was 16.8°. Finally, field trials of prototypes were conducted. The test results showed that the average angle of rotation of the twisting and swinging attitude adjustment device was 15.03°, the maximum longitudinal climbing angle of the tractor was 25.6°, and the maximum transverse climbing angle was 16.2°. The average productivity of rotary farming was 0.65hm2/h, and the average productivity of plow farming was 0.36hm2/h on the ground with a slope of 15°. The tractor met the design and technical requirements, and can better adapt to the hilly and mountainous environment, and it can meet the normal operation requirements of the hilly and mountainous areas.

Abstract:Aiming at the problem that the driver's sitting posture changes due to the cockpit tilt during the tractor ploughing operation on high and low ridges, which affects the driver's ride comfort, a tractor seat leveling control system based on the driver's comfort was designed, a seat leveling control strategy based on the driver's ride comfort was proposed, and a seat leveling system controller was developed to feed back the seat tilt state in real time according to the seat angle sensor. The automatic leveling of the seat was realized. In order to better improve the driver's riding comfort in the leveling process, a test-bed simulating the tilt state of tractor seat was built, and the influence of seat tilt on the driver's upper torso posture under different angles in the test-bed was studied. The test results showed that the projection offset of human thoracic spine and lumbar spine was increased with the increase of seat tilt angle, but the projection offset when the seat was tilted at 3° was far less than that when the seat was tilted at 5°, which determined the operating threshold of the seat leveling system. Through the subjective evaluation test, it was determined that the driver's comfort was higher when the speed of seat leveling was 6~8mm/s. Finally, the real vehicle test was carried out on Dongfanghong LX754 tractor. The comparison test of driver's back muscle electromyography evaluation showed that the lumbar force was uniform compared with that without adjustment, which can well protect the driver's lumbar muscle. The tractor operation accuracy test and the tractor straight-line driving path tracking test showed that the use of the seat leveling system improved the operation efficiency of the steering wheel and gear. After the seat leveling, the average error of operating the tractor straight-line was reduced by 3% and the maximum transverse error was reduced by 8%, which verified the practical value of the system. This research can improve the ride comfort of drivers.

Abstract:Aiming at the problem that the current manual feature detection of assembly robots was susceptible to interference factors such as illumination conditions, background and occlusion, and the feature detection based on point cloud depends on the accuracy of model construction, the method of deep learning was proposed to carry out research on the visual positioning technology of the workpiece based on key point prediction. Firstly, the ArUco pose detection marker and ICP point cloud registration technology were used to construct a set of data for training the pose estimation network model. The depth images from various angles of the workpiece were collected, and the pose information of the workpiece was calculated. The key points on the workpiece surface were selected as the data set. Then the vector field of the key points on the workpiece surface was constructed, and the depth training was carried out to gather with the data set to realize the vector field prediction of the foreground points pointing to the key points. And the direction vectors of each pixel in the vector field pointing to the same key point were divided into two groups, the intersection points of their vectors were taken to generate the hypothesis of the key point, and all the hypotheses were evaluated based on RANSAC voting. The EPnP solver was used to calculate the pose of the workpiece, and the orientation bounding box of the workpiece was generated to display the pose estimation results. Finally, the accuracy and robustness of the estimation results were verified by experiments.

Abstract:The orifice of the hydraulic valve was used to control the flow in the traditional valve controlled hydraulic system, which inevitably led to very large throttling loss. Based on the idea of the digital hydraulic and inspired by the theory that there was no throttling loss in the fully open and fully closed state of the high-speed on-off valve, an innovative configuration of the two-dimensional pulse width modulation rotary valve was proposed to control and distribute the flow of hydraulic system in the way of fluid pulse width modulation, in order to reduce the throttling loss and greatly eliminate the overflow loss through the active overflow. With the rotation of the valve core, the oil could be quickly switched between the high-pressure (the load) branch and the low-pressure (the oil tank) branch, so as to output the discrete flow. And the duty cycle (the ratio of the communication time of the load branch to the total communication time of the oil return branch at a constant speed) was controlled by the axial displacement of the valve core to realize the control of the output average flow. Through the establishment of the mathematical model, the simulation and the experimental research, it was verified that the high frequency two-dimensional pulse width modulation rotary valve can transform the continuous flow of the fluid into discrete and controllable flow. When the output flow of the quantitative pump was 30L/min, the output frequency of the two-dimensional pulse width modulation rotary valve was 200Hz and the control pressure was 13MPa, the output pressure of the two-dimensional rotary valve could be maintained at 21MPa. Taking the duty cycle of 50% as an example, the load flow was about 12.3L/min. This was a flow control method from the perspective of the discretization of the working medium in the fluid system.