Abstract:Aiming at the problems of poor terrain adaptability and difficult control of stubble height of self-propelled green fodder harvester in complex environment, a header profiling system suitable for self-propelled green fodder harvester was designed by using two-point detection-electro-hydraulic control, and the relevant experimental research of profiling system was carried out. On the basis of explaining the overall architecture and specific working principle of the system, the structural parameters of the main key components such as the profiling detection mechanism and the lateral profiling adjustment mechanism were determined through theoretical calculation. The static stress analysis model was established to obtain the relevant mechanical characteristics of the connection between the header and the feeding box. ADAMS simulation software was used to create a cylinder load characteristic model, complete the design and related optimization of the profiling system, and determine the optimal operating parameter range of the cylinder. In order to verify the function of the profiling system, the system was mounted on the 4QZ-30 self-propelled green fodder harvester for testing, and the ratio of the time of the cutting knife front end from the ground height within 100~150mm to the total test time during the test process was used as the test index, the road simulation test and field profiling harvesting tests were arranged, and the control system was used to obtain the header height and response time in real time, and the results showed that the profiling detection mechanism detected the height information reliably. The linear fitting R2 value was 0.9987; the average value of the response time t of the profiling control system was within 0.16s; the profiling system can perform profiling work on the ground with a slope of 0°~6° at driving speed of 0~6km/h, and during the road simulation test, the ratio β of the time when the height of the cutting knife was within the standard range to the total test time was 90.76%, the pass rates of field profiling harvesting tests were 86.67%, 86.67% and 93.33%,which improved the terrain adaptability of the self-propelled green fodder harvester and reduced the difficulty of controlling the stubble height. It can provide a reference for the profiling technology of selfpropelled green fodder harvester.

Abstract:Aiming at the shortage of potato combine harvesters and the poor passability of caterpillar chassis in hilly and mountainous areas, a self-propelled potato harvester with chassis walking device and multistage conveying and separating device was designed, and the passability of chassis and harvestability were analyzed theoretically. Firstly, the performance of harvester chassis driving on slope and obstacle crossing was analyzed theoretically, and the critical parameters of chassis passing were obtained. Secondly, the kinematic analysis of potato during harvest was carried out, and the relevant parameters of key working parts were obtained. At the same time, RecurDyn simulation software was used to simulate the multi-body dynamics of the whole machine, and the relevant motion parameters of the self-propelled potato combine suitable for the horizontal and longitudinal slopes in hilly areas, across trenches and straight walls were obtained. The simulation results showed that the maximum climbing angle of longitudinal slope driving was 28°, the maximum slope angle of transverse slope driving was 20°, the maximum height of the machine crossing vertical obstacles was 150mm, and the maximum crossing trench width was 300mm. The results showed that the rate of damaged potato and the rate of broken potato were 1.92% and 2.86%, respectively, which met the standard of damaged potato and broken potato less than 2% and 3%, respectively. The harvester can meet the longitudinal slope of 25° and drive steadily, cross the trench of 300mm width and climb over the straight wall of 150mm, which was consistent with the simulation results, and verified the accuracy of the simulation. This study can meet the design requirements of crawler potato harvester driving performance, and provide theoretical basis and reference for the design of crawler root harvester in hilly and mountainous areas.

Abstract:In order to solve the problems of high labor intensity, low efficiency, and high cost of manual picking after segmented harvesting of potatoes, a self-propelled potato picking device was designed. The picking device with a double layer reverse chain clamping and conveying function was designed to address the issue of soil accumulation in the feeding part of the picking device, which caused poor potato conveying and affected the overall operation efficiency. To determine the optimal operating parameters of the picking device, a Box-Benhnken test method was used based on the coupling simulation of discrete element software EDEM and multi body dynamics software RecurDyn. The potato flow rate and potato damage rate were used as experimental indicators, and the forward speed of the picking device, the depth of the picking shovel into the soil, the conveying chain speed of the picking device, and the reverse clamping chain speed were used as experimental factors. A four factor and three level experimental was conducted on the working parameters of the device, Design-Expert software was used to establish a quadratic polynomial regression model. After optimizing the regression model, a response surface curve was drawn to determine the optimal operating parameters of the device. Field experiments showed that when the forward speed of the picking device was 0.70m/s, the depth of the picking shovel into the soil was 120mm, the conveying chain speed of the picking device was 1.20m/s, and the reverse chain speed was 1.20m/s, the potato flow rate was 5.94kg/s, and the potato damage rate was 2.10%. Compared with the simulated theoretical values, the errors were 3.30% and 4.48%, respectively. The research result can provide reference for the design of potato picking devices.

Abstract:Aiming at the problems of low mechanical efficiency and high peeling damage rate of peeling equipment in the harvesting process of fresh corn in China, based on the analysis of the structural characteristics of the existing peeling device, a flexible peeling device with the combination of “flexible segmented roller type + spiral adjusting frame” and rubber frequency vibration plate matching was designed. According to the physical characteristics of fresh corn, the peeling process of fresh corn was analyzed mechanically and kinematically, the main factors affecting the peeling performance were determined, and the structural design and parameter analysis of the peeling device was carried out. ANSYS Workbench/LS-DYDA module was used to simulate the peeling process of fresh corn ear, and a peeling prototype was fabricated according to the theoretical analysis and simulation results, and the peeling test was carried out. In order to obtain the best test material for the prototype, a single-factor test was conducted with cob length, ear diameter and ear water content as test factors, and ears with length of 260~280mm, diameter of 64~66mm, and water content of 66.5%~69% were determined as samples for the orthogonal test material of the peeling machine. Design-Expert software was used to design a three-factor, three-level orthogonal test, with peeling roller speed, peeling roller inclination angle and vibrating plate vibration frequency as test factors, and bract peeling rate and grain breaking rate as test indicators. The results showed that the impact on bract peeling rate and grain breaking rate from large to small were peeling roller speed, peeling roller inclination angle, frequency vibration plate; the optimal combination of parameters: peeling roller speed of 478.72r/min, peeling roller inclination angle of 8.05°, vibrating plate vibration frequency of 259.20 times/min, at this time, the bract peeling rate was 92.41%, and the grain breaking rate was 1.47%. Under this condition, the verification test was carried out according to the actual working conditions, and the bract peeling rate and grain breaking rate were 91.75% and 1.55%, respectively, which were basically consistent with the parameter optimization and met the requirements of fresh corn peeling. The research result can provide technical support for the optimal design and selection of fresh corn peeling equipment. 

Abstract:In view of the problems such as high miss rate, high damage rate, high impurity content, low working efficiency, insufficient walking power, poor trafficability and inconvenient collection of the current Hangzhou white chrysanthemum harvesting machinery, a comb belt self-propelled harvester was developed based on the planting mode and picking requirements of Hangzhou white chrysanthemum. The harvester was mainly composed of harvesting components, brush components, lifting devices, walking devices, collection devices, and hydraulic systems. It utilized the comb action of multiple rows of comb teeth to achieve continuous harvesting of Hangzhou white chrysanthemum. The operating speed and working height of the harvesting components can be adjusted through the hydraulic system. Through theoretical calculations, the structural parameters of key components such as conveyor belts and comb teeth were determined. A prototype experiment was built with the driving speed of the harvester, the working speed of the picking components, and the minimum distance between the comb and brush as experimental factors. The picking rate, damage rate, and impurity content of Hangzhou white chrysanthemum were used as the experimental indicators. A three factors and three levels orthogonal experiment was conducted to obtain the optimal parameter combination of the harvester. The experimental results showed that when the driving speed was 0.1m/s, the working speed of the picking components was 60r/min, the minimum distance between the comb and brush was 60mm, the picking effect was the best, with picking rate of 83.1%, damage rate of 15.8%, and impurity content rate of 17.9%. This comb belt self-propelled harvester had stable operation, which met the agronomic requirements of Hangzhou white chrysanthemum picking.

Abstract:The full-feed tangential flow type peanut picking operation as the main way of peanut mechanization harvest has problems on short effective time and high loss rate of pod-picking. In order to solve this problem, a multistage tangential flow peanut picker and conveying device was developed which mainly composed of multistage pod-picking cylinder, front conveying plate, vibration drive shaft and rear conveying plate. The conveying device and the pod-picking device of traditional peanut combine harvester were integrated. The mechanism of seven-stage roller in series combined with vibration conveying was proposed to realize synchronous operation of pod-picking and conveying. The structure and parameters of the key components were designed on the basis of the operating principle analysis. The simulation and optimization method of discrete element software EDEM was used to analyze the motion parameters (direction angle, amplitude, frequency) of the conveying board. Taking the typical peanut variety “Dabaisha” in the main producing area as the research object, the simulation results were verified by field experiments in Zhumadian City, Henan Province. The results showed that when the feeding amount of peanut plants, the rotation speed of the second pod-picking cylinder, the rotation speed of the other six pod-picking cylinders, the direction angle of the conveying plate, the amplitude and the frequency were set to be 5.6kg/s, 325r/min, 239r/min, 25°, 45mm and 7Hz respectively, the picking rate, breakage rate and enrolling loss rate of peanut were 98.41%, 4.76% and 1.46%, respectively. The research results were helpful for the development and promotion of mechanized peanut harvesting technology.

Abstract:Lotus root is an important aquatic vegetable and special agricultural product that is easy to grow and difficult to harvest, with good nutritional value and economic benefits. The lotus root is commonly planted in the Yellow River Delta region at a depth of 30~40cm, but it is mainly harvested manually, which is a harsh environment, labor-intensive, and requires long hours in cold water, which is extremely harmful to the lotus root farmers' health. To solve the low efficiency and high labor intensity of lotus root manual harvesting, an intelligent, effective, and low-damage crawler selfpropelled hydraulic lotus root harvester was designed, which can improve the mechanization and efficiency of lotus root harvesting, and promote the large-scale development of lotus root industry. The harvester was mainly composed of power system, water circuit, hydraulic system, and control system. The total power of the machine was provided by the diesel engine, the moving mechanism was crawler type, which had good stability and flexible steering performance and can adapt to the complex root field operating environment. The mode of flushing slit was swinging jet, which was driven by the lifting hydraulic cylinder and oscillating hydraulic cylinder to ascend and descend the nozzle array and oscillate in a left-right cycle respectively, and was capable of 0°~40°slope transit operation, with an operating width of 2.3m. It can quickly and effectively wash the silt above the surface of the lotus root. Harvesting experiments were conducted in three different lotus root fields. The results showed that the lotus root harvesting depth was 40~50cm, the harvesting ratio was no less than 95%, and the lotus root damage ratio was no more than 5%, and the fuel consumption was no more than 215g/(kW·h). In ordinary fields, speed of the harvester was 3m/min and the average working efficiency was 0.04hm2/h, which was 4~5 times of manual harvesting. In the experimental process, the lotus root harvester had stable working performance, which met the requirements of lotus root harvesting.

Abstract:Aiming at the problems of the existing sugarcane cutting and paving machine, such as one-way harvesting operation, fixed cutting height, no sugarcane supporting mechanism, unreasonable frame structure and low intelligence, a modular sugarcane cutting and paving machine that can be remotely controlled in hilly and mountainous areas was designed. Through the reasonable layout and sidehanging conveying form of the whole machine, the sugarcane was laid to the left and right sides behind the cutter by the laying device with adjustable laying angle, so as to realize the two-way harvesting operation and improve the problem of easy blockage of the conveying channel. Combined with the force analysis of sugarcane in the process of sugarcane movement, the design of unequal pitch spiral drum was put forward. The spiral line equation, the diameter and installation angle of the spiral drum were obtained by the change of spatial coordinates. Through the analysis of sugarcane conveying motion, the working speed of the conveying and laying mechanism and the laying angle of sugarcane were determined. Through the analysis of sugarcane cutting mechanism, the key parameters such as cutting form, cutter diameter and rotation speed of sugarcane cutting mechanism were obtained. The working width of the whole machine was designed to be 1100mm, the working speed was 1.8km/h, and the production efficiency was 0.176hm2/h. The field test results of the prototype showed that when the forward speed was 452.28mm/s, the rotation speed of the sugarcane cutting mechanism was 562.12r/min, and the inclination angle of the cutter head was 12.27°, the orthogonal test response surface analysis was used to obtain the minimum head breaking rate of the sugarcane cutting machine, which was 8.398%, the total working loss rate was 1.71%, and the working state of the whole machine was good during the test, which can meet the design requirements of the whole machine.

Abstract:Ophiopogon japonicus is a major variety of traditional Chinese medicinal herbs, which mainly cultivated in Sichuan Province and Zhejiang Province, and the medicinal part of it is the tuberous root normally barrier underground. Focusing on the labor intensive process of manually removing the tuberous root from Ophiopogon japonicus, as well as the lack of mature mechanical tuberous root removal devices, a tuberous root removal device was specifically designed for Ophiopogon japonicus from Sichuan Province. The device utilized two sets of step type reciprocating cutters, a set of conveying clamping belt, a caliper, and a rack structure. By considering the requirements of the tuberous root of the Ophiopogon japonicus harvesting and measuring the plants' physical parameters, parameters such as overall structure dimensions, caliper interstices, conveying speed, cutting intervals and reciprocating stroke can be determined. The rigid-fliexible coupling simulation analysis between the Ophiopogon japonicus and the tuberous root removal tool was conducted by using ADAMS software. The cutting effect was simulated under the condition of the contact force between the cutting edge and the Ophiopogon japonicus plants exceeded the peak cutting force. The Felx module in ADAMS/view was used to apply flexible treatment to the Ophiopogon japonicas plants, and the changing rules in contact force and plant deformation during the contact collision process was analyzed. Choosing cutting speed, cutting inclination angle and cutting tool configuration as three factors to design a three level orthogonal test. The order of importance for factors affecting cutting effects was as follows: cutting inclination angle, cutting speed, and cutting tool configuration. Cutting speed was significantly correlated with the type of cutting tools. When proceeding high-speed cutting, the peak cutting force generated by the straight-tooth blade tool was greater than that of the other two configurations. Considering the principle of cutting stability reliability, cutting force and tool configuration minimization comprehensively, the optimized solution was carried out: the optimal working parameters was determined as cutting speed of 0.41m/s, cutting inclination angle of 0.3° and a cutting tool configuration of a triangular smooth-edge blade. Verified by simulation, this device had rational structure parameter and met the technical requirements for the harvest of Ophiopogon japonicas tuber, providing a theoretical and experimental references for the further optimization design of mechanical devices.

Abstract:Mainstream fruit and vegetable harvesting end-effectors differ greatly in structure according to the varieties of fruits and vegetables harvested and the wood fiber materials of fruit stems, and all of them have the problems of large space occupation, heavy structure, easy collision with fruits and easy damage, which is not conducive to cutting fruit stems. Therefore, it is very important to study an ultrasonic cutting knife with flexible structure, small space occupation and labor saving for cutting different wood fibers of fruit and vegetable harvesting end-effectors. The theory-finite element-experiment method was used to analyze the mechanism of cutting fruit stalks with ultrasonic cutter. Firstly, the physical field model of cutting fruit stalks by ultrasonic tool was established. Through the time-space analysis of the cutting process, the mathematical models of displacement, speed and acceleration were established, and the time-space discontinuity of the cutting process was analyzed. Based on the fracture mechanics, the cutting mechanism of fruit stalks was studied and the Matlab response surface diagram was drawn. It can be found that the force required for ultrasonic cutting was less than that for conventional cutting. Using finite element analysis software, the modal analysis and harmonic response analysis of ultrasonic tool were carried out, and the vibration modes of the tool at different frequencies and the optimal frequency for cutting were obtained. Finally, the ultrasonic cutting experiments of tangrine and orange stalks with different diameters were carried out by using a selfmade test bench at different cutting speeds. The experimental results showed that the cutting speed and vibration frequency were related to the cutting force. Under the high-frequency and low-frequency vibration, with the increase of fruit stem diameter, the cutting time was changed within 0.4s, and the cutting efficiency was stable. In the experiment, the maximum cutting force of ultrasonic cuttingwas much less than that of conventional cutting, which was less than 1N. Ultrasonic cutting can effectively reduce the cutting force and reduce the impact during cutting. Compared with conventional cutting, ultrasonic cutting had stronger cutting ability, and it was suitable to be applied to the end-effector of harvesting, thus achieving higher recovery efficiency.

Abstract:Sugarcane belongs to annual or perennial tropical and subtropical herbs, and the cutting quality of sugarcane stubble has a crucial impact on the growth of cane in the following year. Sugarcane cutting process is a high-speed, transient, and nonlinear dynamic problem. Aiming to explore the cutting mechanism of sugarcane stalks and study the cracking and propagation process of sugarcane cracks based on the fracture mechanics theory, which was to provide a theoretical basis for improving the cutting quality of sugarcane stalk, the mechanical model (bending moment, cutting force, etc.) of the base cutter cutting the stalk was established by analyzing the interaction between the cutter and the stalk, to explore the influencing factors of cutting force; and the high-speed camera technology was adopted to observe the cutting quality of sugarcane stubble cuts with one and multi-times cutting. It was found that the factors affecting the cutting performance were mainly related to the mechanical properties (diameter, density, stalk inclination angle, etc.) of sugarcane stalks and base cutters, the kinematic parameters (cutting frequency, cutting depth) of base cutters, and soil conditions. However, the multi-time base cutting would increase the probability of stubble damage, which was mainly caused by defects and cracks on the stalks after first cutting, and the cracks would extent under its own gravity and external load. Therefore, the internal cracks of the sugarcane stalk were first studied based on the fracture mechanics theory, the crack initiation conditions and the factors affecting the fracture angle were analyzed, and the influence of crack parameters on the stress intensity factor and the expansion amount were explored. The results showed that the change of crack angle and number led to the composite crack, and theⅠ-Ⅱ composite crack was the main crack type. The stress intensity factor fluctuated greatly with the crack angle changing, and the change of crack direction and extension was due to the angle between the maximum principal stress and the crack propagation surface was affected by the crack angle; when the crack angle was less than or equal to 75°, the crack would continue to expand along the original direction and then bend, however, the crack would expand along the direction perpendicular to the original crack when the crack angle was greater than 75°. During the propagation process of two cracks, the cracks propagated along a 45° direction from the original crack. This was due to the relative slip of the material inside the stem under the influence of the cracks. The shear stress reached its maximum value at the 45° direction. With the increase of number of cracks, the shielding effect between cracks was increased, which would result in the release of the stress in the material at the crack tip, the relative displacement near the crack tip was decreased, and the crack growth was decreased gradually.

Abstract:The working conditions of the large high clearance sprayer are complex and changeable, and the performance requirements of the suspension are different for the working scenarios with frequent switching of different road conditions. The traditional suspension system cannot meet the different vibration reduction requirements of the sprayer under multiple working conditions. In order to effectively improve the vibration reduction performance of large sprayer under different road conditions and effectively reduce energy consumption, a multi working condition active suspension control system based on road condition recognition was proposed. Firstly, the typical working conditions of the sprayer were analyzed, the road roughness characteristics were estimated online in real time, and the mapping relationship between the road surface characteristics and the suspension model was established. Then, the variable step size firefly algorithm was used to optimize the damping coefficient of the skyhook and the groundhook under each working condition, so as to effectively integrate the semi-active and active vibration reduction control methods, and design the upper and lower layers control system of the sprayer hydro pneumatic suspension. Finally, the field dynamic performance of the sprayer equipped with the control algorithm was tested. The test results showed that compared with the traditional suspension, the ride comfort of the active suspension on cement and farmland roads was increased by 11.0% and 22.8%, respectively, and the handling stability was improved by 24.5% and 3.4%, respectively. The control system proposed can effectively solve the contradiction between driving smoothness and handling stability of sprayer under different working conditions. The superiority of multi working condition active vibration control method based on road feature estimation was verified.

Abstract:Sugarcane is a perennial crop, and good cutting quality can promote the growth and germination of sugarcane in the coming year. A large number of experimental studies have shown that under-the-ground basecutting can effectively reduce the stubble damage. Due to the unevenness of sugarcane fields in the hilly and mountainous areas of southwestern China, it is difficult for sugarcane harvesters to keep under-the-ground basecutting, resulting in a large amount of stubble damage. Therefore, based on angle sensors and cutting pressure sensors, the test bench was developed to design a multi-sensor data fusion control system. The basecutter height control strategy based on PID algorithm was developed, and the simulation model of the system was built by Matlab/Simulink. The simulation results showed that the stability time of the system was 0.67s, and the overshoot was 8.6%. To verify the performance of the control system, the bench test was carried out to simulate the sugarcane ground surface with forward speed, ground wavelength and ground amplitude as test factors.The experimental results showed that the minimum average cutting depth error was 3.26mm when the forward speed was 1km/h, the ground wavelength was 1m and the ground amplitude was 4cm. When the forward speed was 3km/h, the ground wavelength was 1m, and the ground amplitude was 12cm, the maximum average cutting depth error was 8.87mm, and the basecutter can keep under-the-ground basecutting. The research provided data support and theoretical basis for the development of the control system for under-the-ground basecutting.

Abstract:In order to solve the problems of high labor intensity, shortage of rubber cutters and low efficiency of natural rubber tapping at night, a fixed automatic natural rubber tapping machine was designed. Moving motor and screw motor were used to drive the rubber cutter to rotate around the tree and move in the vertical direction, respectively. The two movements were combined to form the spiral line operation track of rubber cutter, so that the cutter can adjust the spiral line angle according to the growth of the rubber trees to ensure smooth rubber discharge. The screw motor was matched with the rotary encoder to adjust the rubber consumption of the rubber tapping machine, and double profiling springs were mounted to ensure uniform and stable cutting depth, overall, meeting different rubber tapping requirements, and reducing damage of rubber trees. A three factor and three level orthogonal experiment was conducted with the spiral angle, cutting speed, and rubber consumption as experimental factors, and the cutting machine energy consumption and the variation coefficient of rubber consumption as experimental indicators. The experimental results showed that the optimal results were obtained when the helix angle was 25°, the cutting speed was 0.6m/min, and the rubber consumption was 0.8mm. The average power consumption was 0.48W·h, the variation coefficient of rubber consumption was 7.61%, and the variation coefficient of cutting depth was 4.78%. The experimental results met the requirements.

Abstract:The pollination of pear trees in China requires a large amount of labor and low efficiency. Based on the liquid pollination technology, the liquid pollination experiment of multi-rotor UAV was carried out with horizontal scaffoled pear trees as the research object. The effects of liquid spraying amount per unit area, flight height and pollination mode on droplet deposition distribution and pollination effect of multi-rotor UAV were investigated. The experimental results showed that the performance of liquid spray pollination operation of multi-rotor UAV was stable, and the variation coefficient in groups of droplet deposition distribution was less than 20%. Droplet coverage and droplet coverage density were positively correlated with the spraying amount. When the liquid spraying amount was 6mL/m2, the change of flight height had a significant effect on the droplet deposition distribution. When the flight altitude was 4m, the coverage and coverage density of droplets were 7.06% and 84.77 droplets/cm2, respectively. The fruit setting rate of liquid pollinated flowers and inflorescence were 49.70% and 85.83%, respectively, which were 91% and 43% higher than that of natural pollination. When the amount of liquid pollination was 4.5mL/m2 and 6mL/m2, the fruit setting rate of the UAV liquid pollination was significantly different from that of the natural pollination inflorescence. Meawhile, there was no significant difference between the UAV liquid pollination and the backpack sprayer pollination inflorescence fruit setting rate, which could reach more than 80%. The above results showed that the higher the coverage and density of droplets were, the higher the fruit setting rate of flowers and inflorescences was. The optimal combination of UAV liquid pollination parameters was obtained when the UAV flying height was 4m and the pollen liquid spraying amount was 4.5mL/m2. The research result can provide theoretical guidance and data support for the optimization of liquid pollination operation parameters of multi-rotor UAV.

Abstract:Focusing on the problem of low drug utilization caused by uneven drug distribution when adding granular drugs to pig feed, a precise feed and drug integrated feeder was designed, with a metering inclined flute-wheel as the discharge device, and key components were designed to achieve real-time and accurate dosing under a given mass ratio range. Using the discrete element analysis software EDEM 2018 for discharge stability analysis, it was found that the stability of the feed discharge flute-wheel was good when the speed was 40~60r/min. Based on the (100~250)∶1 material and drug mass ratio, the drug discharge flute-wheel speed was determined to be 10~20r/min. The force chain analysis showed that the external force acting on feed particles and drug particles during the operation of the flute-wheel discharge device were smaller than the crushing force, and it would not cause particle breakage during the entire movement process. Through prototype testing, it was found that within the given speed range, the achievable material and drug mass ratio range was (93.4~251.8)∶1, and the given material and drug mass ratio was within this range. A bench test was conducted on the feed and drug ratio, and the experimental results showed that the precise feed and drug integrated feeder can achieve real-time and precise dosing into the feed, solving the problem of uneven distribution of small particle size and small flow rate drug particles in large particle size and large flow rate feed particles.

Abstract:When harvesting forage seeds, the current harvester has high total loss rate and crushing rate, poor cleaning effect of impurities and insufficient settlement. In order to solve the above problems, theoretical and structural analysis on the basis of the existing settling box were carried out, the cleaning structure of the settling box in combination with specific requirements and the overall structure shape was improved, a circular arc decelerating baffle was designed, the impact force on the seeds was reduced, the size and position of the baffle were adjusted, and more seeds were made clean by settling. The qualified rate of forage seed harvest quality was improved. By means of computational fluid dynamics simulation, the internal flow field of the settlement separation unit was simulated. Reynolds stress model and DPM model were selected to simulate the gas phase and solid phase respectively to obtain the airflow field distribution diagram and the particle field motion trajectory diagram. The results showed that the circular arc deceleration baffle designed compared with the structure of the broken line deceleration baffle increased the seed settlement rate, reduced the seed crushing rate, and increased the bottom capture rate when the inlet wind speed was increased. Field experiments were carried out to harvest alfalfa by forage seed harvester, and the harvest qualification rates of baffle free and broken line reducer were compared. The results showed that the sedimentation loss rate was 0.19% and the crushing rate was 0.9%, both of which met the national standard and proved the effectiveness of the device.

Abstract:There are some problems in the practical application of multi-function forage kneading machine, such as easy blockage and throwing distance can not meet the requirements of use, which are related to the coupling movement characteristics of material and air in the machine. In order to explore the coupling motion law of material and air flow in the kneading machine, the coupling model of air flow and loose material in the process of material crushing was established by using the coupling method of computational fluid dynamics (CFD) and discrete element method (DEM), and the coupling motion law between material and air flow was numerically simulated. The accuracy of the coupling model and numerical calculation was verified by airflow velocity test and throwing distance test. The results showed that the relative errors between the simulated and measured values of the outlet airflow velocity were within 8.1%. The relative error between the numerical calculation and the measured value of the average throwing distance of the material at the three speeds within the rated speed range was less than 5%, which verified the accuracy of the coupling model and the numerical calculation results. After the forage materials were broken, they moved circularly along the wall of the crushing room and were thrown out of the machine along the discharge pipe. After the kneading machine was stable, the maximum speed of indoor particle movement always fluctuated up and down at a certain average speed, and the average speed reflected the kinetic energy obtained by the material particles under the impact force of the hammer. The larger the kinetic energy of the granular material was, the less likely the device was to be blocked, and the farther the average throwing distance of the material was. The research provided a basis for optimizing the coupling motion characteristics of material and air flow in the kneading machine, avoiding clogging and increasing the throwing distance of material.

Abstract:In view of lack of fault diagnosis and fault warning of cotton pickers, a stacked denoising autoencoder (SDAE) fault diagnosis method based on particle swarm optimization (PSO) was proposed. The ratio between the speed of drum and speed of hydraulic drive device and the pressure of hydraulic drive device were used as the input of the fault diagnosis model of the cotton picking head. The PSO algorithm was used to self-adapt the number of hidden layer nodes, sparse parameters and the zero setting ratio of the input data in the SDAE network to determine the network structure. Then, the pre-processed data were input into the PSO-SDAE network for depth feature extraction. After forward propagation and reverse fine-tuning, the fault diagnosis model of picking head was obtained. Through the simulation test on the blockage fault of the cotton head, signals such as the speed of the drum and the pressure of the hydraulic drive device were obtained, and parameters of the normal operation, slight blockage and severe blockage of the drum of the cotton head were obtained. The original blockage fault data sample of the cotton head was formed, and the fault data sample was input into the fault diagnosis model of the cotton head to verify the algorithm. The test results showed that PSO-SDAE network diagnosis method was superior to SDAE network, support vector machines (SVM), back propagation neural network (BPNN) and deep belief network (DBN) in terms of feature extraction and fault diagnosis accuracy. PSO-SDAE fault diagnosis model can be used for fault diagnosis and early warning of cotton picker, which can reduce the failure rate of cotton picker and improve the working efficiency.

Abstract:In view of the insufficient theoretical research on the existing cotton bale wrapping technology and the issues include frequent blockage in the film feeding process and poor edging effect in domestic round baler cotton picker, a point-braking type film wrapping system was designed. The system mainly consisted of film feeding device, point-braking tensioning device, and control unit. By studying the film wrapping mechanism of cotton round bales, the necessary conditions for film feeding, the effective area of successful film feeding on the guide device, and the boundary conditions for the formation of edge wrapping at the ends of the cotton package were theoretically analyzed. A three-factors, three-levels quadratic regression orthogonal test was conducted with packing belt speed, point-braking cycle, the ratio of braking time to film feeding time (referred to as the braking ratio) as the test factors, and the average edge width of the cotton package and the average cotton bale density as the test indexes. A regression model was established to analyze the impact of each factor on the film wrapping system. The optimized working parameters was as follows: packing belt speed of 156r/min, point-braking cycle of 1200ms and braking ratio of 1.2. The combination of these parameters was tested in the field, the average edge width was 160mm and the average cotton bale density was 193kg/m3. The experimental results differed from the theoretically optimized values by less than 5%. The research results can provide reference for the design of cotton packaging and harvesting machinery.

Abstract:The large axial flow pump system ( LAPS ) used in coastal pump stations often needs to be equipped with overflow holes to improve the quality of the transition process. However, due to the unclear mechanism of adding overflow holes in the transition process of LAPS, there are many difficulties in design and application. Totally six kinds of overflow holes with different diameters were designed. Based on the secondary development of Flomaster software, the effects of overflow holes with different diameters on the synchronous start-up, asynchronous start-up, synchronous shut-up and asynchronous shut-up transition process of LAPS were studied by transient simulation method. The results showed that during the asynchronous start-up process, when the diameter of the overflow hole reached 2m, the maximum impact head was 1.67Hr, and the maximum impact power was 1.34Pr. Further increasing the diameter of the overflow hole had no significant gain in reducing the maximum impact head and power of the LAPS. In the process of asynchronous shut-up, the overflow hole can effectively delay the attenuation of LAPS flow and reduce the instantaneous head and power. However, when the diameter of the overflow hole reached 2m, the maximum impact head was 1.18Hr and the maximum impact power was 1.1Pr. Further increasing the diameter of the overflow hole had no obvious effect on reducing the maximum impact head and power during the asynchronous shut-up of LAPS. The larger the diameter of the overflow hole was, the better the effect on improving the quality of the transition process was, but when the diameter of the overflow hole was increased to a certain extent, the effect of continuing to increase the diameter of the overflow hole was not significantly improved.

Abstract:Aiming at the traditional hydraulic design which is difficult to meet the performance requirements of the downhole turbine generator, an inverse problem optimization design method with blade load as the design variable and output power as the objective function was proposed. Based on the blade load distribution law of the downhole turbine generator, “three-stage” was used to parameterize it, and based on the blade load distribution form of the initial model, the load at NC was linearly increased or decreased, and each time the change was 0.2 times, and six different blade load distribution schemes were designed, and the inverse problem method was used to design the impeller model, and the output power of six schemes was numerically calculated, and the output power of six schemes was calculated. The output power of the six schemes was numerically calculated, the highest output power was got for the scheme Ⅱ with value of 118.867W, the initial model output power was 93.2796W, the lowest output power was got for the scheme Ⅳ with value of 80.77W. The front loading point at the load of scheme Ⅱ was increased by 0.2 times compared with the initial model; the blade load of the blade at the load was linearly increased or decreased each time change for 0.2 times; the blade load distribution scheme designed six different blade load distribution, using the inverse problem approach to design the impeller model, the inverse problem approach to design the impeller model. The blade load was analyzed, and the relationship between the output power of the downhole turbine generator and the load at the front loading point was obtained. Based on the relationship between blade load and performance and the blade optimization algorithm designed, the target blade load distribution scheme applicable to the high performance of downhole turbine generator was obtained after continuous calculation and iteration, and the load at the front loading point of the target blade load was increased by 0.28 times compared with the initial model; based on the scheme for the design of the inverse problem, the comparison of the target blade load and the simulated load was closer; after numerical calculations, the target blade output power was increased and then decreased with the load at the front loading point, which was closer to the simulated load. After numerical calculation, the output power of the counter-problem design model with blade load distribution was 129.8W under the same conditions, which was 10.933W higher than the highest output power in the previous six schemes, with an increase of 9.26%; analyzing the pressure cloud diagram, it can be clearly observed that the location of the highest value of the blade pressure was in the vicinity of the front loading point, which was in the highpressure area, proving that the design of downhole turbocharger based on the blade load distribution had the highest value of the blade pressure, which was in the vicinity of the front loading point. The load distribution on the downhole turbine generator was proved based on the feasibility of the inverse problem optimization design method and theory.

Abstract:In order to address the challenge of determining optimal resolutions for capturing images of different features using UAVs, the DJI M600Pro UAV was employed to acquire visible light images of cotton fields during the bud stage. By combining ground survey data and utilizing three supervised classification algorithms: artificial neural networks (ANN), support vector machines (SVM), and random forest (RF), field feature identification was conducted. The analysis encompassed varying resolutions (1.00cm, 2.50cm, 5.00cm, 7.50cm, 10.00cm) to evaluate the accuracy of feature recognition. Additionally, algorithm execution times were considered, with the aim of identifying the best resolution and optimal algorithm for cotton field feature recognition at the field scale in Southern Xinjiang, considering resolution, accuracy, and processing time. Experimental results indicated that at a spatial resolution of 1.00cm, SVM exhibited the highest accuracy in feature recognition, achieving an overall accuracy of 99.857% and a Kappa coefficient of 0.997. As spatial resolution was decreased, both overall accuracy and Kappa coefficient demonstrated a decreasing trend. At resolutions of 2.50cm and 5.00cm, when utilizing the RF algorithm, the shortest execution times were observed. Land, cotton, and drip irrigation lines displayed favorable recognition accuracy, with overall accuracy and Kappa coefficients surpassing 99.137% and 0.983, respectively. With resolutions exceeding 5.00cm, both overall accuracy and Kappa coefficient declined, notably impacting the mapping accuracy of drip irrigation lines (producer's accuracy, PA) and user accuracy (user's accuracy, UA). Images with resolutions lower than 5.00cm effectively identified characteristic features of bud-stage cotton fields, offering guidance for the identification of field feature types and their distribution patterns.

Abstract:To improve the accuracy of apple cultivation area extraction, a CBAM-DeepLab V3+ model based on the fusion of Sentinel-2 and MODIS satellite images was proposed. The main factors affecting the accuracy of apple cultivation area extraction included the quality of remote sensing images and the performance of semantic segmentation models. From the perspective of image quality, a time-series spatiotemporal fusion algorithm called ESTARFM was employed to fuse Sentinel-2 and MODIS remote sensing image data, achieving higher spatial and temporal resolution data. Simultaneously, the training samples were increased from the original 800 to 2400, providing more abundant sample capacity for the subsequent semantic segmentation model. In terms of optimizing the semantic segmentation model, in order to further improve the accuracy of apple cultivation area extraction, a CBAM attention mechanism based on channel and spatial information was introduced into the DeepLab V3+ network, resulting in the development of the CBAM-DeepLab V3+ model. Compared with the original DeepLab V3+ model, the CBAM-DeepLab V3+ model with the addition of CBAM attention mechanism achieved significant breakthroughs in terms of slower fitting speed, less accurate edge target segmentation, inconsistency in segmenting large-scale targets, and existence of holes. These improvements enhanced the training and prediction performance of the model. The original Sentinel-2 images and the spatiotemporal fusion images were used, combined with the datasets of Wanggezhuang Town in Muping District and the apple dataset of Guanshui Town to compare the U-Net, FCN, DeepLab V3+ models, and the CBAM-DeepLab V3+ model. The research findings indicated that in terms of apple cultivation area extraction, the overall accuracy (MIoU) achieved by the optimized CBAM-DeepLab V3+ model was 84.6%, and the accuracy of apple cultivation area extraction reached 90.4%. In comparison, the MIoU of U-Net, FCN, and DeepLab V3+ models were 79.2%, 75%, and 81.2%, respectively. Additionally, the predicted apple cultivation area of Wanggezhuang Town in Muping District was 3433.33hm2, with only 233.33hm2 deviation compared with the data of 3666.66 hm2 published in the Yantai City National Economic and Social Development Statistics Report, resulting in a high prediction accuracy of 93.64%.

Abstract:To solve the problems of low efficiency and high cost in fruits and vegetables recognition, a fruit and vegetable recognition model based on residual block and attention mechanism was proposed, and successfully deployed in fruit and vegetable intelligent recognition equipment. The fruit and vegetable automatic recognition device was composed of Raspberry Pi, STM32F103ZET6, camera, weighing sensor, processor, display screen, micro printer, binding machine and power supply. The central controller interacted with the display screen to display various parameters in real time. The image and quality of the object to be measured were collected through the camera and weighing sensor. The fruit and vegetable automatic recognition model deployed in the Raspberry Pi could accurately identify the fruits and vegetables. At the same time, it cooperated with MCU STM32F103ZET6 to print fruit and vegetable related information and control the tying machine to seal and pack. Based on YOLO v5 network, an automatic recognition model RB+CBAM-YOLO v5 was constructed by adding residual blocks and attention mechanism. The network was trained with the self-made data set, and six kinds of networks were compared, and the optimal network was selected for the device side detection test. The experimental results showed that the accuracy rate, recall rate and mAP0.5 of RB+CBAM-YOLO v5 were 83.55%, 96.08% and 96.20%, respectively, which were 4.47 percentage points, 1.10 percentage points and 0.90 percentage points higher than those of YOLO v5. The RB+CBAM-YOLO v5 model was deployed in the embedded device Raspberry Pi, and the device could realize accurate identification, automatic weighing, printing slip and fast packaging functions, which could meet the needs of fruits and vegetables identification and unsold devices.

Abstract:Aiming at the problems of low accuracy and small number of samples in field amaranth identification, the YOLO v5 amaranth identification model was improved by introducing ASPP attention mechanism of expanding receptive field and extracting context information. The improved model would significantly improve F1 value and mAP index under low data set. The experimental results showed that the F1 value and mAP of amaranth identification model was increased by 13 percentage points and 18.6 percentage points after the introduction of ASPP attention mechanism in low data set. The detection rate of amaranth was increased by 15.4 percentage points with horizontal recording. Therefore, the research provided an effective method for the identification of amaranth or other weeds under low data sets, and prepared for the research of weed identification and management in the agricultural field.

Abstract:Straw coverage rate is one of the most important evaluation indexes for conservation tillage. In view of the difficult problem of different shapes of straw and difficult recognition of the fine straw, a straw coverage rate detection algorithm based on K-means and color space distance gray-scale method was proposed to solve the problem of different straw shapes and difficult identification of broken straw. The color spatial distance method was used to preprocess the straw image. The classification recognition of the straw and soil background was realized based on the K-means algorithm. After the identified images were processed by using the mathematical morphology method, the coverage rate of the straw image was calculated. In 2022, the field experiment was conducted to verify the algorithm on 220 corn straw images collected from Beijing Xiaotangshan National Precision Agriculture Research Demonstration Base. The experimental results showed that the recognition accuracy of low straw coverage rate (0~30%) was 90%. And the recognition accuracy of medium straw coverage rate (30%~60%) was 88%. For the high straw coverage rate (more than 60%), the recognition accuracy reached 86%. The accuracy rate of overall identification grading reached 9818%. The straw coverage rate detection algorithm based on K-means and color spatial distance gray provided a rapid detection method and means for conservation tillage evaluation.

Abstract:Aiming to solve the problems that traditional grain weight classification depends on manual sorting, such as heavy workload, high error rate and lax classification standard, an improved two-stream convolutional neural network model was proposed based on ECA to classify rice by single grain weight. Firstly, images of each group of rice (a group consists seven single rice grains) were taken from two different perspectives: front view and top view. For five traditional supervised models (naive Bayes, decision tree, random forest, K-nearest neighbor, support vector machine), voting mechanism optimization based on genetic algorithm (GA)(GA-SVM) and integrated model (RF+GA-SVM), single grain images were separated through image preprocessing and contour detection. Color moment, local binary pattern (LBP) and Canny operator were used to extract grain color, texture and edge features. And then through principal component analysis (PCA), the principal features were extracted to train each model. For the constructed single-stream convolutional neural network model, two-stream convolutional neural network model (FV-CNN) and the improved two-stream convolutional neural network model were proposed based on ECA (ECA-FV-CNN), the pre-processed images were divided into training set, verification set and test set according to the ratio of 6∶2∶2, and data enhancement were carried out for each data set, and then the models were trained. By comparing and analyzing the above models, the traditional machine learning model, RF+GA-SVM, had the best effect, but its highest accuracy was only 72% when the single grain weight was set for three-graded. Experimental verification showed that the ECA-FV-CNN model proposed had the best performance, and its accuracy for the single grain weight classification of three-graded, four-graded and five-graded reached 94.0%, 92.3% and 71.0%, respectively. However, the accuracies of single-stream convolutional neural network model and FV-CNN model for single grain weight grading were 92.7%, 91.1%, 61.1% and 93.0%, 91.6%, 65.6%, respectively. The grading effect of FV-CNN model was better than that of single-stream convolutional neural network model in three experiments, which showed that the two-branch network training was better than that of single-branch rice single grain weight grading. The accuracy of ECA-FV-CNN model in three grading experiments was 16.2% higher than that of single-stream convolutional neural network model and 8.2% higher than that of FV-CNN model. The results showed that the introduction of ECA module was effective for rice single grain classification, and the improved two-stream convolutional neural network model based on ECA can improve the classification accuracy of rice single grain weight, and the classification of rice single grain weight can be achieved by using computer vision technology, making up for the shortcomings of traditional methods, and improving the classification standard of grain screening.

Abstract:The improved YOLO v8-Pose model was established to identify red ripe strawberries and detect the key points of the stem in greenhouse strawberries under elevated cultivation mode. By comparing the YOLO v5-Pose, YOLO v7-Pose and YOLO v8-Pose models, the YOLO v8-Pose model was determined to be used as the model to identify and predict the key points of red ripe strawberries. Based on YOLO v8-Pose, Slim-neck module and CBAM attention mechanism module were added to its network structure to improve the feature extraction ability of the model for small target objects, so as to adapt to the characteristics of strawberry data set. The P, R and mAP-kp of the improved YOLO v8-Pose were 98.14%, 94.54% and 97.91%, respectively, which can effectively detect red ripe strawberries and accurately mark the key points of the fruit stalk, which was 5.41, 5.31 and 8.29 percentage points higher than that of YOLO v8-Pose. The model memory footprint was 22MB, which was 6MB less than that of the YOLO v8-Pose footprint.In addition, according to the unstructured characteristics of the orchard, the influence of light, occlusion and shooting angle on the model prediction was explored. Compared with the recognition and stem prediction of the improved YOLO v8-Pose model in the complex environment, the mAP-kp of the improved YOLO v8-Pose under the influence of occlusion, light and angle was 94.52%, 95.48% and 94.63%, respectively. Compared with YOLO v8-Pose, it was 8.9, 10.75 and 5.17 percentage points higher, respectively. The improved YOLO v8-Pose can ensure the accuracy of the network model, and at the same time, it had good robustness to the effects of occlusion, light and shooting angle, etc., which can realize the identification of red ripe strawberries in complex environments and the prediction of key points of fruit stalk.

Abstract:Cotton is an important economic crop in China，and the prediction of cotton yield helps in economic regulation and regulation of planting patterns，thereby improving production returns. At present，traditional manual production measurement methods have problems such as high labor intensity and low measurement accuracy. To solve this problem，cotton images after spraying defoliant were selected as the research object，and relevant datasets were constructed. At the same time， the calculation formula for the number of cotton plants， cotton bolls， and single boll seed cotton quality per unit area and the improved YOLO v5 algorithm model were used as the core algorithm to design a cotton yield prediction system based on Android mobile devices. Image information was obtained by choosing to take photos on a mobile phone or calling an album， and performing data analysis and processing on the target image to achieve cotton yield prediction. Using the detection box of cotton in the image to detect cotton bolls， the cotton yield per hectare was automatically calculated based on different soil types. Compared with the actual yield， the average error between the actual yield and predicted yield of seed cotton and lint was 122.01kg/hm2 and 57.98kg/hm2， and the model had high accuracy on the mobile phone. Compared with the original YOLO v5 model， the accuracy P and recall R were increased by 19.58 percentage points and 16.84 percentage points， respectively， with values of 90.95% and 73.16%. After comparative testing on three types of mobile phones， the system ran smoothly and the yield prediction results did not differ significantly. The results indicated that the designed cotton yield prediction system had good performance in field yield measurement and algorithm operation， and can provide technical reference for cotton yield prediction.

Abstract:In order to meet the needs of rapid detection of field crop growth and guiding variable management, a field crop chlorophyll intelligent detection system based on multi-spectral imaging was designed and developed with maize as an example. It included visible light (RGB) and near-infrared (NIR) image acquisition module, main control processor module, model acceleration module, display and power module, which was used to realize intelligent identification of corn plants and integrated detection of chlorophyll index. Firstly, the canopy image data set of maize seedling stage and jointing stage were collected, and two deep learning models of plant canopy instance segmentation and plant center target detection were compared. A corn plant location detection model based on MobileDet+SSDLite (single shot multibox detector lite) lightweight network was constructed to realize corn plant identification. Secondly, the identified plant heart RGB-NIR images were extracted, the matching and segmentation of RGB and NIR images were carried out, and the gray values of R, G, B and NIR were extracted to calculate the vegetation index. SPXY algorithm (sample set portioning based on joint X-Y distances) and SPA (successive projections algorithm) were used. The samples of the dataset were divided and the characteristic variables were screened, and then GPR (Gaussian process regression) algorithm was selected to establish the chlorophyll index detection model. The results showed that the recognition rate of the model reached 88.7% in the complex environment of occlusion overlap, and the recognition accuracy reached more than 90% in the non-overlapping environment. The model determination coefficient R2 of the modeling set of the chlorophyll content index detection model was 0.62, and the model determination coefficient R2 of the test set was 0.61. Field tests on the developed system showed that the detection rate of the system can reach 14.6 frames per second, and the average accuracy was 92.9%. The research results can effectively solve the problem of corn nutritional status detection in field environment, meeting the real-time detection requirements of field environment, and providing solutions and technical support for intelligent perception of crop production.

Abstract:In order to meet the needs of rapid detection of crop growth and guide variable management, a portable crop chlorophyll detector was developed based on the selection and optimization of characteristic wavelength of crop chlorophyll spectral response. Firstly, the reflectance spectra of field leaves were collected by ASD spectrometers, and the true chlorophyll content of leaves was extracted to screen the wavelength of chlorophyll sensitive response based on hyperspectral reflectance. The Monre Carlo uninformative variables elimination (MC-UVE) algorithm was used to select 10~100 characteristic wavelengths, which showed that the optimal chlorophyll content detection ability was achieved when 50 characteristic wavelengths were used.Secondly, the AS7265x spectral sensor was selected to cover 50 wavelength positions screened in 12 intervals with FWHM (full width at half maximum) of 20nm. The chlorophyll detector was designed to include modules such as sensor, main controller, display and control, and realize the functions of collection, processing, display and storage of reflected spectral data of crop canopy.Sensor reflectivity calibration and field application tests were carried out, based on the reflectivity of 12 bandwidths, a partial least squares regression detection model of chlorophyll content was constructed, and the coefficient of determination of the verification set was 0.628. The normalized difference red edge index (NDRE: 730nm, 900nm) and the green normalized difference vegetative index (GNDVI: 535nm, 900nm) were further combined, and the accuracy of the detection model was improved to be 0.69. By embedding the model into the system, the rapid detection of chlorophyll content in the field was realized, which provided technical support for the efficient analysis of crop growth.

Abstract:Aiming at the potential change caused by chloride leakage of Ag/AgCl reference electrode in in-situ soil monitoring, a flexible Ag/AgCl reference electrode modified with liquid metal fiber mat was prepared. The influence of preparation parameters of liquid metal fiber mat on the potential stability of the reference electrode was studied and the parameters were optimized. In order to improve the stability (reduce the standard deviation (s)), one solution parameter and four process parameters were selected for single factor test. Finally, three key factors were selected, namely the mass ratio of polystyrene (styrene-block-butadiene-block-styrene) (SBS) to 1, 2-dichloroethane, the collection distance of electrospun fibers and the number of tension required for activation electrode. Then based on Box-Behnken principle, the 3-factor and 3-level response surface optimization experiments were carried out. The results showed that the order of influence was as follows: collection distance, mass ratio and drawing numbers. The coupling effect of collection distance and drawing number was significant. When the mass ratio was 1∶5, the collection distance was 19cm, and the drawing number was 1150 times, the model prediction results were better, and the average relative error of the verification set was no more than 4.4%, indicating that the parameter optimization model was reliable. In addition, the flexible Ag/AgCl reference electrode was prepared based on the optimal parameters and applied to the soil column nitrate nitrogen and pH monitoring test. The absolute error between the nitrate nitrogen detection results and the commercial electrode measurements was less than 5.55mg/L, the relative error was less than 7.2%, and the root-mean square error was 1.98mg/L. The absolute and relative errors between the pH test results and the commercial electrode measurements were less than 0.21, less than 2.8%, and the root-meansquare error was only 0.17. At the same time, the data of the two groups of flexible electrodes were consistent with that of commercial electrodes.

Abstract:A set of crop root growth monitoring minirhizotron device was designed and developed to achieve real-time monitoring and image collection of crop roots at different depths. The collected images can be displayed and stored in real-time through upper computer software. The device consisted of a monitoring tube and a control box. The monitoring tube used acrylic material to make a transparent shell, and a sliding track was set up inside the tube through screws and guide rods. The motion control of the camera on the track was achieved by using a stepper motor. The control box was based on the STM32 microcontroller as the core control board, and corresponding peripheral modules were selected according to actual needs. The tomato root system was used as the research object for 98d of image collection, and the root image was analyzed by using the software RhizoVision Explore. The experimental results showed that the overall growth rate of tomato roots was faster in the first 70d, and gradually stabilized in the last 28d, and distributed densely at depths of 6~10cm. The root length density reached its maximum value of 1.22cm/cm3 in the 91st day at a depth of 10cm. The analysis results were consistent with the growth pattern of tomato roots, indicating that this root growth monitoring minirhizotron device can complete long-term online monitoring of crop roots without affecting the continuous growth of roots, meeting the requirements of crop root monitoring.

Abstract:Aiming at the existing detection methods, it is difficult to accurately detect the information of kiwifruit pests and diseases on single plants in orchards over a large area, and the information obtained by ground or remote sensing data alone is incomplete. By building the ground data collection equipment, together with the remote sensing images collected by the UAV, more comprehensive information on kiwifruit canopy leaf pests and diseases was obtained from both air and ground perspectives. The Pytorch deep learning framework was selected and the YOLO v5s model was used for target detection of pest and disease leaves. When calculating the infestation rate of a single fruit tree, the pixel values of infested leaves and canopy leaves were counted by image processing instead of number counting. During the calculation of canopy pixel values, K-means cluster analysis and Otsu method threshold segmentation algorithm were compared, and both methods were more accurate, with the latter taking less time and being simpler to operate. As a result, the precision rate of the detection model was 99.54%, the recall rate was 99.24%, and the mean values of target detection and classification loss in the validation set were 0.08469 and 0.00083, respectively. Meanwhile, totally 20 disease and pest data from UAV and ground were selected, respectively, and the predicted values of the number of pest and disease leaves obtained from the detection model were compared with the real values labeled manually, and the mean absolute value errors of the disease and pest detection models from remote sensing and ground were 3.5, 2.5, 0.9, and 0.45, respectively. The detection effect of the ground-based data was better than that of the remote sensing data. The research result can provide a basis for the establishment of kiwifruit orchard pest and disease detection system, and also provide guidance for the fine management of kiwifruit orchards.

Abstract:Crop diseases and pests have a significant impact on agricultural yield and quality. Digital image processing technology plays an important role in identifying crop diseases and pests. Deep learning has achieved significant breakthroughs in this field, with better results than traditional methods. The issue of crop pest and disease detection was defined. The deep learning method had stronger feature extraction ability, which can accurately capture subtle features, improve detection accuracy and reliability. Deep learning provided strong support for agriculture. The research of crop pest detection based on deep learning was summarized from three aspects: classful network, detection network and segmentation network, the advantages and disadvantages of each method were summarized, and the performance of existing research was compared. On this basis, the challenges that deep learning based crop disease and pest detection algorithms may face in practical applications were further explored, and corresponding solutions and research ideas were proposed. These findings and reflections had important guiding significance for promoting the development of crop pest detection technology in practical applications. Finally, the future trends of crop disease and pest detection based on deep learning were analyzed and prospected.

Abstract:Wheat diseases have frequently threatened the yield and quality of wheat production. In order to quickly and comprehensively monitor wheat diseases in the field and identify diseases in different growth parts of wheat based on the characteristics of wheat disease, a dual camera wheat disease detection device based on a lightweight model was designed. The device was composed of a dual camera acquisition module and a main control module, and it can collect and detect wheat powdery mildew at multiple locations in cooperation with the disease detection software system. In order to ensure the feasibility of the model deployment in the detection device, a lightweight improved powdery mildew target detection model based on YOLO v7-tiny model (YOLO v7tiny-ShuffleNet v1, YT-SFNet) was proposed. To verify the accuracy and detection speed of the lightweight model, it was trained and compared with the YOLO v7-tiny model. The results showed that the YT-SFNet model improved the average accuracy by 0.57 percentage points compared with YOLO v7-tiny model. The detection time and model size were decreased by 2.4ms and 3.2MB, respectively. Finally, the lightweight model and software system were transplanted to the main control module of the device, and a test set was created to test the performance of the devices detection accuracy and detection speed. Its recognition accuracy for the test set was 86.2%, with good stability in detection speed, and the average time spent on the entire process of processing, detecting, and displaying and saving a single disease image was 0.5079s.

Abstract:The dynamic highprecision measurement of fertilizer application rate is a prerequisite for implementing variable fertilization. Although there are various methods for measuring fertilizer quality and flow in the field, there are still some problems such as inaccurate measurement and inability to adapt to the working environment when applying to the field. A microwave-based particle fertilizer mass flow measurement system was developed. A flow mass measurement model and method were proposed, and the agricultural granular fertilizer Stanley 15-15-15 and Sacokufu 15-15-15 were used as the experimental object. The microwave sensor distance and fertilizer discharge rate were controlled, and the data was smoothed by using a Kalman filter. The experiment achieved good results. The analysis of experimental data showed that the dominant frequency of the particle fertilizer echo signal was only related to the distance between the electric fertilizer discharge device and the sensor, while the power spectral density was only related to the number of fertilizer particles. By using the least squares method, the response relationship between the two compound fertilizers, Qt and Sval, was established. The determination coefficient R2 of the response relationship between the two types of compound fertilizers was not less than 0.9858, and the response relationships were validated. The measurement range for Sacokufu 15-15-15 was 1119.8~2065.9g/min, with a maximum measurement error of 6.35%. The measurement range for Stanley 15-15-15 was 1071.9~1877.9g/min, with a maximum measurement error of 4.85%. Its measurement performance fulfilled the operational needs of the task.

Abstract:Humus (HS) is an important indicator of the nutrient quality of organic fertilisers. During the composting process, the addition of biochar and cotton straw wood vinegar liquid leads to changes in the internal environment of the heap, but the changes in the chemical properties of HS are not clear. The morphology of heavy metals (HMs) (e.g. Pb, Cr, Cd, Ni) was detected by flame atomic absorber, and Fourier transform infrared spectroscopy (FTIR) method and three-dimensional excitation-emission matrix fluorescence spectroscopy (3D-EEM) were used to characterise the complexes of HS and HMs from different angles. Meanwhile, mathematical statistics, correlation analysis and redundancy analysis (RDA) were used to compare the decay indices (temperature, pH value), humification capacity (HS, fulvic acid (FA), huminic acid (HA), huminic acid and fulvic acid ratio (H/F)) and functional groups of the test groups. The results of the study showed that the wood vinegar liquid treatment could make the humification of the compost relatively high, and the peak located at 876~835cm-1 was significantly enhanced, indicating the accumulation of aromatic structure, and the H/F finally reached more than 2.3 in all cases. The HS content of the T1 biochar treatment was located between T4 and T3, and the H/F finally reached 3.67. The passivation process of Cr by the T1 vs exchange state final passivation proportion to 2%. T4 treatment group in the Pb passivation process, the proportion of the final residue state was as high as 68%. Cd was more affected by T4, and eventually showed the transfer of 2%, 10% and 11% from the oxidation state to the exchange state, reduction state and residue state, respectively. However, the passivation of Ni by the addition of either biochar or wood vinegar liquid did not show any significant trend during the stacking process, and the proportions of exchange, reduced, oxidized and residual states were stable at 1%~2%, 5%~7%, 26%~35% and 56%~68%, respectively, indicating that the effect of HS on Ni was relatively small in this experiment. FTIR further confirmed the role of HS as a core agronomic and carboxylate-rich basic properties of functional substances. The aromaticity of HS was gradually increased during the composting process, which enhanced the complexing ability with Pb, Cr and Cd ions. In addition, it was found that the wood vinegar liquid with a mass fraction of 1.75% performed better overall in the composting process of the pig manure base. In summary, the following adsorption mechanisms existed for wood vinegar liquid and biochar: the special functional groups of wood vinegar liquid complexed with HMs ions; biochar relied mainly on adsorption with HMs; and the mechanism of Ni in composting may be more inclined to bind with nitrate ions.

Abstract:The high concentration of salt inhibits the growth of microalgae in wastewater. Indole-3-acetic acid (IAA) was added to promote the growth of microalgae under salt stress. At the same time, the effects of different algae species and salt ion on the growth, nitrogen and phosphorus removal of microalgae cultured in high-salt wastewater were investigated. The results showed that the maximum biomass of Chlorella vulgaris, Desmodesmus sp. and Scenedesmus quadricauda could reach 0.55g/L, 0.66g/L and 0.75g/L under the flow-feeding mode. The maximum specific growth rates of Desmodesmus sp.and Scenedesmus quadricauda were 0.18d-1 and 0.17d-1 on the 4th day, while Chlorella vulgaris were 0.14d-1 on the 8th day. The removal rates of TN and NH+4 N were 70.7%, 88.5%, 79.7% and 90.7%, 92.6%, 92.4%, respectively. The removal rates of TP were more than 90%. The growth of Scenedesmus quadricauda in the high-concentration IAA (20 mg/L) was the best, and the biomass was up to 0.403g/L. The growth of Scenedesmus quadricauda in the low-concentration IAA (≤2mg/L) was not significantly promoted or even inhibited. The removal rates of TN and NH+4 N were 20% and 44.1% in 20mg/L IAA, respectively. The TP removal rate was 97.2% in 20mg/L IAA, and the other five groups were maintained at about 15%. In addition, the biomass under different salt conditions was 0.667g/L (blank group), 0.750g/L (Cl-group), 0.898g/L (NH+4 group) and 1.037g/L (NH+4+Cl-group), respectively. There was no significant difference in the removal efficiency of N and P among all groups. The removal rates of TN and NH+4N was about 30% and 50%, and TP was more than 90%. The above studies indicated that high concentration of IAA could significantly promote the growth of Scenedesmus quadricauda under salt stress, and the biomass was the highest in mixed group (NH+4+Cl-). Therefore, the research result can provide data support for the subsequent application of microalgae in the treatment of high-salt wastewater.

Abstract:Promoting of renewable energy and low-carbon approaches in rural China has become essential to achieving the goals of low-carbon villages. Therefore, considering the Xinxing Village, Hubei Province as a case study. Based on the analysis of rural energy demand and renewable energy resources, a renewable energy cogeneration system integrated with wind, solar, and biomass energy was proposed. A rural renewable energy cogeneration model was established based on the HOMER Pro simulation system with the lowest cost and lowest average electricity price as the system objective functions, and the thermal power coverage and biomass utilization rate as the system evaluation indicators. The construction plan of the renewable energy cogeneration system in Xinxing Village was obtained through simulation analysis and capacity optimization, which consisted of 674kW photovoltaic, 200kW wind turbine, 500kW CHP cogeneration unit and external heat source. Moreover, the lowest system cost and electricity price of renewable energy were 2.30×107CNY/a and 0.986CNY/(kW·h). Further sensitivity analysis results indicated that the increased power load could increase system costs and electricity prices. While increasing the nominal discount rate reduced costs but led to an increase in electricity prices. Daily biomass input had no impact on economic indicators. The emission reduction assessment results showed that compared with traditional energy supply methods, the annual emission reduction of the system reached 410.77t, equivalent to 28.19% of annual emissions of traditional energy supply method in Xinxing Village. The overall economic and practical performance of this renewable energy supply system was excellent. The research result can provide a scientific reference for the energy revolution and the construction of low-carbon villages and towns in rural areas.

Abstract:In response to the problems of excessive parameter quantity, high computational cost, and low accuracy in the recognition model of soybean abnormal seeds, an improved lightweight neural network MobileNetV3 model was proposed, which reduced the number of layers, accelerated the training and inference speed of the model, increased the nonlinear discrimination ability of the model by adding fully connected layers and softmax layers, and facilitated the output of multiple classification tasks, by using global average pooling instead of global maximum pooling to reduce information loss, and increasing the model's generalization ability by adding a Dropout layer and removing the SE Block attention mechanism in MobileNetV3. The experimental results showed that after comparing the soybean seed image data with the traditional convolutional neural networks AlexNet, VGG16, and lightweight neural network MobilenetV3, the AlexNet algorithm's final mean average precision (mAP) was 87.3%, and the VGG16 algorithm's mAP was 87.7%. The difference in mAP between the two was small, but there was a significant difference in model size and training time during the training process, the AlexNet model had a model size of 7070kB and a training time of 5420.59s, while the VGG16 model had a model size of 19674kB and a training time of 8282.68s. Overall, AlexNet was relatively better. The recognition and training of the lightweight neural network MobileNetV3 model resulted in a model size of 32153kB, a training time of 6298.29s, and an mAP of 90.6%, which was higher than that of the two traditional algorithms and more suitable for the classification and recognition of abnormal soybean seeds. In order to improve training accuracy and speed, the structure of the MobileNetV3 network model was adjusted and improved. The optimized Opt-MobileNetV3 network model mAP reached 95.7%, which was 5.1 percentage points higher than that of the traditional MobileNetV3 neural network mAP. The model size was 9317kB, reduced by 22836kB, and training time was saved by 696.57s. The optimized model achieved reducing model size, improving accuracy, and faster training speed, which can meet the task of identifying abnormal soybean seeds.

Abstract:In response to the problems of simple structure, poor air separation effect, and single air separation test object in tea air separation equipment, in order to improve the air separation effect and tea quality, a green tea air separation equipment was designed and tea air separation performance tests were conducted under different processes. Firstly, the basic structure and working principle of the tea air separation equipment were elaborated, and its key components were theoretically analyzed. The relevant design parameters of the air separation room, fan, transmission device, and material transfer device were determined. Then, based on fluid mechanics, the simulation of fluid in the air separation room was carried out, and the movement of particles in the air separation process was simulated by discrete element simulation. It was concluded that the streamline was stable when the fan speed was 6~6.5m/s, and the air separation effect was good. Finally, validation tests were conducted on the air separation performance of famous and high-quality green tea air separation equipment. Prototype performance tests and air separation effect tests were conducted to explore the impact of different operating frequencies and air separation sequences of fans on tea quality. The experimental results showed that the wind speed variation coefficients of the wind separation equipment were all less than 8%, and the optimal operating frequency of the fan frequency converter was 35Hz. The productivity per unit effective width was greater than 7kg/(cm·h),and the output per kilowatt hour was greater than 420kg/(kW·h). The fresh and killed leaves of tea was upgraded from B grade to A- grade and A grade, and the selection rate of tea was higher than 90%, indicating a significant improvement in the air selection effect. In each experimental group, the main tea outlet was overall superior to the other tea outlets, and the tea quality of the main tea outlet was the best in the green killing grading, with the highest sensory evaluation score for tea.

Abstract:As the final output product of tobacco agriculture, the appearance quality of cigarette is the key link to control. However, there is no special detection equipment for the whole appearance defect detection of cigarette, while many finely categorized defects exist on the external surface of cigarette, which cannot be detected manually in a comprehensive, detailed and systematic manner. And the standardization of quality testing and the digitization of quality control are severely inhibited. A cigarette appearance inspection device was designed and tested after analyzing the technical systems of the testing equipment by using the black box approach. Firstly, the equipment was designed by combining innovative mechanical design and visual inspection function, which consisted mainly of device for storing and supplying cigarettes, device for forming units, full appearance imaging device of cigarette, sorting device and defects detection algorithm programs. Secondly, key factors affecting cigarette running stability, cigarette integrity, and imaging quality were tested on the equipment. The experimental results showed that the equipment can image cigarette completely and operate stably under the optimum parameters: the lifting speed of slide plate was 0.3m/s, the angle of transition plate was 40°, the displacement speed of roller was 0.045m/s, the expansion speed of the cylinder was 20mm/s, the spring stiffness was 3N/mm, the angle of the light source was 10°, and the height difference between the light source and the cigarette was 30mm. Thirdly, cigarette of 54mm×100 size was imaged in the equipment, and the resulting images were processed by improved HourglassNet-YOLO v4 network model, while the result showed that the accuracy of defect detection and defect classification can reach 98% and 95.4%, respectively. The experimental results showed that the device can meet the needs of cigarette appearance detection and can provide a reference value for cylindrical object appearance detection.

Abstract:To enhance the field adaptability and stability of agricultural robot chassis, a four-wheel independent drive steering chassis was specifically designed for acquiring wheat phenotypic information in Shandong Province. Taking into account the agricultural requirements for wheat cultivation and the driving terrain conditions, comprehensive layout plans and main technical parameters of the chassis were determined. The design focused on the drive components, steering components, and swing arm balance components of the chassis, followed by parameter verification and component selection. An ANSYS finite element model was constructed to analyze stress deformation in the swing arm balance mechanism and simulate the vibration modes of the frame. The simulation results indicated that the swing arm balance mechanism exhibited sufficient strength and stiffness to meet the design requirements, while the frame effectively mitigated resonance caused by terrain excitation. Furthermore, an ADAMS dynamic simulation model of the chassis was established to conduct longitudinal and lateral stability analysis and assess the chassis ability to traverse single side protrusions and dents. The simulation outcomes demonstrated that both the transverse and longitudinal stability of the chassis met the design criteria, and the swing arm balance mechanism effectively compensated for changes in centroid height caused by unilateral obstacles, thereby enhancing driving stability. Field experiments confirmed the excellent driving performance of the robot chassis, with an average deviation rate of 0.51% on hard ground during straight driving and 1.13% on field terrain. The center point offset for turning in place was measured as 3.1mm, and the minimum turning radius for Ackermann turning was determined to be 1.125mm. Additionally, the longitudinal tilt angle reached 34°, while the lateral tilt angle reached 28°. It was noteworthy that the maximum height for traversing unilateral obstacles and the maximum depth for crossing unilateral pits were both 160mm.

Abstract:To address the issue of estimating tractor side-slip in hilly and mountainous terrains, a multi-sensor information fusion algorithm that integrated machine vision and the global navigation satellite system (GNSS) was proposed. Initially, a simplified kinematic model of the tractor was presented, followed by separate discussions on skid estimation methods based on GNSS and machine vision technologies. The feasibility of the skid estimation methods was validated through joint simulation using CarSim and Simulink. Kalman filtering and weighting functions were introduced to dynamically fuse and adjust sensor data. An experimental platform mimicking hilly terrains was set up to conduct tests under varying road slopes, GNSS coverage conditions, and road surface conditions. The experimental results showed that under dry road conditions and GNSS blockage, the total skid amounts for tractors driving on 9° and 18° slopes were 0.322m and 0.432m, respectively, with relative error of 7.86% and 6.00%, which indicated that accurate skid estimation was still achievable even when GNSS signals were obstructed. The research result can provide methods and experimental foundations for precise lateral control of tractors.

Abstract:The real-time, synchronous, and suitable-frequency acquisition of working condition parameters of tractor field operation is of great significance for reliability analysis and optimization. A tractor-operating condition parameter testing system based on the NI-C DAQ controller was developed, the required sensors were selected, designed, and installed, and the detection software and remote monitoring platform were developed with the LabVIEW platform. The system consisted of sensors, data acquisition controllers, and data acquisition and monitoring platforms, which can realize the parameter measurement of various mechanisms such as engines, wheels, axles, suspension systems, and implements. In addition, the system can be remotely controlled and monitored in real-time via a portable touch screen. To verify the accuracy and stability of the testing system, signal error tests, and typical parameter field tests were carried out. The results of the signal error test showed that the acquisition error, packet loss rate, and initial error of all types of signals can meet the requirements of the parameter testing system. In the field test, the maximum relative error between the measured tractor wheel speed and the actual speed was 3.1%. The maximum relative error between the calculated and measured values of the horizontal traction force of the suspension system was 4.5%. The accuracy of recognizing the field operation ground type based on the measured wheel acceleration was 96%. The R2 for fitting the tillage depth according to the suspension position was 0.99156. Finally, a 24h continuous operation test of the testing system for field operation was carried out, and the system was able to maintain stable operation and accurate data at all times. Compared with similar systems, the tractor operating condition information testing system developed by this system collected more parameters, which was more convenient to operate, and provided an effective means of data acquisition for reliability analysis and optimization.

Abstract:Aiming at the serious problem of emission pollution of agricultural tractors, especially limiting the emission of NOx and Soot, taking a certain model of agricultural diesel engine of YITO as the research object, relevant research on engine emission optimization was carried out by using a combination of system modeling and simulation, bench test verification and simulation analysis. Firstly, a three-dimensional model of the combustion chamber of the agricultural tractor was constructed and imported into CONVERGE for combustion emission simulation and emulation, and through the comparison of the test values of the model cylinder pressure and heat release rate with the simulation values, it was proved that the model could describe the internal combustion and emission process of the engine better and had a high degree of accuracy. After that, an artificial neural network was established as an agent model with the shrinkage rate, camber depth, and combustion chamber depth as inputs, and NOx and Soot emitted from the engine as outputs. The goodness of fit R2 and the mean relative error (MRE) were calculated to verify the accuracy of the artificial neural network. Then an improved particle swarm optimization algorithm was proposed on this basis to obtain the optimal parameter combinations of the combustion chamber shrinkage rate, camber depth, and combustion chamber depth to form combustion chamber structure and imported into CONVERGE software to perform emission simulation calculations and compare the emissions with the original combustion chamber. It was found that the combustion chamber structure can reduce the engine NOx and Soot emissions, which provided reference and ideas for the design and development of the combustion chamber system of the related agricultural tractors.