Abstract:Intelligent agricultural machinery equipment is a complex of informatization, automation, and intelligence. Its intelligence level is mainly reflected in the application of advanced technologies such as state perception, data analysis, scientific decision-making, and autonomous control in the operation process. At present, there are mainly problems in the operational services of agricultural machinery equipment in China, such as poor practicality of fault automatic monitoring, imbalanced allocation of maintenance service resources, and high service scheduling costs. The current situation and characteristics of typical technologies related to the operation and maintenance service management of combine harvester at home and abroad were summarized, and focusing on the research status and development trends of remote operation and maintenance service platform for agricultural machinery equipment, operation data monitoring, multi machine collaborative operation, and operation and maintenance service optimization. Firstly, the overall architecture and technological progress of the operation and maintenance service management platform were elaborated. Then the research progress of remote operation and maintenance data monitoring technologies such as operating conditions, job quality perception, and vehicle terminal management was elaborated, as well as the research progress of multi machine collaborative operation technologies such as path planning, collaborative control, and task allocation. The research progress of operation and maintenance service optimization technologies such as fault diagnosis prediction, maintenance strategy, and group scheduling was also elaborated. Finally, combined with practical application scenarios, the challenges and opportunities faced by agricultural machinery equipment operation and maintenance service management technology were summarized and analyzed, and the future development direction of agricultural machinery equipment information management technology was proposed. Future research directions for informationization of agricultural equipment operation and maintenance service management in China were pointed out.

Abstract:A split and combined vibration seeding and leveling device was designed to address the problem of poor seeding performance of the vibrating seed metering device in the super rice seedling cultivation and sowing process that made it difficult to achieve precision seeding, and a vibration seed leveling control method based on image recognition was proposed. The single factor discrete element simulation analysis of the key structural parameters of the vibration plate was carried out. The results showed that the variation coefficient of the uniformity of the vibration plate and the variation coefficient of the supply uniformity at the exit of the vibration plate was increased with the increase of depth of box. With the increase of angle of steering groove, the depth of the seed storage box and the angle of the steering groove were determined to be 12mm and 48°, respectively. A seed stream image detection and control system was designed and built. The test of piezoelectric vibration monomer and homogenizing unit image detection and rectification showed that when the proportion of white low pixels in the detected image was less than 20%, after rectification, the proportion of white pixels could meet the design requirements. The results showed that the qualified rate was not less than 93.47% and the missed rate was not more than 1.00% when the operating voltage was 150~200V. The qualified sowing rate of the three rice seeds was not less than 94.17%, and the missing sowing rate was not more than 0.67%. The device can meet the requirements of precision sowing of super rice and had good adaptability to different super rice seeds.

Abstract:When the seed flow in the pneumatic centralized drainage system enters the distributor along the side wall of the seed pipe, the mixing uniformity of the gas and the seed is low, and the consistency of the displacement of each row is poor. The spiral booster pipe was proposed, and the overall structure and working principle were expounded. The design and calculation of the seed pipe and the spiral booster pipe were carried out. Taking buckwheat as the object, the force, motion, and pressure loss analysis of the seed group in the spiral booster pipe was carried out. Based on the CFD-DEM coupling simulation method, the effects of blade number, torsion angle, spiral booster tube length, and conveying airflow velocity on the consistency coefficient of variation of each row displacement of buckwheat seeds were studied by single factor test, steepest climbing test, and central composite design test. The simulation results showed that when the number of blades was 3, the length of the spiral booster tube was 210mm, the torsion angle was 383°, and the conveying airflow velocity was 29.30m/s, the consistency coefficient of variation of each row of buckwheat was 9.83%, which was optimal. The performance verification tests of different types of booster tubes were carried out on the test bench of the pneumatic collecting and discharging system. The results showed that the variation coefficients of the consistency of each row of the buckwheat pneumatic collecting and discharging system with spiral, corrugated, and eyelet booster tubes were 5.58%, 6.85%, and 9.65%, respectively. The difference between the simulation results and the bench verification test results under the optimal parameter combination condition was 4.25 percentage points. Compared with the traditional booster tube, the spiral booster tube proposed enhanced the uniformity of gas mixing, improved the operation quality of the pneumatic collection system, and provided technical support for the design of the pneumatic collection system.

Abstract:Aiming at the problem of low seedling picking success rate, high damage rate and poor stability of pepper transplanter, a kind of clamp-puncture end effector and pneumatic system was proposed based on the operation mode of taking seedlings in whole row and throwing seedlings at intervals. The structural configuration and working principle of device was analyzed and the kinematics and dynamic model was established. Based on the physical characteristics of pepper plug seedlings and analysis of boundary condition, the main parameters range of the newly improved end effector was determined. Simulation test based on ADAMS and FluidSIM-P 3.6 were carried out to verify the reasonability of the structure and response time sequence of pneumatic actuator. The coupling simulation test based on EDEM and RecurDyn turned out that the improved structure had a better performance. Aiming at the success rate of seedlings picking and the loss rate of substrate, single factor test and analysis were carried out on the key factors such as needle interval, needle insertion stroke and needle insertion angle. According to the results of single factor test, the orthogonal test, variance analysis and response surface analysis were carried out. The optimal parameters were obtained as follows: needle interval was 12.25mm, needle insertion angle was 50.37°, needle insertion stroke was 14.21mm, the seeding picking success rate, substrate loss rate were respectively 97.44% and 2.13%. Bench verification test showed that the success rate of seedling extraction was 97.92% and the matrix loss rate was 2.21%. The relative error between verification result and the predicted value was less than 5%.

Abstract:The shape of Cyperus esculentus seeds is irregular and the surface is uneven. Most of the domestic seed equipment of the seed ball is improved based on the large seed ball seed feeder. Due to the lack of accurate simulation parameters during the discrete element simulation experiment, this results in errors in the simulation and optimization analysis of the seed dispenser. To solve this problem, the intrinsic parameters and the contact parameters between seeds and seeding device were measured, and the seed particle filling model was constructed by reverse engineering technology. According to the results of simulation accumulation test of square shell and cylinder, Plackett-Burman test was used to screen out the interspecific static friction coefficient and interspecific rolling friction coefficient which had significant influence on population accumulation angle. A binary regression equation was established with these two significant influence factors as independent variables and population accumulation angle as response value. After solving the equations with Matlab, the calibrated interspecific static friction coefficient and interspecific rolling friction coefficient were 0.246 and 0.083, respectively. To verify the reliability of parameters after calibration, simulation verification test was carried out in EDEM, and the results showed that the error between simulation and physical accumulation angle of square shell device and physical accumulation angle of cylinder device was 0.43%, and the error between simulation and physical accumulation angle of cylinder device was 0.55%. To further verify the accuracy of parameters after calibration, the pneumatic precision seed disperser was used and the three key feature sizes and the ratio of stack crosssectional area were selected. After Cyperus esculentus seeds was stabilized, the simulation and physical test were compared, finally, the relative errors of the parameters H1 from the maximum convex point of the upper boundary of the population to the horizontal center line of the seed platen, H2 from the intersection point of the population profile boundary and the vertical line on the left side of the transparent observation window to the horizontal center line of the seed platen, and the ratio r of the accumulation crosssectional area were all within 8.33%.

Abstract:In response to the requirements of soil-covering with large width, thin thickness and high uniformity for mechanized soil-straw layered mulching technique in orchards, a supporting soil-covering device of orchard straw mulching machine was developed, which mainly consisted of a soil-taking screw, two soil-throwing wheels, a soil-collecting trough, and a soil-covering screw. The soil-taking screw with width of 1.5m provided a flat foundation for straw layer laying, ensuring a flat-covering of straw layer and uniformity of the thin soil layer. The soil-throwing wheels on both sides of the soil-taking screw threw the soil back into the soil-collecting trough, avoiding the stuck problem of the device caused by foreign objects such as gravel. With discrete element simulation, it was determined that the terminal double helix structure was adopted for the soil-taking screw, the number of the soil-throwing wheel blades was 10, and the blade deflection angle of the soil-throwing wheel was 10° to improve the soil-covering uniformity and reduce the power consumption. The field tests showed that the device had a soil-covering width of 1.5m and a thickness of 20.8~31.7mm. The coefficient of determination of the regression model for the standard deviation of the soil-covering thickness was 0.9342 with error of 1.8%. The optimal combination of operating parameters with the standard deviation of the soil-covering thickness as the response value was decided as the rotation speed of the soil-throwing wheel was 300r/min, the soil-throwing angle was 40°, and the soil-taking depth was 20mm. Under the combination, the soil-covering thickness was 22.9mm with standard deviation of 3.7mm, which met the requirements of soil-covering with thin thickness and high uniformity. The research realized the mechanization of the orchard soil-straw layered mulching agronomy and provided a method and equipment for orchard management in arid areas.

Abstract:Due to the lack of management, Camellia oleifera forests would encountered problems such as soil compaction and mixed growth of grass and shrubs. In order to achieve integrated operations of soil loosening, weeding, and clearing shrubs in low and intensive Camellia oleifera forests, a vertical spiral machine for soil loosening and weeding was developed. The vertical spiral cutter was designed by analyzing the physical and mechanical characteristics of common weeds and miscellaneous shrubs roots in Camellia oleifera forests, and combined with motion analysis of cutting tools. The root-soil composite model was established by using Hertz-Mindlin bonding contact model, with the goal of reducing power consumption, the parameters of the cutting tool were simulated and optimized, and the disturbance of root-soil composite were analyzed by the optimized cutter. In addition, an actual cutter was manufactured and the soil bin experiment was conducted, and it verified the reliability of the simulation results. The simulation results showed that the disturbance of soil-root complex by the optimized vertical spiral cutter can reach 91.41%, the maximum power consumption when cutting soil alone was 0.16kW and the maximum power consumption when cutting the root-soil composite was 0.77kW. The soil bin experiment results showed that the coefficient of stability of the plowing depth was 92.34%, the soil fragmentation rate was 95.00%, and the root removal rate was 95.30%;the maximum torque measured in the case of cutting soil alone was 7.93N·m and it was 6.57% lower than the simulation value; the actual measured maximum torque was 34.84N·m in the case of cutting the root-soil composite, and it was 4.91% lower than the simulation value; the correctness of the design was verified by the close agreement between the experimental and simulation values. The vertical spiral soil loosening and weeding machine met the requirements for soil loosening, weeding, and irrigation operations in Camellia oleifera forest in hilly and mountainous areas.

Abstract:For the large monopoly wide potato cultivator due to the width of the larger, equipment structure was huge and the potato biologica morphological characteristics of the longitudinal and horizontal stability of the potato poor, serious deflection, pressure seedling injury and other problems,the optimized design and test of the large monopoly wide potato cultivator were carried out. Through kinematic analysis of the potato cultivator, the main factors affecting the longitudinal and lateral stability of the tiller was determined and the design of its components was optimized, the EDEM discrete element simulation software was used to conduct a secondary orthogonal rotation simulation, test with the horizontal distance from the mulching share to the hinge point of the parallel four-bar and the frame, the lateral width of the parallel four-bar and the operating speed as the test factors, and the monopoly straightness and the monopoly height stability coefficient as the evaluation indexes. Using Design-Expert 11.0 software to optimize and analyze the results of the simulation test. When the horizontal distance from the mulching share to the hinge point of the parallel four-bar and the frame was 702.7mm,the lateral width of the parallel four-bar was 255.4mm and the operating speed was 1.29m/s, the monopoly straightness of the cultivator was 2.39cm and the stability coefficient of the monopoly height was 94.12%.The optimized data were verified in field tests, results showed that the straightness of monopoly direction was 2.4cm, the pass rate of cultivation was 95.7%, the injury rate was 1.5%, the stability coefficient of monopoly height was 94.2%, and the height of cultivation was 7.8cm. The optimized large monopoly wide potato cultivator was more reasonable, with good working effect,which had better longitudinal stability and lateral stability than the pre-optimized large monopoly wide potato cultivator.

Abstract:Soil cultivation, the part and parcel of crop farming, can provide a good peripheral environment for tobacco growth. In order to solve the problems of traditional soil cultivation's operation, including the lack of centralized throwing soil and the poor effect of fixed-point covering soil at tobacco fields in southern China. The soil-banking machine with centralized soil feeding and double-sided lateral delivery was designed for tobacco based on the mode of breaking up the soil after rotary tilling, centralizing throwing soil and soil lateral delivery. The soil engaging component was researched and the process about centralized soil throwing and lateral delivery on the basis of the theory of soil milling and soil auger conveying was analyzed, and the main structure and operation parameters of the equipment was confirmed. Based on the EDEM discrete element simulation technology, a component-soil simulation model was established. The forward speed, rotary tillage depth, and rotation speed of the auger shaft were the test factors. The soil height, soil width, and soil fetching rate were used as test indicators, and the simulated quadratic orthogonal rotation regression test was carried out. Field experiments were conducted on the basis of simulation. On the basis of simulation test values, the test factors were optimized and analyzed. It illuminated that when the forward speed was 0.4m/s, the rotary tillage depth was 60mm and rotation speed of the auger shaft was 339r/min, and the soil height was 27.76mm, the soil width was 132.94mm, and the soil fetching rate was 23.66%. The field test results showed that the soil at the trench was loosened and cultivated soil working was stable, and its soil height was 24.37mm, soil width was 130.42mm, and soil fetching rate was 21.19%. The deviation of the test index was respectively 12.21%, 1.90% and 10.44%, and the research result would provide a way to strengthen integration of agricultural machinery and agronomy about tobacco soil cultivation technology.

Abstract:In response to the problems of high potato damage rate, high peeling rate, long transportation distance, and large volume of existing potato combine harvesters, combined with the harvesting mode in the main potato production areas of northern China, circular loss reducing and collecting potato lifting device suitable for potato lifting operations was designed based on bagging potato combine harvesters. On the basis of expounding the overall structure and working principle, combined with the analysis of potato kinematics model and impact characteristics, it was found that the main factors affecting the lifting efficiency and potato damage of the device were the height of the lifting baffle, the angle between the lifting baffle and the lifting belt, the distance between two adjacent lifting baffles, and the running speed of the lifting belt, a simulation model of potato and circular loss reducing and collecting potato lifting device was constructed through DEM-MBD coupling, and single factor and multi factor experiments were conducted on the influencing factors. The optimal parameter combination for the circular loss reducing and collecting potato lifting device was obtained through parameter optimization: the height of the lifting baffle was 199.21mm, the angle between the lifting baffle and the lifting belt was 75.86°, and the distance between adjacent lifting baffles was 240.35mm. The results of the inter bench test showed that when the loading rate was 24t/h and the conveyor belt speed was 0.8m/s, 1.0m/s and 1.2m/s, the average values of the five collision acceleration peaks collected by the electronic potato were 636.63m/s2,593.29m/s2, 685.63m/s2, with skin breaking rates of 1.13%, 1.06%, 1.21%, respectively. The collision acceleration peaks were all smaller than the critical damage threshold of the potato. Field experiments showed that at operating speeds of 0.6m/s, 0.7m/s and 0.8m/s, the potato damage rates were 0.94%, 1.06% and 1.12%, and the skin breaking rates were 1.09%, 1.21% and 1.33%, respectively, the circular loss reducing and collecting potato lifting device operated normally without any phenomenon of potatoes falling. The coordination of various components ensured that the bagging type potato combine harvester was in an efficient and stable operating state.

Abstract:Aiming at the randomness of sugarcane growth, which causes the feeding amount of the harvester to vary from large to small, which can easily lead to blockage or low efficiency in the transportation of the whole rod sugarcane harvester, a conveying control system for the whole rod sugarcane harvester was designed. By building a sugarcane harvester experimental platform, based on dynamic torque sensors, PLC control system, hydraulic system and servo motor system, real-time adjustment was made to match the conveying capacity of the sugarcane harvester with the feeding amount, reducing sugarcane blockage. The results of the sugarcane conveying control experiment showed that after installing the conveying control system, the average conveying speeds of sugarcane at each experimental level were 4.06m/s, 3.42m/s, 3.04m/s, and 2.42m/s, which were significantly improved compared with those without the control system. Moreover, the conveying capacity of the harvester (the walking speed of the harvester and the speed of each conveying roller) returned to its initial value after the peak feeding rate, ensuring the efficiency of field harvesting work. The average blockage rate of the regulation experiment was reduced to 5%, and the conveying regulation system had a significant effect on alleviating conveying blockage.

Abstract:Camellia oleifera is a unique woody oil tree species in China, which plays an important role in grain and oil security. However, there are problems with harvesting of Camellia oleifera fruits in hilly and mountainous areas such as low efficiency in manual harvesting, difficulties in flower and fruit synchronization, and difficulties in large-scale mechanical, limiting the development of mechanized harvesting of Camellia oleifera fruits. A handheld impacting combtype Camellia oleifera fruit harvester was designed. The impact component of the device was provided with a buffer sleeve and an angle adjustment mechanism. The harvesting of Camellia oleifera fruit was completed by using the collision and brushing action of the impact finger with the Camellia oleifera fruit. The goal was to minimize the quality of harvester. The topology optimization section of Ansys Workbench 19.2 software was used to lightweight design the harvester. After optimization the weight reduction of the rack was nearly 30.59%. Meanwhile, the “impact finger-Camellia oleifera” collision model and “main branch-secondary twig-fruit” three pendulum dynamic model were established. By analyzing, it can be seen that the main factors affecting the abscission of Camellia oleifera fruit were the mass and pressing deformation of Camellia oleifera fruit, the quality and rotation speed of the impact finger, and the length and quality of the branches after collision. The rotational speed of the impact finger, the number of comb strokes, and the angle between the impact fingers were taken as experimental factors. The rate of harvesting efficiency and net harvesting rate of camellia fruits as well as the rate of flower bud abscission were calculated under different experimental conditions. The experiment results were analyzed by response surface analysis. The results indicated that the performance of the picking device was best when the number of combs was 4.5, the rotating speed of the impact finger was 409.8r/min, and the impact finger included angle was 4.1°. In this case, the picking rate and net picking rate of Camellia oleifera fruit were 43.67kg/h and 86.42%, and the falling rate of bud was 8.89%, which can meet the working requirements of high net picking rate and low damage rate of flower buds.

Abstract:In order to improve the control performance, work quality, and reduce energy consumption of the hay cutter cutting operation, an objective function was established based on linear predictive control and combined with the characteristics of the hay cutter cutting operation. The sampling period was derived from the cutting kinematic error model to solve the control robustness, and the control time domain and prediction time domain were derived from the cutting dynamics model to improve control responsiveness. A load controller for the hay cutter cutting roller was designed. Simulink simulation showed that when the prediction parameter group (sampling period, prediction time domain, control time domain) selected through calculation and regression optimization was 0.8s, 15s, and 2s, the control accuracy and robustness were the best, the operation ability was the highest, the response speed to disturbances was the fastest, the suppression ability was the strongest, and the energy consumption was the lowest (9.27×106J). The on-site test results showed that the optimized predictive parameter group model cont-roller can effectively track and control the cutting load of the hay cutter, and the product quality met the standard requirements. At the same time, it improved the operation capacity of the hay cutter, making the system control response faster, production efficiency higher, and unit operation energy consumption smaller (1.382×107J). The method for establishing the parameter model of this controller provided a reference for the design of the control system of a generic forage crop harvester.

Abstract:With the problems of severe tool wear and lack of fault monitoring device, leading to the failure of the work during the working process of stalk chopping, an intelligent monitor system which can be mounted on the returning stalk chopping machine was designed. Taking STM32 microcontroller as the main controller, multiple sensors fusion technology was applied, and tool wear condition monitoring was realized based on machine learning algorithm. In order to solve the problem of difficult extraction of nonlinear feature signals of straw crushing tool wear, a method of tool wear monitoring IBOA-SVM integrating improve butterfly optimization algorithm (IBOA) and support vector machine (SVM) was proposed. The monitoring method used the rotational speed, left side vibration frequency, and right side vibration frequency of the crushing knife roll as input eigenvectors to the model, and the wear condition of the tool (normal, worn and lost) as outputs. Compared with the unoptimized SVM algorithm, the identification accuracy of tool wear condition was improved from 95.61% to 98.83% by optimizing the parameters of the SVM algorithm with the IBOA algorithm. In order to verify the effectiveness of the IBOA-SVM model, the repeated comparison experiments of multiple models were conducted under the same parameter setting environment, which showed that the average value of the recognition accuracy of the IBOA-SVM model was improved and the accuracy of a single trial was maintained at a high level as compared with the five models of SVM, PSO-SVM, WOA-SVM, BOA-SVM and CWBOA-SVM. The IBOA-SVM model was embedded into the monitoring system and field test was conducted, in which it was shown that the designed tool wear condition monitoring system had good performance both in terms of recognition accuracy and robustness.

Abstract:In the process industry, there are many cases where the residual energy liquid has a high-pressure head and low flow rate, it is necessary to use the multistage guide vane centrifugal pump as turbine (PAT) to recover the liquid residual energy. To adapt to the impact of production regulation on the performance of pumps as turbines in the process industry, multistage PAT is required to have a small variation in output power with the flow, i.e., a flatter output power-flow curve. Based on the Euler equation and according to the retention of the velocity distance, the relationship between the output power and geometrical parameters of the guide vane PAT was derived, the output power was related to the geometrical parameters of the guide vane (throat area, outlet angle, guide vane number, and base circle diameter) and the geometrical parameters of the impeller (blade outlet diameter, outlet width, blade outlet angle, and blade number). Using mathematical derivatives, the influence of geometric parameters of the guide vane on the flatness of the output power curve was determined. A two-stage PAT was used as the object research, and the research scheme was designed by changing the geometric parameters of the positive guide vane, which were experimented with and simulated by Fluent software. The computational results were consistent with the theoretical derivation, the output power curve can be flattened by appropriately increasing the positive guide vane throat area, positive guide vane outlet angle, positive guide vane blades number, or appropriately reducing the positive guide vane base circle diameter; at the design condition point, the order of influence of positive guide vane geometric parameters on the slope of the power curve was throat area, outlet placement angle, blade number, base circle diameter, and the slope of the power curve was reduced by 0.17, 0.11, 0.05 and 0.03, respectively; a proper increase in throat area can lead to an increase in multistage PAT efficiency by 1.65 percentage points and a shift in the high-efficiency point toward high flow rates.

Abstract:Extracting cropland accurately and efficiently from high-resolution remote sensing images is of great significance to agricultural production and agricultural resource investigation. Cropland with different areas, ground covers and cultivation types in remote sensing images have large differences in features, whereas the insufficient generalization ability of traditional supervised learning models also leads to poor extraction of heterogeneous cropland with the above features. To solve this problem, the semi-supervised semantic segmentation with mutual knowledge distillation (SSS-MKD) model as the base model and incorporated an online hard example mining strategy based on a weighted loss function. The proposed model was evaluated on the Vaihingen dataset and achieved the highest overall accuracy of 87.1% and an average F1 score of 85.0%, The model had the best extraction accuracy compared with other semi-supervised models. In addition, for the task of large-area cropland extraction, the feature information of the unannotated images that were heterogeneous and homogeneous with the annotated images were added to the training of the semi-supervised learning model by designing two sets of experiments using Jilin-1 cropland image dataset, respectively, in order to improve the cropland extraction accuracy in the proposed extraction area. The experimental results showed that the proposed model could achieve the highest overall accuracy of 84.0% by using the cropland images to be extracted for assisted training. In addition, by using unlabeled images with strong similarity to the cropland in the target region, the overall accuracy could be further improved by 2.1~6.1 percentage points. The maximum overall accuracy achieved using unlabeled images with strong similarity to the cropland features in the training set was 81.6%, which was 6.6~8.5 percentage points higher than the accuracy achieved by using conventional supervised learning. Taking Xian County area in Hebei Province as an example, the model used the Jilin-1 cropland image dataset (part 1) as the labeled training set, and achieved the highest accuracy of 88.7% overall accuracy after training with both the Jilin-1 cropland image dataset (part 2) and the unlabeled images from the Xian County areas Gaofen-2 image dataset, compared with the extraction accuracy by using these two datasets alone and without the unlabeled images, respectively 1.0 percentage points, 4.9 percentage points, and 8.1 percentage points. Under the realistic background of abundant cropland remote sensing data and insufficient corresponding labeled information. The semi-supervised model learned the feature information in the images of cultivated land in the proposed extraction area and the training set area, and improved the extraction accuracy of heterogeneous cultivated land. A method that maximized the learning of heterogeneous cropland unlabeled images and a minimum amount of labeled data using a semi-supervised learning model for the task of wide-area cropland extraction was proposed, and the extraction results were effective.

Abstract:Yield prediction models can be improved by better integration of data and algorithms, and the accuracy of yield prediction can be further improved by incorporating other factors such as those affecting yield into the model. The research situation was addressed that the wheat-maize rotation system lacked the direct incorporation of the previous crop information into the yield prediction and management of the seasonal crop, a multi-temporal and multimodal crop yield prediction model based on GPR was established by using remote sensing information of the growing season and yield information of the previous maize crop, fusing multi-temporal and multimodal data such as remote sensing information of wheat growing season at the jointing stage, filling stage and maturity stage, fertilization information before sowing and soil properties. The results showed that the performance of the yield prediction model based on the multiple growth periods was improved compared with that based on the single growth period, in which the decision coefficient R2 of the yield prediction model was improved by 0.01~0.03. The accuracy of the yield prediction model based on the spectral indexes of wheat growing season at the jointing stage was higher than the accuracy of the yield prediction model based on the spectral indexes of wheat growing season at the filling stage, and the accuracy of the yield prediction model based on the spectral indexes of wheat growing season at the maturity stage was the lowest, and the accuracy of the yield prediction model based on the jointing stage was slightly lower than that of the yield prediction model based on the multiple growth periods, but the accuracy was similar. In the yield prediction models based on the multimodal parameters fusion, the yield prediction models based on two-modal parameters fusion had higher accuracy than the unimodal yield prediction models, except for the yield prediction model constructed by fusing maize information with soil properties. The accuracy of the yield prediction models with four-modal parameters fusion and three-modal parameters fusion was higher than that of the corresponding yield prediction models with low-modal parameters fusion. The GPR model with four-modal parameters fusion had a decision coefficient R2 of 0.92 and RMSE of 213.75kg/hm2, which improved R2 by 0.02 to 0.41 compared with the wheat yield prediction models based on other modalities. For wheat yield prediction models based on multimodal parameters fusion, from large to small, the influence of each modal parameters was as follows: fertilization information, wheat remote sensing information, soil properties information, maize crop information. Maize crop information had the least improvement in the accuracy of the yield prediction models based on the multimodal parameters fusion, which improved R2 by 0.02~0.07. Maize crop information characterized the soil fertility condition of post-harvest to a certain extent, and it was a high spatial resolution supplement to soil properties information, which could further improve the ability to quantify soil fertility, then combined with other parameters, they can improve the accuracy of wheat yield prediction. In conclusion, the research result provided a scientific basis and method for the comprehensive utilization of soil-crop data and the comprehensive management of wheat-maize rotation system.

Abstract:Machine learning models have been applied for monitoring crop growth condition and estimating crop yield, it is difficult to understand the internal mechanisms of complex models. In order to estimate crop yields accurately and make understandable explanations at the same time, LightGBM was used to develop yield estimation models of winter wheat in the Guanzhong Plain, PR China by using vegetation temperature condition index (VTCI), and interpretable methods such as local interpretable model-agnostic explanation (LIME), submodular pick-LIME, partial dependence plot (PDP), and individual conditional expectation (ICE) at global and local scales were used for further interpretations of the yield estimation models. Compared with other models, the results of LightGBM optimized by grid search showed that the R2 between the estimated and official yield records of winter wheat was 0.32, the RMSE was 809.10kg/hm2, and the MRE was 16.55%, which reached the extremely significant level (P<0.01), indicating that the model had high prediction precision and strong generalization ability. The interpretability of the experiments showed that the model can extract the knowledge in the data. In global interpretation, VTCI at the jointing stage for yield formation was the most important, followed by VTCI at the heading to filling stage and VTCI at the dough stage, and VTCI at the turning green stage had the least effect, which were consistent with prior knowledge. In local interpretation, based on the spatial characteristics of winter wheat yield that was high in the west and low in the east, the local interpretable methods further provided the reasons for the differences in the yield formation of different counties (districts), which provided references for field management in the Guanzhong Plain, PR China. These methods had application value for increasing and stabilizing the yield of winter wheat.

Abstract:Previous research on the extraction of winter wheat distribution information has mostly relied on single phenological periods or individual vegetation indices, neglecting the characteristics of different phenological periods and their interconnections, which has limited the classification accuracy. To enhance the extraction accuracy, a method for winter wheat identification was proposed based on corresponding feature indices for the sowing period, overwintering period, growth period, and maturation period. The method was applied to extract the winter wheat area in Jiaozuo City in 2020. By comparing the results under different phenological periods and classification methods, the findings indicated that the inclusion of the overwintering period led to varying degrees of improvement in overall accuracy and Kappa coefficients for both random forest and support vector machine classification methods, with respective reductions in root mean square error (RMSE) by 19.3% and 9.8%. The error percentage in winter wheat area extraction was reduced by 8.64 percentage points and 4.42 percentage points, respectively. Among different classification methods, random forest outperforms support vector machine and minimum distance in terms of overall accuracy and Kappa coefficient. Compared with support vector machine, random forest classification reduced RMSE by 19.6%. When compared with single feature indices, the overall accuracy and Kappa coefficient of the multi-phenological feature index based on random forest were higher, with RMSE of 1.84×103hm2, representing 33.6% reduction compared with single feature indices and 7.14 percentage points decrease in the error percentage for winter wheat area extraction.

Abstract:The accuracy of the data captured by UAV multispectral remote sensing for winter wheat yield prediction is still not high, and in order to guide the accurate prediction of winter wheat yield at the field scale, a high-precision winter wheat yield estimation model needs to be constructed. The corrected near-ground hyperspectral data (acquired by Field-Spec 3 analytical spectral devices, ASD) was used to verify the low-altitude UAV multispectral remote sensing data (acquired by DJI Phantom 4 multispectral camera, P4M), and the vegetation index calculated by the UAV multispectral image was combined with empirical statistical methods, and unvariate regression and multiple linear regression were used to estimate yields based on a single vegetation index and the combination of multi-vegetation index at the panicle stage, flowering stage and filling stage, respectively. Among them, the combination of multi-vegetation index included the normalized difference vegetation index (NDVI), the optimized soil adjusted vegetation index (OSAVI), the green normalized difference vegetation index (GNDVI), the leaf chlorophyll index (LCI) and the normalized difference red edge index (NDRE). The results showed that the winter wheat yield estimation model based on a single vegetation index had the highest accuracy, while the multiple linear regression model based on five vegetation indices had better fitting effect than the single vegetation index model in the three growth periods. Univariate or multiple regression models fit best during the spike extraction period. The coefficients of determination (R2), root mean square error (RMSE) of the modeling set of winter wheat based on the GNDVI index of the univariate quadratic regression yield estimation model were 0.69 and 428.91kg/hm2, respectively, and the R2, RMSE and relative root mean square error (RRMSE) of the validation set were 0.76, 418.14kg/hm2 and 11.56%, respectively. The R2 , RMSE and RRMSE of modeling set of the multiple linear regression yield estimation model based on the combination of five vegetation indices were 0.80, 340.14kg/hm2, and the R2, RMSE and RRMSE of the validation set were 0.69, 466.75kg/hm2 and 12.90%, respectively. In summary, the data captured by the P4M had broad application prospects in estimating winter wheat yield. The optimal model for winter wheat yield estimation was a multiple linear regression model based on the combination of multiple vegetation indices at the ear pumping stage.

Abstract:Leaf potassium content (LKC) is an important indicator to characterize the potassium nutritional status of crops, and efficient and accurate acquisition of potato LKC can help precision agriculture fertilization management. The aim was to improve the accuracy of potato LKC estimation by combining vegetation indices (VIs) and vegetation cover (FVC) extracted from RGB images during the critical fertility period of potatoes. Firstly, VIs and FVC were extracted from the RGB images of potato tuber formation stage (S1), tuber growth stage (S2), and starch accumulation stage (S3). Then the correlation between VIs and FVC and potato LKC was analyzed for each fertility period separately. Finally, the correlation between VIs and FVC, and LKC was analyzed by using a support vector machine (SVM), least absolute shrinkage and selection operator regression (Lasso), and ridge regression used to construct the estimation model of potato LKC. The results showed that the accuracy of potato FVC extracted based on RGB images was high, and the first two fertility periods were higher than that of the third; the estimation of potato LKC using VIs was feasible, but the accuracy needed to be further improved; and the combination of VIs with FVC can improve the estimation accuracy of potato LKC. The research result can provide technical references for crop growth and potassium nutrient status monitoring.

Abstract:Canopy information is an important element of tea field management and an important basis for the design of related equipment. Aiming at the traditional methods of obtaining crop canopy information, which are time-consuming, subjective and prone to damage, a method of obtaining and estimating the height and outline of the tea tree canopy was proposed. Firstly, the point cloud data of the tea field was collected from multiple sites by 3D LiDAR, and the original point cloud was pre-processed with attitude correction, ROI selection, alignment, noise reduction, and elevation normalization to obtain the elevation-normalized tea tree point cloud. Secondly, the canopy height model (CHM) of tea trees was generated by inverse distance weight (IDW) and triangulation irregular network (TIN) at different spatial resolutions, among which, the CHM of tea trees generated by IDW at 0.05m spatial resolution had better interpolation accuracy and the model produced relatively fewer pits. Finally, the raster values of CHM were extracted from 21 percentiles between 90 and 100 as the canopy height of tea trees and compared with the measured values. The results showed that the estimated value was most accurate when the percentile was 98.5, and the correlation coefficient with the true value was 0.88, with an average absolute error of 3.17cm, and a root mean square error of 4.16cm. In addition, totally 20 canopy section point clouds were extracted from the elevation-normalized tea tree point clouds and their outlines were fitted by elliptic, Gaussian and quadratic polynomial models, respectively. The results showed that the quadratic polynomial model could better reflect the characteristics of the tea tree canopy outline, and the mean value of the average minimum distance between the points and the fitted curves was 2.60cm with a variance of 0.21cm2. The research can provide theoretical support for the modern management of tea fields and the design of related equipment.

Abstract:To address the challenges of mutual occlusion and accurate differentiation between rice and weeds in real-world environments, an improved method for rice-weed recognition was proposed based on DeepLabv3+. The research focused on images of rice field weeds captured by UAV in complex backgrounds, the MobileNetv2 was used as the backbone feature extraction network to reduce the number of parameters and computational complexity of the model; channel and spatial dual-domain attention modules were integrated to strengthen the model's attention to important features. A multi-branch receptive field cascade fusion structure was proposed based on dense sampling to improve the ASPP module to expand the sampling range. In addition, improvements to the decoder were made. Experimental results demonstrated that the improved model achieved the best performance in rice-weed recognition, with a mean intersection over union (MIoU) of 90.72%, mean pixel accuracy (mPA) of 95.67%, and F1_score of 94.29%, which were 3.22, 1.25, and 2.65 percentage points higher than that of the basic model. The improved model had a size of 11.15MB, which was 1/19 of the original model's size, and achieved an average network inference speed of 103.91 frames per second per image. The results demonstrated that the improved model can accurately segment rice and weeds in complex backgrounds, supporting precise pesticide application using UAV.

Abstract:Rice grain detection plays an important role in grain storage, directly affecting the price of grain sales. In response to the problems of difficult recognition, large network model parameters, slow detection speed, and high cost of general machine vision detection algorithms in dense scenes with small rice grain targets, a rice grain detection algorithm was proposed based on YOLO v7 optimization. Firstly, some efficient layer aggregation network (ELAN) modules were replaced with lightweight network module GhostNetV2 and added them to the backbone and neck network sections, achieving precise simplification of network parameters while reducing feature redundancy in channels. Secondly, the attention module (ACmix) that combined convolution and self attention was added to the MP module, balancing global and local feature information and fully paying attention to the details of feature mapping. Finally, wise intersection over union (WIoU) was used as the loss function to reduce penalty term interference such as distance and aspect ratio. The design of monotonic focusing mechanism improved the positioning performance of the model. The improved model detection level was verified on the rice grain image dataset, and the experimental results showed that the improved YOLO v7 model was high, mAP@0.5 was up to 96.55%, mAP@0.5:0.95 reached 70.10%, and the training model parameters were also decreased. In practical scenarios, the effect of rice impurity detection with a dark black background was better than other models, meeting the real-time detection requirements of rice grains. This algorithm can be considered for application in automated grain detection systems.

Abstract:Accurate identification of multi-category targets in tomato images is the technical premise for automatic picking. Aiming at the problems of low segmentation accuracy and the large number of model parameters in existing networks, a multi-category segmentation method based on improved DeepLabv3+ was proposed for tomato images. The method used GhostNet and coordinate attention (CA) to construct CA-GhostNet as the backbone feature extraction network of DeepLabv3+, reducing the number of parameters in the network. And a multi-branch decoding structure was designed to improve the segmentation accuracy of the model for small target categories. Then the weight parameters of the synthesized dataset were used for migration training based on the single and binocular small sample dataset. Eight semantic categories such as fruit, trunk, branch and thin line were segmented. The results showed that mean intersection over union (MIoU) and mean pixel accuracy (MPA) of improved DeepLabv3+ model were 68.64% and 78.59% on the monocular dataset, respectively. The MIoU and MPA were 73.00% and 80.59% on the binocular dataset. In addition, the memory occupation of the proposed model was only 18.5MB, and the inference time of a single image was 55ms. Compared with the baseline model, the MIoU on the monocular and binocular datasets was increased by 6.40 percentage points and 6.98 percentage points, respectively. Compared with HRNet, UNet and PSPNet, the memory occupation was reduced by 82%, 79% and 88%, respectively. The research result can provide reference for intelligent picking and safe operation of tomato picking robot.

Abstract:In response to the issues of a wide variety of crop types, poor resource information standardization, and low model training accuracy in China, focusing on seven crops: wheat, rice, maize, soybeans, cotton, peanuts, and rapeseed, using parameters like variety, morphology, yield, and quality as indicators, totally 83 crop variety entities were constructed. A manual annotation approach was adopted and an information extraction four-layer network model (BERT-PGD-BiLSTM-CRF) was introduced by incorporating adversarial training techniques. The model utilized the bidirectional encoder representation from transformers(BERT) model, based on a deep bidirectional transformer, as a pre-training model to acquire semantic representations of words and phrases. It employed projected gradient descent (PGD) adversarial training to introduce perturbations to the samples, thereby enhancing model robustness and generalization. Additionally, it leveraged a bidirectional long short-term memory (BiLSTM) network to capture long-distance text information and combined conditional random fields (CRF) to learn label constraint information. Comparing the training results with 18 different information extraction models, the research indicated that the proposed BERT-PGD-BiLSTM-CRF model achieved a precision of 95.4%, a recall of 97.0%, and an F1 score of 96.2%. This suggested that the BERT-PGD-BiLSTM-CRF model, utilizing adversarial training techniques, was effective in extracting crop variety information and also provided a technological reference for agricultural information extraction.

Abstract:The RGB-D SLAM augmented reality log measurement system was constructed, which used a smart phone embedded with a ToF camera, RGB camera and IMU as the hardware system, and used the depth map and pose acquired by RGB-D SLAM technology as the data source. Specifically, the method for online estimating the pixel depth of RGB images was designed based on ToF images in order to preliminarily estimate the position of log end faces; secondly, a denoising algorithm based on discrete partitioning and a log end face curvature estimation algorithm were designed for precisely filtering the log end point cloud and evaluating the reliability of the filtering results; then, the PCA algorithm was used to estimate the length and diameter direction vector of the log, which was used to estimate the value of the length and diameter of the log; and finally, the algorithm was used to build a log measurement system on the mobile phone platform, so as to realize the online measurement of the log diameter by using the smart phone, and the online supervision of the measurement results by using the augmented reality scene. The system was tested by measuring 334 logs in six regions to evaluate the measurement accuracy. The results showed that the bias and root mean square error (RMSE) of the log diameter estimates were -0.13cm (-0.35%) and 1.05cm (3.34%) respectively; the bias and RMSE of the log stepping diameter estimates were -0.10cm (-0.22%) and 1.33cm (4.43%) respectively; the bias and RMSE of the log volume estimates were -0.007m3(-0.27%) and 0.0939m3(7.23%) respectively; the absolute value of the relative error of the volume of log pile was no more than 2.23%; and the error of the total volume of all log piles was -0.68%. Obviously, no matter from the point of view of a single log or a pile, the measurement results were unbiased and high-precision, which meant that the new log measuring system was a potential highprecision, high-robust real-time log measuring potential solutions.

Abstract:Cucumber downy mildew is caused by the spores of cucumber downy mildew from Cuba through infection, which seriously affects the quality and yield of cucumber. The number of the spores is closely related to the severity of the disease. Accordingly, there urgently needs to establish a rapid, simple and efficient quantitative detection method for the spores of cucumber downy mildew, in order to explore the way forward to achieve the control of cucumber downy mildew. Based on YOLO v5 model, an exploratory model was proposed for quantitative detection of cucumber downy mildew spores by Faster-NAM-YOLO. Firstly, a feature extraction module C3_ Faster was proposed, which was used to replace the C3 module in YOLO v5, which effectively reduced the calculation amount of model parameters and the depth of the model, and also improved the detection speed and accuracy of cucumber downy mildew spores. Secondly, the NAMAttention module was added to the backbone network and also improved the model's feature extraction ability and computational efficiency by applying weight sparsity penalty to suppress insignificant weights. In the end, the quantitative detection of the spores caused by cucumber downy mildew was realized. Faster-NAM-YOLO model on the test set mAP@0.5 and mAP@0.5:0.95 reached 95.80% and 60.90%, respectively, to compare with the original YOLO model. It can be seen that the final results were increased by 1.80 percentage points and 1.20 percentage points, respectively, reducing the model size and FLOPs of the original YOLO v5 model by 5.27M and 1.49×1010, respectively. It was found that Faster-NAM-YOLO had significant advantages in detection accuracy, model size, FLOPs, and inference time compared with single stage target detection models such as YOLO v3, THP-YOLO v5, YOLO v7, YOLO v8, Faster RCNN, and SSD. In addition, under the comparison for the three different resolution scales of 1200 pixels×1200 pixels, 1500 pixels×1500 pixels and 1800 pixels×1800 pixels, as well as the different specifications and the varied amount of images, which suggested that the Faster-NAM-YOLO model was further validated to have strong robustness and generalization ability. The research result not only provided a more accurate basis for early online monitoring of cucumber downy mildew, but also laid a foundation for further exploring the relationship between the dynamic changes of spore number and morphological characteristics and the severity of the disease.

Abstract:To achieve early detection of crop diseases, a crop disease early detection model was proposed based on infrared thermal imaging and improved YOLO v5. The CSPD-arknet was used as the main feature extraction network, and the YOLO v5 stride-2 convolution was replaced by the SPD-Conv module, which were respectively the five stride-2 convolution layers in the main network and the two stride-2 convolution layers in the Neck. This can improve its accuracy while maintaining the same level of parameter size and outputting three different scales of feature layers in the downstream stage. In order to enhance the interdependence between modeling channels, channel feature responses were adaptively recalibrated and SE mechanism was introduced to enhance feature extraction ability. In order to reduce model calculation and improve model speed, SPPF was introduced. After testing, the improved YOLO v5 algorithm had the best detection performance with an mAP of 95.7%, which was respectively 4.7 percentage points, 8.8 percentage points, 19.0 percentage points, and 3.5 percentage points higher than that of YOLO v3, YOLO v4, SSD, and YOLO v5 networks. Compared with the improved network before improvement, it also improved the detection of crop diseases under different temperature gradients. The mAP of five gradients were 91.0%, 91.6%, 90.4%, 92.6%, and 94.0%, which were higher than those before improvement by 3.6 percentage points, 1.5 percentage points, 7.2 percentage points, 0.6 percentage points, and 0.9 percentage points, respectively. The size of the improved YOLO v5 model was 13.755MB, which was lower than 3.687MB of the basic network before the improvement. The results showed that improving YOLO v5 can accurately and quickly detect early diseases, which can provide certain technical support for the development of early disease detection instruments.

Abstract:Accurate detection of tea diseases is crucial for a high yield and quality of tea, thereby increasing production and minimizing economic losses. However, tea disease detection faces several challenges, such as variations in disease scales and densely occluded disease areas. To tackle these challenges, a novel method for detecting tea diseases called multi-scale guided self-attention network (MSGSN) was introduced, which incorporated multi-scale guided self-attention. The MSGSN method utilized a VGG16-based module for extracting multi-scale features to capture local details like texture and edges in tea disease images across multiple scales, effectively expressing the local multi-scale features. Subsequently, the self-attention module captured global dependencies among pixels in the tea leaf image, enabling effective interaction between global information and the disease image's local features. Finally, the channel attention mechanism was employed to weight, fuse, and prioritize the multi-scale features, thereby enhancing the model's ability to characterize the multi-scale features of the disease and improving disease detection accuracy. Experimental results demonstrated the MSGSN method's superior detection performance in complex backgrounds and varying disease scales, achieving an accuracy rate of 92.15%. This method served as a valuable reference for the intelligent diagnosis of tea diseases. In addition, the method can provide a scientific basis for the prevention and control of tea diseases and help farmers take timely and effective control measures. At the same time, the method can also provide technical support for the development of the tea industry.

Abstract:To systematically analyze the effects of nitrogen application on cotton yield and water use efficiency under different climatic conditions, soil foundation conditions, and farmland management measures, totally 103 published Chinese and English literatures from China were collected, among which 37 papers were selected, and a total of 301 sets of yield and 127 sets of water use efficiency data were obtained. Based on the Meta-analysis method, the impact of nitrogen application on cotton yield and water use efficiency under different production conditions was quantitatively analyzed, and partial correlation analysis was used to identify the main influencing factors of cotton yield and water use efficiency under nitrogen application conditions. The results showed that with no nitrogen application as the control, nitrogen application can significantly improve the yield and water use efficiency of cotton. The effect of nitrogen application on yield and water use efficiency was most significant in areas with an average annual precipitation of 200~500mm, with effects of 34.02% and 54.15%, respectively. The effect of nitrogen application on yield and water use efficiency was increased with the increase of sunshine hours. When the soil pH value was 6~8, nitrogen application had the most significant effect on improving cotton yield and water use efficiency, with the effective amounts being 28.52% and 24.59%, respectively. The effect of nitrogen application on yield and water use efficiency in different soil textures was the highest in sandy soil, with 46.71% and 26.29%, respectively. As the frequency of fertilization was increased, the effect of nitrogen application on cotton yield and water use efficiency was gradually increased. The effect of nitrogen application on yield was increased with the increase of irrigation amount, and the effect on water use efficiency showed a trend of first increase and then decrease with the increase of irrigation amount. When the nitrogen application rate was 300~450kg/hm2, nitrogen application had the most significant effect on improving yield and water use efficiency. When the planting density was 1.5×105~2.5×105 plants/hm2, nitrogen application had the most significant promoting effect on yield and water use efficiency, with effective amounts of 38.48% and 16.46%, respectively. The nitrogen application rate and soil pH value were the main influencing factors on cotton yield under nitrogen application conditions. The nitrogen application rate and irrigation amount were the main influencing factors of cotton water use efficiency under nitrogen application conditions. The research results can provide reference for achieving high yield and efficiency in cotton cultivation under different production conditions.

Abstract:Aiming at the unclear effect of climate change and planting structure adjustment on agricultural production in Huang-Huai-Hai region, based on denitrification decomposition model (DNDC), the effect of climate and planting structure change on agricultural greenhouse gas (CO2, CH4 and N2O) emission and irrigation water demand in the Huang-Huai-Hai area was assessed by scenario analysis. The results showed that from 1995 to 2015, the climate in the study area developed towards the direction of warm and humid, in which the average annual maximum temperature did not change significantly, the average annual minimum temperature increased by 0.7℃, and the annual precipitation increased by 46.5mm.In 1995, the planting area of maize, wheat and rice was about 7.9×106 hm2, 1.4×107hm2 and 2.9×106hm2, respectively. In 2015, the planting area of three crops was increased, while the planting proportion of rice and wheat was decreased. Under the influence of climate change, agricultural greenhouse gas emissions were increased and irrigation water demand was decreased slightly in the Huang-Huai-Hai region. Compared with that in 1995, the emission intensity of CO2, CH4 and N2O was increased by 4.7%, 0.8% and 26.2% respectively to 3730.5kg/hm2, 443.2kg/hm2 and 5.9kg/hm2 in 2015, and the irrigation water demand was decreased to 499.3mm with decrease of 6.6%. The adjustment of planting structure changed greenhouse gas emissions and irrigation water demand. The evolution of planting structure in the Huang-Huai-Hai region increased the total emissions of CO2, CH4 and N 2.O to 9.9×107t, 1.4×106t and 1.3×105t, which was increased by 13.8%, 8.6% and 13.3%, respectively. Meanwhile, in some areas, the irrigation water demand was increased with the increase of corn planting proportion as a high waterconsuming crop. The research results can provide theoretical basis for the formulation of future agricultural water saving and emission reduction policies in the Huang-Huai-Hai region.

Abstract:In order to solve the problem of massive deposition or blockage of particles in the labyrinth runner pressure-compensated irrigator to the extent of affecting the normal operation of the irrigator, after simulating the deformation of the fixed gasket by FSI, and based on the coupled simulation of CFD-DEM, and after verifying its reliability through experimental comparisons, the design of the one-factor and Box-Behnken response surface test was carried out to analyze the effects of the three structural parameters of the sub-runner cross-sectional area in the pressure-compensated chamber, the outlet diameter of the pressure-compensated chamber, the diameter of the pressure-compensated chamber and their interactions on the anti-clogging performance of the irrigator, and to make comprehensive judgments on the anti-clogging performance of the irrigator through the establishment of a regression model for the prediction of the particles retention rate in the irrigator. The results showed that the particle retention rate was decreased by 5-09 percentage points when the cross sectional area of the secondary flow channel was increased from 0.018mm2 to 0.054mm2; and the particle retention rate was decreased by 2.87 percentage points when the diameter of the pressure-compensated outlet was decreased from 1.4mm to 0.8mm. Particle deposition was significantly affected by two interactions, the cross-sectional area of the secondary flow channel and the outlet diameter of the pressure-compensated chamber, and the cross-sectional area of the secondary flow channel and the diameter of the pressure-compensated chamber, with the lowest particle retention rate of 7.67% under the interaction. The regression equations of particle retention rate and three structural parameters, namely, cross-sectional area of the secondary flow channel in the pressure-compensated chamber, outlet diameter of the pressure-compensated chamber, and diameter of the pressure-compensated chamber, were fitted, which can be used to judge and predict the anti-clogging performance of the irrigator, and a set of optimal anti-clogging parameters with a secondary flow channel area of 0.051mm2, an outlet diameter of the pressure-compensated chamber of 0.894mm, and a diameter of the pressure-compensated chamber of 6.923mm were recommended. In conclusion, by studying the clogging mechanism of the pressure-compensated irrigator, a judgment model that can predict the particle retention rate was established, which reduced the probability of clogging of the irrigator and improved its stable filling time, and provides theoretical support for the design of anti-clogging performance of this type of irrigator.

Abstract:Aiming to promptly obtain soil moisture content (SMC) in the root zone of field crops for precise irrigation, hyperspectral technology was utilized. Over a 2year period spanning from 2019 to 2020, during the winter wheat jointing stage, SMC data at varying soil depths and hyperspectral data were collected. Three categories of vegetation indices were created, comprising ‘trilateral’ spectral parameters related to blue, yellow, and rededge areas, any two-band vegetation indices with the highest correlation to winter wheat root zone SMC, and empirical vegetation indices showing good correlation with crop parameters from previous studies. The vegetation indices exhibited the highest correlation with SMC at different soil depths were selected. Subsequently, random forest (RF), back propagation neural network (BPNN), and extreme learning machine (ELM) were employed to construct SMC estimation models, using the selected vegetation indices as model inputs. The results revealed that a majority of the ‘trilateral’ spectral parameters spectral indices, any two-band vegetation indices, and empirical vegetation indices displayed stronger correlations with SMC in the 0~20cm soil layer in comparison with the 20~40cm and 40~60cm layers. The two-band combinations in the 0~20cm layer exhibited the highest correlations with SMC, all surpassing 0.8. Among which, RI showed the highest correlation with SMC at 0.851, utilizing a wavelength combination of 675nm and 695nm. The RF model emerged as the most effective modeling method for SMC, with the highest accuracy observed in the 0~20cm soil layer. The coefficient of determination (R2) for the validation set of the estimation model in the 0~20cm layer reached 0.909, and the root mean square error (RMSE) was 0.008, while the mean relative error (MRE) was 3.949%. The outcomes can serve as a foundation for hyperspectral monitoring of winter wheat root zone SMC and provide valuable insights for the rapid assessment of crop growth under water stress.

Abstract:The presence of inferior bowl seedlings in hole tray seedling cultivation would affect the survival rate of later seeding transplantation. It is urgent to clean the substrate of these holes. The existing mechanical removal methods often have the phenomenon of particle scattering and omission, while the air suction removal method can effectively compensate for this defect. In order to analyze the mechanism of air suction removal of bowl seedling substrate, parameter calibration experiments for discrete element simulation research were conducted. Totally 100g of matrix was selected for particle size distribution detection and then a funnel was used to stand still. Based on image processing, the actual stacking angle on both sides of the matrix was obtained, the average value was calculated which was used as the response value. Through the Plackett-Burman experiment to screen for four significant factors that affected the stacking angle of the matrix. The maximum response area of significant factors was determined through the steepest climbing experiment. A second-order regression model was establish based on the Box-Behnken experiment and solved for the optimal parameter combination. The results showed that on the premise of taking the intermediate value of insignificant factors, when the matrix particle-particle collision coefficient of restitution A was 0.142, the matrix particle-particle rolling friction coefficient C was 0.097, the matrix particle-stainless steel static friction factor E was 0.223 and the matrix JKR surface energy G was 2.325J/m2, the simulated stacking angle was 33.4°, and the relative error with the actual stacking angle of 34.19° was 2.31%. The research result can meet the experimental requirements, and the obtained calibration parameters can be used for discrete element simulation of bowl seedling matrix.

Abstract:In order to solve the problems of slow precooling speed and high energy consumption of existing lychee origin precooling devices, a mobile lychee storage and spray precooling device control system was designed to improve the guarantee of post-harvest precooling effect of lychee. The system mainly consisted of STM32 main control system, pump driving system, refrigeration system, data acquisition system and auxiliary control system. The intelligent serial screen interface was designed based on USART HMI software. The intelligent serial screen communicated with STM32 microcontroller through TTL serial port, which can complete the setting of lychee spray precooling parameters and display the operation status information of the control system to realize the accurate control of lychee spray precooling device. The test hardware platform was built to evaluate the precooling effect with the test factors of spray flow rate of water pump and load of lychee at single spraying, and the precooling time and uniformity of lychee as the test index. The test results showed that before 1/2 precooling time (HCT), when the spray flow rate exceeded 70L/min, the spray flow rate had little effect on the cooling rate; after HCT, the precooling time was reduced by 170s (90L/min), 260s (110L/min) and 262s (130L/min), respectively, compared with that under 70L/min, so that when spraying flow rate was more than 110L/min,the effect of increase of spraying flow rate on the lychee cooling rate was not significant; test on different loads of lychee found that when the lychee load was greater than 50kg, increasing the lychee load has a significant effect on the lychee cooling rate; when the lychee load was 50kg, the uniformity of the lychee precooling completed with the increase in spraying flow first become larger and then reduced; when the spraying flow was 90L/min, the uniformity of lychee precooling increased and then trended to stabilize with the increase of lychee load. The research results can provide help for the optimization of the control system of lychee spraying precooling device to achieve rapid precooling of lychee after harvesting, so as to protect the lychee after harvesting.

Abstract:Aiming at the problems of traceability data differentiation between upstream and downstream blockchains, difficulties in fine-grained sharing, and data privacy protection existing in the existing supply chain of fruits and vegetables and agricultural products, an attribute-based cross-chain traceability access control model of fruits and vegetables supporting heterogeneous multi-chain was proposed by analyzing the business processes of various links in the supply chain of fruits and vegetables and agricultural products, and in the context of fruit, vegetable and agricultural products quality and safety traceability scenarios. The model used relay chain-based cross-chain technology to standardize cross-chain information to achieve cross-chain communication between heterogeneous multi-chains, and combined the cross-chain traceability requirements of fruits and vegetables with the attribute-based access control (ABAC) model to achieve flexible and fine-grained access control on data resources. In order to verify the effectiveness of the model, BitXHub relay chain technology was used to realize cross-chain data access in the heterogeneous chain composed of Hyperledger Fabric and Ethereum, and the attribute-based access control process was implemented in the cross-chain contract, so as to construct a prototype system of the cross-chain access control model for fruit and vegetable traceability. The system test results showed that the average maximum value of the sending rate of the relay chain for processing cross-chain transactions was about 600t/s and 400t/s, respectively, and the policy determination time did not fluctuate greatly with the increase of the number of policies, and was basically stable at about 2000ms, which was able to satisfy the needs of the fruit and vegetable supply chain for the differentiated and fine-grained sharing of cross-chain data among heterogeneous blockchains, and also ensured the data sharing process of the data privacy.

Abstract:Peanut oil is susceptible to aflatoxin B1 (AFB1) contamination during the production process. In response to the problems of tedious operation and poor timeliness of the traditional detection method for AFB1, fluorescence spectroscopy was utilized for the rapid determination of AFB1 in peanut oil. Firstly, the optimal excitation wavelength was determined by three-dimensional fluorescence spectroscopy. K-means and self organizing map (SOM) clustering algorithm were used to qualitatively identify the AFB1 content in peanut oil with an accuracy of over 95%. Nextly, two preprocessing algorithms and two dimensionality reduction algorithms were used. Competitive adaptive reweighted sampling (CARS) was selected as the best wavelength selection method. The echo state network (ESN) was then used for quantitative modeling of AFB1. Compared with other models, the results showed that the CARS-ESN model obtained the best prediction of AFB1 content. Finally, the sparrow search algorithm (SSA) was used to find the optimal ESN parameters. The final test set coefficient of determination reached 0.984, with root mean square error of 2.13μg/kg, demonstrating the feasibility of fluorescence spectroscopy technique combined with ESN to predict the AFB1 content in peanut oil. The research result can provide a theoretical basis for the development of an online system for the detection of fungal toxin content in edible oils.

Abstract:A hydraulic omnidirectional leveling system was designed based on “three-layer frame” and a composite Q learning-BP neural network-PID (QBP-PID) omnidirectional leveling control algorithm was proposed, and the crawler machine was taken as the research object. Firstly, the structural scheme and working principle of the whole omnidirectional leveling machine were given. On this basis, a dynamic model of the entire crawler machine, including the omnidirectional leveling system, was established. Then aiming at the problem that PID control parameters were difficult to be adjusted, an omnidirectional leveling compound QBP-PID controller was established. The PID control parameters were updated in real time through the BP neural network, and the Q-learning algorithm was introduced to update the neural network connection weights online. Results of simulation showed that the leveling time under QBP-PID control was 2.8s for 20° transverse leveling and 3.2s for 25° longitudinal leveling. Compared with PID and BP-PID control, the leveling time was reduced, and no overshoot occurred. In the end, the complete machine test on transverse slope road and longitudinal slope road was carried out. Compared with the simulation results, the errors of transverse and longitudinal leveling time were 0.6s and 0.4s. And the body inclination angle on the horizontal ground was less than 1.5°, which satisfied the leveling performance demand of hilly mountainous agricultural machinery.

Abstract:Aiming at the difficulty of path planning for mobile robots in complex environment, a hybrid algorithm combining ant colony algorithm and artificial potential field method for local path planning was proposed. Firstly, multi-factor heuristic function and ant travel mechanism were used to solve the problem that the path quality of traditional ant colony algorithm was poor and it was easy to fall into diagonal obstacles. Secondly, in view of the slow convergence of traditional ant colony algorithm, the adaptive volatilization coefficient and dynamic weight coefficient were designed. Then, the concepts of virtual target point, relative distance and safe distance were introduced to solve the problems of local minimum, unreachable target and excessive obstacle avoidance in traditional artificial potential field method. Finally, the turning point of the path planned by the improved ant colony algorithm was used as the local subentry point to invoke the improved artificial potential field method for secondary planning. The simulation results showed that the improved ant colony algorithm optimized the path length by 9.9% and 2.0%, the path turning times by 81.8% and 63.6%, and the convergence speed by 94.2% and 63.6% compared with that of the traditional algorithm and other literature algorithms. The improved artificial potential field method effectively solved the shortcomings of unreachable target, easy to fall into local minimum and excessive obstacle avoidance. The hybrid algorithm based on the two methods effectively combined the advantages of the two methods, and had high environmental adaptability and path planning efficiency in complex static and dynamic environments.

Abstract:In order to solve the coupling problem of a spatial rigidflexible parallel robot with flexible moving platform, a high-order flexible triangular thick plate element and its continuity constraints were proposed based on the Bézier triangle and absolute nodal coordinate formulation (ANCF). The deformation of the platform was divided by the element and its effects on dynamics were analyzed. The robotic dynamics was established by natural coordinate formulation (NCF) and ANCF. The second-order gradients of the fourth area coordinate were introduced to describe the element deformation in thickness. Besides, the Poisson locking problem was solved accordingly. The Lagrange dynamics equations were solved via the generalized α method and the Newton’s method. The statics and dynamics models of the system were simulated. The results showed that the periodic concave deformation of the moving platform had an impact on the spatial posture of the robot which was exactly consistent with the mechanism layout, mass and load distribution. The coupling mode between the rigid components and flexible moving platform conformed to the nonlinear characteristics of multibody dynamics. The trajectory error was less than 1.2×10-12mm. The errors of dynamics and constraint equations were less than the preset thresholds of 10-6and 10-14, which met the requirements of engineering applications. Meanwhile, the validity and versatility of the method were verified by comparative analysis based on different system parameters.

Abstract:In order to grasp the nonlinear dynamic characteristics of space parallel mechanism considering spherical joint clearances and three-dimensional revolute joint clearances, the dynamic characteristics analysis method of space parallel mechanism multi-body system considering compound clearances was studied. Firstly, taking 4-UPS/RPU spatial parallel mechanism as research object, the models of three-dimensional revolute joint clearances and spherical joint clearances were established, and the dynamic equation of 4-UPS/RPU spatial parallel mechanism considering both spherical joint clearances and three-dimensional revolute joint clearance was derived. Then the dynamic equation was solved numerically by Runge-Kutta method, and the effects of different clearance types, clearance values, driving speed and friction coefficient on the dynamic response of parallel mechanism were analyzed. The correctness of dynamic model and numerical calculation was verified by simulation of ADAMS virtual prototype. Finally, nonlinear characteristics of 4-UPS/RPU parallel mechanism with compound clearances were considered by phase diagram, Poincare map and bifurcation diagram. The research provided a basis for dynamic modeling and nonlinear dynamic analysis of spatial parallel mechanism considering compound joint clearances.

Abstract:In order to solve the geometric nonlinear problem of the positive stiffness module in the compliant constant-force mechanism, the positive stiffness module was designed based on the semicircular corrugated periodic beam, the negative stiffness module adopted a bistable beam, and the compliant constant force mechanism was designed by using the principle of positive and negative stiffness superposition. The theoretical model of the positive stiffness module was established by using the Mohr integral method and the compliance matrix method, and the theoretical model of the negative stiffness module was established by the elliptic integral method. The force-displacement curve of the compliant constant force mechanism was analyzed by ANSYS finite element simulation software to verify the theoretical model, and the relative error between the two was within 10%. In order to expand the range of constant-force, the response surface method was used to optimize the geometric parameters of the compliant constant-force mechanism, and the sensitivity analysis of the geometric parameters affecting the constant-force range was carried out. The prototype of the compliant constant-force mechanism was processed by 3D printing technology for experimental verification. The experimental results showed that the mechanism can maintain a constant-force of about 18.8N within the input displacement range of 1.1~6.2mm. The feasibility of the designed compliant constant-force mechanism, the validity of the optimization method and the accuracy of the theoretical model were verified.