Abstract:In the current development process of agricultural informatization, most sub-domains of agriculture face challenges such as dispersed data resources, difficulties in information integration, and low efficiency in knowledge utilization. As an emerging knowledge representation technology in recent years, knowledge graph has demonstrated powerful capabilities in semantic reasoning and data integration in specific agricultural domains. Simultaneously, it has enhanced the performance of some upper-level applications in agriculture. To systematically summarize recent research on the construction and application of knowledge graphs in the agricultural domain, the fundamentals of knowledge graphs and the process of agricultural knowledge graph construction were introduced. Furthermore, it summarized the key technologies involved in constructing an agricultural knowledge graph from four aspects: ontology modeling, information extraction, knowledge fusion, and knowledge processing. Subsequently, an overview of the current applications of agricultural knowledge graphs was provided and discussed in five aspects: information retrieval, question-answering systems, recommendation systems, expert diagnostic systems, and crop prediction. In conclusion, the research status of agricultural knowledge graphs was summarized and it was suggested that future research in agricultural knowledge graphs should explore areas such as multimodal knowledge reasoning, timely knowledge updating, multilingual knowledge queries, cross-domain data fusion, and sub-domain knowledge graph construction.

Abstract:The spaceborne LiDAR measurement system offers a solution by covering remote areas that are challenging to access via airborne systems. This system effectively addresses the technical limitations of optical imaging and synthetic aperture radar measurements, thereby providing a reliable data source for swiftly and accurately obtaining forest structure parameters. However, the existing review on applications of spaceborne LiDAR has a lack of analysis that measures the accuracy of forest structural parameters estimated by spaceborne LiDAR as a whole and the content of the analysis is not systematic. The existing spaceborne LiDAR technology observation system were initially reviewed, and then the applicability of spaceborne LiDAR data estimation for multi-scale forest structure parameters was explored. The current development trends, as well as the existing advantages and disadvantages of satellite-borne LiDAR were also analyzed. Finally, the current challenges and anticipates the future prospects and development directions of spaceborne LiDAR technology were summarized. The research result can serve as a valuable reference for the design and application of spaceborne LiDAR remote sensing satellites in forestry. It was suggested that further research can deepen the in-depth study on the inversion of different forest structure parameters, product systems and standard specifications, accuracy evaluation of forestry applications, and design of LiDAR parameters for forestry.

Abstract:To address the problems of high labor intensity, limited operating speed, and low efficiency in long-distance row-follow operation of traditional manually operated cotton pickers, based on the prior characteristics of parallel precision sowing of cotton using satellite navigation, an automatic row-follow control method of cotton picker based on parallel trajectory navigation was proposed. Firstly, the cotton picker was manually driven to complete the first row alignment operation, the RTK positioning trajectory data was recorded, and the linear fitting was carried out to obtain the slope of the reference navigation line. Then, when harvesting the subsequent rows, the cotton rows to be harvested were aligned at any starting point, and the parallel navigation baseline was calculated by the starting point coordinates and the reference slope. Finally, a speed-adaptive pure tracking algorithm was adopted to realize automatic row-follow control of the cotton picker. Field experiments demonstrated that this method could automatically align rows from the starting position of any row. At different speeds, the average absolute lateral deviation was 2.91cm, and the maximum absolute lateral deviation was 9.22cm, and the standard deviation was 1.80cm, which met the requirements of the alignment accuracy of cotton picker. Compared with the mechanical touch-based automatic row-follow methods, this method was not affected by missing or broken plants and maintained good alignment accuracy at high speed of 7km/h.

Abstract:To address the challenges of limited feature point matching, vulnerability to loss, and sparse point cloud in dark lighting conditions in orchards, the ORB-SLAM2 was improved by proposing an adaptive threshold-based algorithm for dense construction of binocular 3D orchard maps. Firstly, a FAST corner extraction method with adaptable threshold values was introduced in the tracking thread, and ORB features were extracted from left and right eye images by calculating the average pixel solution threshold across images captured under different lighting conditions, which effectively enhanced the number of feature point matches under different lighting conditions. Subsequently, local map tracking was performed based on camera pose estimation by using feature points and accomplished local map construction through bundle adjustment optimization of key frame map points derived from the tracking thread. Based on the original algorithm, a dense mapping module was incorporated by utilizing ZED-stereo binocular deep fusion to acquire image pairs through feature matching of key frames from the left and right eyes. Depth information was obtained by solving the image pairs, camera pose was determined via depth optimization, and local point clouds were constructed and stitched together based on the camera pose. Finally, global BA optimization was applied to refine the resulting point cloud map, enabling the construction of a three-dimensional dense map of an orchard. The improved ORB-SLAM2 algorithm demonstrated enhanced convergence in terms of absolute trajectory error when evaluated on the KITTI data set sequence. Specifically, the standard deviation of trajectory error was decreased by 60.5% and 62.6% in sequences 00 and 07, respectively, while also exhibiting varying degrees of improvement in other sequences. These results indicated a notable enhancement in positioning accuracy compared with the original algorithm. The results demonstrated that in comparison with the original algorithm, the proposed algorithm exhibited excellent adaptability to diverse lighting conditions. Specifically, it achieved an average increase of 5.32%, 4.53%, 8.93% and 12.91% in feature point matching under strong light, normal light, dark light, and rainy day respectively. The results demonstrated that the yaw angle exhibited enhanced convergence, resulting in higher positioning accuracy. Moreover, the proposed algorithm reduced the number of extracted key frames by 2.86% and decreased average tracking time by 39.3% compared with the original approach. Additionally, it achieved a favorable dense mapping effect, accurately reflecting both robot pose and real environmental information within the orchard. Consequently, this method satisfied the requirements for constructing a 3D dense point cloud map of an orchard and provided essential support for realizing navigation path planning for orchard robots.

Abstract:Aiming at the problems of low control accuracy and frequent steering switching of single-side braking steering track chassis in automatic walking, a three-tangent local path dynamic planning algorithm based on single-side braking steering track chassis was proposed. According to the uncontrollable characteristics of steering braking torque of track chassis, a local path planning method of “steering-straight-steering” was designed. The three-tangent local path dynamic planning algorithm was composed of a straight line and two arcs, the first arc was tangent to the vehicle direction, and the vehicle gradually turned to the target route along the arc. The second straight line was tangent to the two arcs, and the vehicle gradually approached the target route along the straight line. The third section of the arc was tangent to the target course, and the vehicle turned gradually to the target course along the arc. The turning radius of the vehicle was calculated by weighted recursive average filtering through the lateral deviation and heading deviation of the historical movement, and the current stage was judged for adjustment. The results of field automatic straight line driving test showed that in the initial state with a lateral deviation of 0.25m and a heading deviation of 0°, the driving speed of 0.4m/s was advanced. The mean value and standard deviation of the absolute transverse deviation of the algorithm were 0.085m and 0.104m, respectively, which were 23.19% and 19.54% higher than that of the bang-bang algorithm, respectively. The mean value and standard deviation of the absolute course deviation were 3.31°and 3.74°, respectively, which were 25.95% and 25.64% less than that of the bang-bang algorithm, respectively. The total steering control times were 9 times, which was 43.75% less than that of the bang-bang algorithm. The results showed that the three-tangent local path dynamic planning algorithm was suitable for single-side braking steering track chassis, with higher path tracking accuracy and lower control frequency, and can meet the needs of field operations.

Abstract:Unmanned operation of grain harvesting machinery can address the issue of labour shortage effectively during the harvesting season. The whole-field automatic navigation is a significant indicator of intelligent unmanned harvester. To address the issue of automatic dotting for unmanned rice harvesting, the automatic navigation of the outermost circle of the rice field was realized based on cut boundary fitting and headland region detection. The trajectory of rice harvester was recorded by using Beidou real-time kinematic (RTK) while working in the outermost circle of the paddy field, which enabled automatic dotting for the remaining operation region (ROR). A dynamic region of interest was constructed by employing the priori morphology knowledge of the target area, which not only improved the robustness of cut boundary fitting but also reduced its computational complexity. Once the rice harvester entered the headland region, the uncut boundary in the front of the machinery was extracted and the intersection point between two uncut boundary lines was fixed. Finally, the position of the vertex corresponding to the ROR was calculated based on the antenna position and operating width. The proposed method was verified based on an automated trolley, and results indicated that the average straight line tracking error was 4.1cm, and the maximum tracking error was 6.3cm at speed of 0.8m/s. The processing time for single image was less than 50ms. The average error of automatic dotting was 3.5cm, and it can realize automatic dotting of rice field of right-angled trapezoid. It can be concluded that the designed automatic dotting system satisfied the needing of whole-field unmanned operation of rice harvester.

Abstract:Aiming at the vulnerability of multi-rotors drones to various external disturbances during flight, an active disturbance rejection control (ADRC) system was designed for a rotor control system of composite wing unmanned aerial vehicle (UAV) as the research object. Firstly, the kinematics and dynamics models of the multi-rotor drone system were established. Secondly, the characteristics of the ADRC algorithm were studied, and position and attitude controllers were designed in combination with the drone model. The extended state observer (ESO) was improved by introducing a more accurate dynamic model, which enhanced the disturbance observation speed and estimation accuracy. At the same time, the order of the extended state observer was reduced, and the simplicity of controller parameter tuning was improved. Furthermore, based on the six-degree-of-freedom force and moment balance equation, a control allocation model was built for the research object. Finally, Matlab/Simulink was used to complete the simulation model design and parameter adjustment. Internal gravity disturbance and external wind disturbance were added to the control targets. The simulation results showed that the controller designed could not only estimate and compensate for the internal and external disturbances of the system, but also had strong anti-interference ability, which could ensure that the drone reached the target position quickly and smoothly from the initial point, and maintained stable hovering. The steady-state error of attitude control was within 0.05°.

Abstract:The contact parameters affect the discrete element simulation results of controlled-release fertilizer particles. In order to accurately simulate the mechanical behavior and motion law of controlled-release fertilizer particles, the contact parameters of controlled-release fertilizer particles were calibrated and experimented based on the discrete element method. Firstly, the basic model of controlled-release fertilizer discrete elements was established, and the contact parameters between controlled-release fertilizer particles and PVC panels were calibrated in EDEM by combining bench and simulation experiment. Secondly, the collision recovery coefficient, static friction factor and rolling friction factor between controlled-release fertilizer particles and PVC plate were measured by collision bounce experiment, slope slip experiment and inclined rolling experiment to be 0.539, 0.507 and 0.105, respectively. Finally, through the accumulation experiment, the steepest climbing experiment and the orthogonal rotation combination experiment, the collision recovery coefficient, static friction factor and rolling friction factor between controlled-release fertilizer particles were obtained as 0.38, 0.25 and 0.09, respectively, and verified by bottomless cylinder lifting experiment and fertilizer discharge bench experiment. The experiment results showed that the relative error between the actual value of the accumulation angle and the simulation result was 1.54%, and the relative error between the actual value of the discharge amount and the simulation result at the four speeds was 4.38%, 4.23%, 4.41% and 4.36%, respectively, and the contact parameters of the calibrated controlled-release fertilizer were accurate and effective, which can provide data and model support for the simulation of controlled-release fertilizer discrete elements.

Abstract:Aiming at the existing heavy-duty precision planter soil-covering device at high speed, there are problems such as high impact on seed spacing uniformity, seed being overhead, etc., the proposed combination of Archimedean spiral-type and extruded soil cover technical ideas, an Archimedean spiral-type arc double-toothed disc soil-covering device was designed, to achieve suitable and uniform mulch thickness and reduce the impact of mulching operations on seed spacing uniformity. The structural parameters of the mulching device and the amount of mulch were analyzed, and the range of values of the main parameters affecting the effect of mulching was determined;the simulation and optimization analyses of the designed components were carried out by using EDEM software to determine the optimal parameter combinations of the arc double-toothed disc;and the operational performance of the device was verified through field experiments. The results showed that the main structural parameters affecting the operational performance of the arc double-toothed disc were the center distance, installation inclination, and embedded depth of the arc double-toothed disc. When the center distance of the arc double-toothed disc was 142.6mm, the installation inclination was 20°, and embedded depth was 55mm, the device passed well with suitable and uniform mulch thickness, the average mulch thickness was 47mm, and the standard deviation and coefficient of variation of qualified spacing differed from the blank control group by 0.22mm and 1.81 percentage points, which was a small influence on the uniformity of sowing and met the agronomic requirements of precision sowing.

Abstract:In order to solve the problem of poor high-speed sowing quality of mechanical soybean seeder, an independent part-filling soybean double-row brush high-speed precision seeder was designed, which increased the number of holes by using the double-row seed tray structure, and independently divided the filling to avoid the mutual disturbance of the population backlog, so as to realize the high-speed precision seeding operation. Through theoretical analysis, the influencing factors of seed filling and seeding performance were clarified, and the relevant parameters of the molding hole and diversion groove were determined. The results showed that when the soybean plant spacing was 8cm and the seed distributor operating speed was 9km/h, the optimal combination was the hole inclination angle of 54° and the diversion trough inclination angle of 39°, the pass index was 95.5%, the missing seeding index was 2.0%, and the plant spacing variation coefficient was 13.1%. The results of the comparative test showed that when the operating speed was 8~12km/h, the qualification index of the independent sub-filling seed dispenser was increased by more than 1.3 percentage points, the missed seeding index was reduced by more than 0.8 percentage points, and the plant spacing variation coefficient was reduced by more than 2 percentage points. The results of the agronomic adaptability test of narrow row dense planting showed that under the condition of operation speed of 8~12km/h, the qualification index of independent sub-filling seed discharger under narrow row dense planting agronomy was above 90.1%, the missed seeding index was no more than 4.6%, and the coefficient of variation of plant spacing was no more than 20.1%, and the agronomic adaptability of narrow row dense planting was high.

Abstract:An air-assisted linear seed seeding device for wheat was created in order to address the issue that the pressure gradient’s change direction in the air-assisted seed seeding device is inconsistent with the direction of seed movement, leading to seed backflow or collision and decreasing the stability of grain spacing. To simulate the seeding process, a CFD-DEM unidirectional coupling simulation model was developed, which was based on the mathematical model of gas-solid two-phase flow. According to the simulation results, the length of the pipeline and the inlet pressure had a big impact on the flow field’s pressure distribution and overall pressure loss. The seeding device’s flow field had a uniform pressure distribution, and the direction of the pressure gradient change corresponded with the direction in which the seeds migrated.Wheat seeds followed a “straight-curve-straight line” motion path that avoided collisions and backflow. Inlet pressure, pipeline length, and operating speed were used as test factors in the response surface optimization test, and the grain distance variation coefficient served as the test evaluation index. The test findings indicated that inlet pressure, operating speed, and pipeline length were the primary and secondary factors influencing the grain distance variation coefficient. Operating speed and pipe length were influenced by inlet pressure. Through parameter optimization, the following parameters were found to be the best combinations: operation speed of 0.11m/s, pipeline length of 24.2cm, and inlet pressure of 5.1kPa.The soil tank test confirmed that under these conditions, the average grain spacing was 5.3cm and the coefficient of grain spacing variation was 6.3%, meeting the agronomic requirements for precise, uniform wheat sowing. By addressing the issues of wheat seed backflow and collision, air-assisted linear seeding may greatly enhance seeding performance and offer technical assistance for accurate and consistent wheat sowing.

Abstract:In order to realize the directional seeding of pumpkin seeds, a combined direction method was proposed to standardize the movement attitude of pumpkin seeds and then orient them, and a combined directional seed metering device for pumpkin seeds was designed. The key components of the device to realize the direction of pumpkin seeds included two parts: limit clearing plate and deflector plate. Based on the geometric characteristics of pumpkin seeds, the structural parameters of the limit cleaning plate were determined. The kinematic equations of the direction process of pumpkin seeds were established, and the relationship between the attitude change and the structural parameters of the deflector plate during the direction process of pumpkin seeds was clarified, and the structural parameters of the deflector plate were determined. The coupling analysis method of CFD-DEM was used to verify the feasibility of directional seeding of the seed metering device. The value range of rotation speed of seed discs was determined. A test bench was built. With the flat adsorption rate as the test index, the diameter of the hole in the seed discs was determined by a single factor test. The single seed rate was the test indicators, and the value range of negative pressure and the number of holes in the seed discs were determined by the single factor test. The negative pressure, the number of holes in the seed discs and the rotation speed of the seed discs were selected as the test factors, and the single seed rate and directional success rate were the test indicators. A three-factor and three-level secondary center combination test was carried out, and the primary and secondary factors influencing the single seed rate of the seed arranger were the negative pressure, the number of holes in the seed discs and the rotation speed of the seed discs. The primary and secondary factors influencing the directional success rate was the rotation speed of the seed discs, the negative pressure, the number of holes in the seed discs. Field validation experiments were carried out on the optimized data, and the results showed that when the negative pressure was 10kPa, the number of holes in the seed discs was 16, and the rotation speed of the seed discs was 4r/min, the single seed rate was 94.7%, the directional success rate was 82.7%, and the positive and negative deflection angle of the seeds falling into the cave basin was 18.5°.

Abstract:To explore a technology solution capable of achieving high-speed seedling retrieval followed by uninterrupted rapid seedling deployment and synchronized seedling reception, an analysis was conducted on the four-stage movement process of rapid seedling deployment after high-speed seedling retrieval by using a pneumatic down-press type high-speed seedling retrieval and deployment device, obtaining the movement time of tray seedlings. High-speed photography was used to capture the process of uninterrupted rapid seedling deployment by the research group’s custom-built high-speed seedling retrieval and deployment device, with PCC software utilized for capture analysis and calculation, examining the relationship between displacement and time during the continuous seedling throw and deploy process. Through limit analysis and intersection operation of synchronous seedling reception, the range of advance phase angles for hanging cup synchronous seedling reception at different seedling retrieval frequencies was determined, achieving the spatio-temporal matching relationship between high-speed seedling retrieval and deployment device, high-speed seedling retrieval, uninterrupted rapid seedling deployment, and hanging cup synchronous seedling reception. Verification experiments of the hanging cup synchronous seedling reception effect were conducted on a high-speed transplanter prototype, with results showing a success rate of 96.4% for hanging cup synchronous seedling reception at a retrieval frequency of 180 seedlings/min, providing foundational data for the spatio-temporal matching relationship between the high-speed seedling retrieval and deployment device and the hanging cup.

Abstract:Based on the fact that existing seedlings pick-up test bench cannot meet the needs of various types of seedlings bowls and seedlings pick-up mechanisms for seedlings pick-up test, a kind of multipurpose seedlings pick-up test bench which was driven by multiple stepper motors was designed and it can adjust seedlings pick-up mechanisms inclination and program in the control system. It can adapt to different types of bowls and meet the test requirements of various types seedlings pick-up mechanisms by adjusting the position and posture of the mounting bracket, changing the distance between steel wire and the diameter of the circumference of the wire in the driving cage of the bowl. By editing the controller program, the linkage of three stepper motors can be controlled and the horizontal and longitudinal seedlings feed process of the seedlings feeding device can be matched with seedlings pick-up process of the seedlings pick-up mechanism. Modeling, assembling and virtual simulation of the test bench were carried out in 3D modeling software. Two types of seedlings bowls and three types of seedlings pick-up mechanisms were tested on the physical prototype produced by processing. The results of the three tests showed that the seedlings pick-up test could be successfully completed on the seedlings pick-up test bench, which proved that the test bench could meet the test needs of various types of seedlings bowls and seedlings pick-up mechanisms.

Abstract:In order to accurately analyze the unsteady flow characteristics in the impeller of a double-suction pump as a hydraulic turbine, the SST k-ω turbulence model was used for numerical simulation under design conditions. The dynamic mode decomposition (DMD) of a periodic unsteady velocity field was carried out, and the first four main modes and their corresponding spatio-temporal information were obtained by combining the Q criterion. The analysis results showed that the DMD method decomposed the complex flow field characteristics in the impeller into dynamic and static interference modes, basic modes and dissipative modes. Among them, the dynamic and static interference modes were dominant, and the frequency was the rotation frequency of the impeller, which reflected the flow characteristics of the flow in the impeller disturbed by the static components. The vortex structure was mainly point vortex and discontinuous tubular vortex. The basic modal frequency was zero, which reflected the steady-state flow field characteristics caused by the geometric characteristics of the blade flow channel. The vortex structure was mainly continuous tubular vortex. The third-order and fourth-order modes were dissipative modes, which reflected the characteristics of flow separation and unstable vortex structure shedding on the blade caused by the interference of static components in the impeller, mainly sheet vortex and discontinuous tubular vortex. The DMD method can decompose the flow field structure of the impeller at a specific frequency, and can clearly analyze the unsteady characteristics of the complex flow field in the impeller of the double-suction pump as a hydraulic turbine.

Abstract:In order to improve the supervision efficiency and accuracy of the supplementary cultivated land projects, and ensure that the quantity of the supplementary cultivated land is accurate, the land type meets the requirements, and the location is reasonable. It was based on spatial big data to research on the automatic discriminant technology of supplementary cultivated land compliance. The automatic identification rules and indicator system for the compliance of supplementary cultivated land was designed based on the big data framework, parallel computing technology, GIS spatial analysis, and the technical process, algorithm and software for the automatic identification of the compliance of supplementary cultivated land were developed. Daily and special supervision and verification of supplementary cultivated land were carried out, which became an increasingly important technical means for supervision and verfication of supplementary cultivated land projects. Practical operations demonstrated that the average analysis time for the project was 2~4 minutes, and the average of over 5700 problematic projects were prevented from being included in the database each month. This research provided technical means for compliance identification in the initiation, inplementation, and acceptance of supplementary cultivated land projects, enhancing the rationality of project initiation and improving the technical level of information verification, supervision of supplementary cultivated land. It played a crucial technical support role in ensuring the implementation of the system of the cultivated land requisition compensation balance in China.

Abstract:The rapid and accurate extraction on cropland information by using remote sensing technology is a key aspect of cropland protection. Taking Shanghe County of Shandong Province as an example, a cropland information extraction method of Landsat 8 OLI images based on multi-seasonal fractal features was proposed. Firstly, the upper fractal signal and lower fractal signal of each pixel of multi-seasonal remote sensing images were calculated by using a blanket covering method, and the fractal characteristics of cropland and other land use types were compared and analyzed. The third scale of the upper fractal signal was selected as the feature scale to extract the spatial distribution of cropland in Shanghe County. Secondly, the land use vector data, Esri land cover data and statistics at the same period were used to evaluate the extraction accuracy of cropland information. Finally, comparative experiments between multi-seasonal fractal extraction with the single season fractal extraction and the existing land use data products were set up to evaluate the accuracies based on the point matching degree and area matching degree, respectively. The results showed that the multi-seasonal data can better reflect the complexity of crop growth and improve the extraction accuracy of cropland information. Different land use types had different signal values at different fractal scales, and their fractal features can clearly depict the differentiations among them at different scales. The evaluated point matching degree and area matching degree of cropland extraction results by using multi-seasonal fractal features based on the land use vector data and Esri land cover data were 87.13% and 89.83%, 99.73% and 97.91%, respectively, which were higher than that of single season fractal extraction. Considering the point matching degree, area matching degree and spatial distribution characteristics, the research method could effectively distinguish cropland and other land use types, which had much better extraction results and a higher consistency with the statistical data. The method developed can accurately extract the cropland information and provide technical supports for the dynamic monitoring and damage assessment of cropland.

Abstract:In order to improve the recognition accuracy of irrigation water bodies in irrigated farmland, the Jiefangzha Irrigation Area of Hetao Irrigation District was taken as the study area, and the surface water body extraction model (WatNet) was improved based on Sentinel-2 remote sensing images, combined with the actual situation of the irrigation area, to obtain the MWatNet model and extract irrigation water bodies. Overall accuracy (OA), mean intersection over union (MIoU), F1 value and other water body extraction accuracy indicators were used for comprehensive evaluation. The results showed that the improved surface water body extraction model (MWatNet) had good recognition accuracy in the extraction of farmland irrigation water bodies in Jiefangzha Irrigation Area, the overall accuracy of the model reached 96%, the mean interaction over union reached 83%, the F1 value was 80%, and the accuracy of the field research validation was 85.7%;comparing with the original WatNet, the semantic segmentation model of water bodies (Deeplabv3_plus), and the water body extraction model (Deepwatermapv2), MWatNet showed better results and model operation efficiency in terms of connectivity of irrigation water body extraction, and elimination of road and town interference. The quantitative characterization of irrigation water bodies can be achieved by using this model, which provided data support for irrigation water scheduling.

Abstract:Forest aboveground biomass (AGB) is an important indicator for evaluating forest growth. Based on the 2D and 3D data generated by digital aerial photography (DAP), totally 41 point clouds height variables and 16 visible light vegetation indices were calculated respectively, and AGB estimation models were developed with single variable set and comprehensive variable set respectively by using six regression algorithms (random forest, RF;bagged tree, BT;support vector regression, SVR；Cubist;categorical boosting，CatBoost；extreme gradient boosting，XGBoost) to explore the contribution of different variables to the AGB estimation model. The results showed that the highest accuracy AGB prediction models for spectral and point cloud datasets were Cubist and XGBoost, with R2 of 0.5309 and 0.6395, respectively, and the highest accuracy model for the combined dataset was XGBoost, with R2 of 0.7601, and the XGBoost model had a higher stability of AGB estimation. The result also showed that the contribution of the six machine learning models mainly depended on the regression method considered, and the number of features chosen and the importance of the features to the model were not consistent across the models. DOM spectral features had a higher importance in the estimation of AGB. Overall, the combination of 2D and 3D data can effectively improve the accuracy of forest AGB estimation, and the RGB images acquired based on UAV tilt photography can realize the fast and nondestructive estimation of forest AGB.

Abstract:In order to optimize nutrient management and ensure normal plant growth, UAV remote sensing technology was used to efficiently and non-destructively collect crop seedling information in the field, and to monitor the leaf area index (LAI) and the relative chlorophyll content (SPAD) of oilseed rape during the seedling stage. It is difficult to balance the monitoring efficiency and monitoring accuracy due to the constraints of flight altitude and image resolution of UAVs. A super-resolution reconstruction method was adopted to integrate the high-resolution images taken at lower flight altitudes and reconstruct the images taken at higher flight altitudes, so that LAI and SPAD could be monitored by the flight images taken after the modeling was completed. Three nitrogen fertilizer gradients, three sowing periods, and three planting densities were set up, and the UAV was used to collect the images of oilseed rape seedlings at 20m and 40m flight altitudes respectively in seedling stage, and SRRestnet method was used to analyze the seedling images at 40m and 40m flight altitudes respectively. SRRestnet method, and super-resolution reconstruction was performed on the 40m images. Based on the three combinations of features extracted from the 20m, 40m and 40m reconstructed images, three machine learning methods, namely partial least squares regression (PLSR), random forest (RF), and support vector regression (SVR), were utilized to monitor LAI and SPAD. The results showed that the super-resolution reconstructed images performed well in phenological seedling monitoring, and PLSR monitoring of LAI and RF monitoring of SPAD had the highest monitoring accuracy, and the operational efficiency of the 40m reconstructed images was 48.6% higher compared with that of the 20m images.

Abstract:The basic seedling number is an important basis to reflect the health level of rice. Accurately estimating the basic seedling number at tillering stage can guide the fertilizer and nitrogen amount in later stage, so as to regulate the optimal tillering number of rice. At the same time, it is of great significance for rice growth monitoring and yield forecasting. Considering that traditional manual field statistics on the number of basic seedlings are time-consuming and costly, this experiment took rice at tillering stage in Zhenjiang Runguo Farm, affiliated farm of Jiangsu University, as the research object, and used DJI UAV (M600 Pro) equipped with multi-spectral camera (Rededge-MX) to obtain multi-spectral data of rice at tillering stage. After image splicing, radiometric correction, geometric correction and other pretreatment operations were carried out on the original image, the soil end elements and vegetation end elements were extracted according to the pixel purity coefficient, and the spectral library was established. Then the mixed pixel decomposition was performed according to the fully constrained least square method, and the regression model of vegetation coverage and the number of basic rice seedlings was constructed. The model determination coefficient obtained by this method was 0.891, and the root mean square error RMSE was 4.6 plants/m2. In the traditional pixel dichotomy model (based on NDVI, VDVI and GNDVI vegetation index), the determination coefficients of R2 were 0.834, 0.744 and 0.642, and the RMSE were 5.7 plants/m2, 7.1 plants/m2 and 8.4 plants/m2. The experimental results showed that the evaluation indexes of the model based on the hybrid pixel decomposition method were superior to the pixel dichotomy model. The statistical accuracy of rice basic seedlings can be effectively improved based on the decomposition of mixed pixel decomposition, and the inverse map of rice basic seedling number is generated, which can directly count the basic seedling number and provide guidance for rice seedling replacement and thinning at tillering stage.

Abstract:The Asian citrus psyllid (ACP) serves as the primary vector for Huanglongbing (HLB), a citrus tree disease with potentially devastating consequences for citrus orchards. In order to achieve efficient monitoring of ACP populations, an intelligent monitoring system capable of insect trapping, pest identification, and result visualization was developed. A monitoring device equipped with an automatic renewal mechanism for the insect trapping tape and real-time image capturing was designed. To improve the performance of the YOLO v8 model for ACP recognition, targeted cropping and Mosaic data augmentation techniques were employed to effectively expand the ACP dataset, addressing issues related to limited sample size and constrained positioning in the datasets. The application of a coordinate attention (CA) mechanism guided the model to comprehensively consider both channel and spatial information, thereby enhancing its ability to accurately locate the target psyllids. Additionally, the Web interface and mobile APP were developed to enable data visualization and remote control. During the model testing phase, the improved YOLO v8-MC achieved significant better performance than the baseline model, reaching 91.20%, 91%, and 90.60% in terms of recall rate, F1 score, and precision, respectively. In the field experiment, the model exhibited a recall rate of 88.64%, an F1 score of 87% and a precision of 84.78%, and the system operated effectively, meeting the requirements for field applications. In conclusion, the intelligent monitoring system developed enabled remote monitoring of ACP populations in orchards, providing an efficient mehtod for the management and control of such pest infestations.

Abstract:In response to large number of parameters and poor robustness of object detection algorithms in complex orchard environment, an improved YOLO v7 network for apple maturity (immature, semimature, mature) detection was proposed. With YOLO v7 as the baseline network, a window multi-head self-attention mechanism (Swin transformer, ST) was adopted into the feature extraction structure to greatly reduce the parameters and computational complexity. In order to improve the ability of the model for detecting small targets in distant images, adaptively spatial feature fusion (ASFF) module was adopted into the feature fusion structure to optimize the Head part, effectively utilizing shallow and deep features and enhancing the performance of the feature scale invariance. Wise intersection over union (WIoU) was used to replace the original complete intersection over union (CIoU) loss function, thus accelerating the convergence speed and detection accuracy. The experimental results showed that the improved YOLO v7-ST-ASFF model had significantly improved the detection speed and accuracy on the test set of the apple images. The average detection precision, recall, mean average precision (mAP) for different maturity levels can reach 92.5%，84.2% and 93.6%, all of which were better than that of Faster R-CNN, SSD, YOLO v3, YOLO v5, YOLO v7 and YOLO v8 object detection models. The detection effects were good for multi, single, frontlight, backlight, distant and close targets, as well as bagged and unpacked targets. The size of the model was 53.4MB, and the ADT was 45ms, which was also better than that of other models. The improved YOLO v7-ST-ASFF model can meet the detection of apple targets in complex orchard environment, providing effective exploration for automated fruit and vegetable picking by robots.

Abstract:With the industrialized application of livestock breeding digital intelligent monitoring technology, further enhancement of the classification of livestock and poultry breeding policy fine management has become a new demand for fine and efficient breeding management in modern livestock industry. Adopting fixed camera position and multi-angle video acquisition technology, the wandering behavior of sheep in the process of grazing in real time was recorded;a multi-target detection model of sheep was designed based on YOLO v5 model, and multi-target real-time tracking and identification of sheep in the process of wandering was realized in response to the complex situation that was prone to be blocked in the video of sheep wandering, and the identification rate of small and medium-sized sheep can reach 90.63%.Then the DeepSORT algorithm was adopted for sheep wandering multi-target trajectory tracking, through extracting the depth of sheep target epigenetic features, the sheep wandering trajectory graph and the sheep wandering variable tempo change data were obtained. The experimental results showed that the wandering behaviors of sheep were usually in three different combinations: slow walking, fast walking and sprinting, and the wandering behaviors of a single sheep were usually in random combinations that were not fixed. In medium to large-scale sheep flocks, due to the complexity of their kinship structure, the flocks often differentiated into multiple small groups, which made it exceptionally difficult to observe and analyze their behavior holistically. In order to overcome this difficulty, it was turned to a small-scale target flock, and the synchronization phenomenon was initially verified on the beat cycle of sheep wandering through the empirical analysis of the three wandering processes of sheep dispersal, aggregation and synchronization.

Abstract:Accurate extraction of the udder region of dairy goats was the key to realize non-invasive temperature detection of dairy goats. Due to the occlusion of breast area and the low quality of thermal infrared image, the detection accuracy needs to be further improved. Based on thermal infrared imaging technology, an improved YOLO v5s based detection method for key parts of milk goat udder was proposed. By replacing some convolutional modules of Backbone network in the original model with ShuffleNetV2 structure, the number of parameters in network deployment and training process was reduced, and the purpose of lightweight network design was realized. By introducing CBAM attention mechanism into the head of the Neck network detection head, the complexity of the network has been reduced and the detection accuracy of the breast region of dairy goats was ensured. Totally 4611 infrared images of breast of pregnant dairy goats containing complete information, incomplete information and blurred edges were collected, and the model was trained after location labeling. After testing, the accuracy of the model was 93.7%, the recall rate was 86.1%, the mean average precision was 92.4%, the number of parameters was 8×105, and the floating point computation was 1.9×109. Compared with the YOLO v5n，YOLO v5s，YOLO v7-tiny，YOLO v7，YOLO v8n and YOLO v8s target detection network, the accuracy of this network was increased by 1.9 percentage points，1.2 percentage points，1.6 percentage points，4.3 percentage points，3.5 percentage points and 2.7 percentage points, the recall rate was increased by 3.4 percentage points，5.0 percentage points，0.1 percentage points，2.6 percentage points，0.9 percentage points and 1.5 percentage points, the number of parameters was decreased by 1.1×106，6.2×106，5.2×106，3.6×107，2.4×106 and 1.0×107, and floating-point calculations were reduced by 2.6×109，1.4×1010，1.1×1010，1.0×1011，6.8×109 and 2.7×1010, respectively. It met the detection requirements of the key parts of milk goat udder, and significantly reduced the number of parameters of the network without losing the detection accuracy, which was conducive to the deployment and use of the network in different environments, and had reference significance for the design of non-contact temperature monitoring system for milk goat body temperature.

Abstract:In order to promote the intelligence of fishery equipment, video streaming-based fish feeding behaviour recognition has received extensive attention in recent years. The model of traditional recognition methods based on video streaming is too complex to be realized on edge computing devices. To address this problem, a lightweight 2D-convolutional motion feature extraction network, Motion-EfficientNetV2, was proposed which can effectively recognize fish feeding behaviour by using video streams as input. The proposed model used EfficientNetV2 as the backbone network, constructed the motion feature extraction module Motion based on TEA and ECANet, and embeded the Motion module into each Fused-MBConv module of EfficientNetV2, in order to give EfficientNetV2 the ability to extract motion features. The MBConv in the EfficientNetV2 network was also improved by using ECANet to enhance its channel feature extraction capability. Null convolution was used in Motion-EfficientNetV2 to expand the receptive field and improve the wide-range feature extraction capability. The experimental results showed that after introducing the designed Motion module and a series of improvements, the number of parameters and FLOPs of Motion-EfficientNetV2 was 9×106 and 1.31×1010, respectively, which were reduced compared with EfficientNetV2. Comparison experiments using the same dataset in the algorithmic models of TSN-ResNet50, TSN-EfficientNetV2, C3D, and R3D, respectively, showed that the present algorithm achieved an accuracy of 93.97% while the number of parameters and FLOPs were lower than the rest of the models. Therefore, the model proposed can effectively identify fish feeding behavior and guide aquaculturists to develop fish feeding strategies.

Abstract:In response to the problems of insufficient utilization of pretrained language models, low utilization of external knowledge injection, and low recognition rate of nested named entities in the process of named entity recognition in the field of agricultural diseases, a named entity recognition model continuous prompts for machine reading comprehension (CP-MRC) was proposed based on continuous prompt injection and pointer network. This model introduced the bidirectional encoder representation from transformers (BERT) pretraining model, which freezed the original parameters of the BERT model and retained its text representation ability obtained during the pretraining stage. To enhance the applicability of the model to domain data, continuous trainable hint vectors were inserted into each layer of Transformer. To improve the accuracy of nested named entity recognition, a pointer network was used to extract entity sequences. A comparative experiment was conducted on a self built agricultural disease dataset, which included 2933 text corpora, 8 entity types, and a total of 10414 entities. The experimental results showed that the accuracy, recall, and F1 values of the CP-MRC model reached 83.55%, 81.4%, and 82.4%, which was superior to other models. The recognition rate of nested entities in pathogens and crops was increased by 3 percentage points and 13 percentage points in F1 value compared with that of others, and the recognition rate of nested entities was significantly improved. The model still had good recognition performance with only a small number of trainable parameters, providing ideas for the application of large-scale pretrained models in information extraction tasks.

Abstract:The expansion of the trading scale of aquatic products has led to the widespread application of blockchain technology in product traceability and data sharing. However, the inability of data exchange and sharing between blockchains constructed in different enterprises or different links within the same supply chain has led to the persistence of information silos. Therefore, a cros-chain transaction model for aquatic products was proposed based on a notary mechanism, aiming to achieve cross-chain transactions between different links of aquatic products, solve single point of failure problems, and enhance transaction efficiency. Meanwhile, addressing the characteristics of cross-chain transactions for aquatic products, a two-stage cross-chain transaction process was proposed, along with an improved transaction error handling mechanism, to address the requirements of data consistency and atomicity during cross-chain transactions. A system prototype was built based on the Hyperledger Fabric platform. Experimental results demonstrated that under high transaction volumes, the average success rate of transactions exceeded 99%, with an average latency of around 0.21 s. Thus, the model met the requirements of cross-chain transactions.

Abstract:The application of blockchain technology in the online trading architecture of aquatic products can provide basic protection for the privacy information of both parties involved in the transaction. However, currently, blockchain-based aquatic product online trading models and systems are suffering from problems such as large data storage loads, high maintenance costs, and low data query efficiency. To further alleviate the above problems, based on the sorting and analysis of the aquatic product trading process, and according to the technical requirements of aquatic product trading business, a trading matching model for aquatic products based on multi-chain storage optimization was proposed. This model achieved a highly efficient multi-attribute online trading matching process for aquatic products through greedy algorithms in smart contracts, and constructed a multi-chain architecture for aquatic product online trading through blockchain multi-channel technology, achieving distributed storage of user transaction information and thus improving the efficiency of transaction information query. Meanwhile, this trading matching model adopted a dual storage technology of blockchain and local database, which alleviated the load of massive data storage at various nodes in the blockchain network. Then, a prototype system for aquatic product online trading based on multi-chain storage optimization was implemented on the Hyperledger Fabric platform. The performance test results of the prototype system indicated that it took 900s to complete 1296 transaction matchings on average, indicating that the system can operate normally when processing a volume of thousands of transaction data, meeting the needs of the online trading platform for aquatic products. At the same time, when storing 1600 contract information on the chain, the average time to query a user’s contract information was 4.018s, which indicated that the multi-chain data storage structure improved the speed of on-chain data queries.

Abstract:Field experiment was conducted in 2023 to investigate the effects of different straw returning methods with N fertilization on soil CO2-emission and carbon balance in black soil maize fields. The experiment included three straw return methods: straw leaving the field (S0, control), straw mulching returning to the field (S1), and straw rotary tillage returning to the field (S2), and two modes of N fertilization: regular N fertilizer application (N, 250kg/hm2) and no N fertilizer application (W, 0kg/hm2, control), totaling six treatments. The soil CO2-emission flux and the contents of soil organic carbon (SOC), dissolved organic carbon (DOC) and microbial biomass carbon (MBC) after maize harvest were measured under different treatments, and the relationship between soil CO2-emission and the contents of SOC, DOC, MBC were explored, and the carbon balance of black soil maize field ecosystem was analyzed. The results showed that the cumulative soil CO2-emissions in each treatment were S2N, S1N, S0N, S2W, S1W,S0W, and the cumulative soil CO2-emissions of S2N treatment were increased significantly by 70.31% compared with S0W treatment (P<0.05). Under the same nitrogen application mode, straw returning to the field effectively increased the contents of SOC, DOC, MBC. Additionally, the cumulative soil CO2-emissions were positively correlated with the contents of SOC, DOC, MBC. Under different straw returning methods and nitrogen fertilizer, the yield of maize in S1N treatment was the highest (13534.4kg/hm2), and the crop carbon emission rate was the lowest (0.122kg/kg). The carbon balance value of black soil maize field ecosystem under different straw returning methods and nitrogen fertilizer application were all positive, showing a strong carbon “sink”, and the carbon balance value and soil carbon sequestration potential of S1N treatment were the largest, which were increased by 13.12%~94.05% and 3.49%~25.32%, respectively, compared with other treatments. In summary, the results of this experiment suggest that straw mulching and returning to the field, in combination with conventional nitrogen application (S1N), can effectively achieve the goals of soil carbon sequestration, emission reduction, and increased crop yield in black soil corn fields.

Abstract:To investigate the impact of water and nitrogen management on the carbon budget of rice field ecosystems, field experiments were conducted, with two irrigation modes: dry-wet-shallow irrigation (D) and flooded irrigation (F), as well as three fertilization levels: 110kg/hm2 (local fertilization standard, N1), 99kg/hm2 (nitrogen reduction of 10%, N2), and 88kg/hm2 (nitrogen reduction of 20%, N3). The dry matter quality and carbon content of different organs of rice harvested in residual rice fields were observed, the CO2 and CH4 emissions fluxes from rice fields were simultaneously monitored, and the net primary productivity (NPP) of rice and the net ecosystem primary productivity (NECB) of rice fields were calculated. The results showed that water and nitrogen management would affect the dry matter and carbon content of various organs in rice plants. Among all treatments, the NPP of DN2 treatment was the highest (8918.02kg/hm2), and the NPP of dry-wet-shallow irrigation mode was greater than that of flooded irrigation mode, increasing by 12.13%, 36.73%, and 8.01%, respectively. The dry-wet-shallow irrigation mode increased the CO2 emission flux of rice field soil respiration, reduced the application of nitrogen fertilizer reduced the CO2 emission flux, and reduced the application of nitrogen fertilizer in dry-wet-shallow irrigation reduced the CH4 emission flux. The total emissions of CO2 and CH4 from each treatment under both irrigation modes dwere ecreased with the decrease of nitrogen fertilizer application. The total emissions of CH4 from each treatment under flooded irrigation mode were significantly higher than those under dry-wet-shallow irrigation mode (P<0.05). The net carbon budget of the rice field ecosystem under each treatment was positive,and the rice field ecosystem with high residual stubble in the black soil area showed a carbon “sink”, with the highest NECB of 1950.96kg/hm2 in the DN2 treatment. Overall, the carbon sink of the rice field ecosystem treated with dry-wet-shallow irrigation mode and 10% nitrogen reduction treatment was the strongest, and the research results can provide theoretical reference and technical support for the protection of black soil in cold regions.

Abstract:Taking Sanjiang Plain, a typical black soil region, as the study area, and the least squares method was used to reveal the change trend of terrestrial ecosystem carbon stocks under climate change in 1990—2020. By means of barycentric analysis and cold hot spot analysis, the spatial and temporal differentiation of terrestrial ecosystem carbon stocks in study area was revealed, and the influence of climate factors on carbon stocks was analyzed by geographical weighted regression method. CMIP6 data were used to simulate and predict terrestrial ecosystem carbon stocks under SSP585 and SSP245. The results showed as follows: under the background of climate change, carbon storage in study area showed a fluctuating trend of decline in 1990—2020, and carbon storage loss was about 2.66×107t. The regions with high carbon storage value were mainly distributed in the northwest and east, while the regions with low carbon storage value were mainly distributed in the north and southeast. The rate of carbon storage decline in the study area was changed greatly in 1990—2020, and the rate of carbon storage decline in Hegang City was the largest, while the rate of carbon storage decline in Jiamusi City was the minimum. The center of gravity of carbon storage in study area was shifted 1340m to the northeast in 1990—2020, and 1680m to the northeast in 2000—2020. The distribution pattern of carbon storage hotspots was sheet-like and blocky, and the distribution pattern of cold spots was sheet-like and banded, the range of hot spots was basically unchanged, and the range of cold spots was reduced. The relationship between annual temperature, annual precipitation and carbon storage in study area was significant in 1990—2020, showing a positive and negative interleaving spatial pattern. Climate and land use change determined the spatio-temporal pattern of carbon storage, and forest-forest land and cultivated land-cultivated land led to the greatest loss of carbon storage. Under the SSP585 and SSP245 scenarios, the carbon reserves in study area in 2030 were 2.22×107t and 2.26×107t, respectively, which were reduced by 2.17×107t and 2.13×107t compared with that in 2020, respectively. The spatial distribution pattern of carbon reserves was not changed significantly, but the areas of cold hot spots were reduced. The center of gravity would continue to shift 6525m and 6000m to the northeast.

Abstract:In the biogas project, determining how to treat the biogas slurry with large volume is one of big issues. Additionally, biogas upgrading is always required for increasing the heat value of biogas. A concept was proposed by coupling the water recovery using forward osmosis and CO2 chemical absorption, in which the CO2-rich solvent from CO2 chemical absorption process was adopted as the draw solution to recover water from biogas slurry in forward osmosis. The feasibility of water recovery from biogas slurry by using CO2-rich solvent as the draw solution in addition to concentrating the biogas slurry was experimentally investigated. In terms of the water flux, concentration ratio of biogas slurry, rejection ratio of ammonia nitrogen and CO2 absorbent transfer flux from the draw solution to biogas slurry, effects of the type and mass fraction of draw solution and other key operation parameters on water recovery performance in forward osmosis were explored. Results showed that CO2-rich solvent can be adopted as the draw solution to successfully reclaim water from biogas slurry and simultaneously concentrate the biogas slurry in forward osmosis. Additionally, the water flux transferred from biogas slurry from the draw solution increases with the mass fraction, flow rate and temperature of draw solution. Correspondingly, the concentration ratio of biogas slurry was increased, however, the rejection ratio of ammonia nitrogen in biogas slurry was decreased and the absorbent transfer flux of CO2 absorbent was increased. When the CO2-rich potassium glycinate with 2.5mol/L and 0.5mol/mol CO2 loading was adopted as the draw solution in the forward osmosis, the initial water flux from biogas slurry was about 8.05L/(m2·h) at the conditions of 70℃ and 150mL/min for draw solution, and ambient temperature and 150mL/min for biogas slurry. After 4h operation, the concentration ratio and ammonia nitrogen rejection ratio of biogas slurry can reach 1.18 and 84.13%, respectively. Notably, the CO2 absorbent transfer flux was only 2.94g/(m2·h). The research result may provide a technical support for concentrating biogas slurry and recovering water from biogas slurry.

Abstract:Aiming at the insufficient conditions of the current research platform for the internal heating mobile bed biomass pyrolysis carbonization test, an internal heating up and down suction mobile bed pyrolysis carbonization test device was proposed, the continuous pyrolysis carbonization process was simulated, and the gas supply system, carbon sampling collection system, product collection system, gas purification system and temperature monitoring system were designed. Using rice husk as raw material, up-suction and down-suction pyrolysis carbonization experiments were carried out to study the effects of residence time and inspiratory mode on internal thermal pyrolysis carbonization products. The results showed that with the increase of residence time, the volatile and fixed carbon contents of rice husk carbon were decreased, the ash content was increased, and the calorific value was decreased. The calorific value of rice husk carbon was decreased from 20.7MJ/kg and 22MJ/kg at 10min to 14.6MJ/kg and 15.2MJ/kg at 60min, respectively. With the increase of residence time, the content of C in rice husk carbon was decreased from 51.9% and 65.3% at 10min to 36.5% and 54.8% at 60min, respectively. The content of N element was decreased from 1.1% and 2.3% at 10min to 0.5% and 0.7% at 60min, respectively. The content of O element was increased from 46.0% and 28.7% at 10min to 61.7% and 42.2% at 60min, respectively. The change trend of H element was not obvious. With the increase of residence time, the specific surface area of up-suction and down-suction pyrolysis rice husk charcoal was increased from 0.73m2/g and 0.78m2/g at 10min to 3.84m2/g and 3.95m2/g at 60min, respectively, and the pore structure of biochar was developed. The test system had reasonable structure, stable and reliable operation, good sealing effect, and can effectively control the test factors such as air intake mode, air intake volume and heat preservation carbonization time. The development of the internal thermal pyrolysis charring platform provided an important condition support for the experimental research of internal heating charring process parameters.

Abstract:In view of the problems of poor cutting quality, high breakage rate and low degree of automation, the structure of potato cutting machine was improved. Laser sensor and feed shunt device were used to detect the feed speed and regulate the feed speed to improve the feed uniformity and cutting stability. Taking the speed of centrifugal drum, the inclination angle of returning blade and the installation angle of cutting blade as test factors, and the cutting pass rate as test index, the three factors and three horizontal response surface tests were conducted. The test results were analyzed by Design-Expert 10.0.1 software, and the influence of interactive factors on the test index was analyzed through the response surface. The optimal combination of test factors and the verification test were obtained. The results showed that when the speed of centrifugal drum was 224.6r/min, the inclination angle of returning blade was 19.4°, and the installation angle of slicing knife was 292°, the qualified rate of cutting strip was 96.7%, and the damage rate of dividing cutting was 2.0%. On this basis, the feed speed control and adjustment test was carried out. The test results showed that the qualified rate of cutting strip was 96.2%, the breakage rate was 2.4%, the yield was 3.3t/h, the qualified rate of cutting was increased by 7%, the breakage rate was reduced by 4%, the cutting efficiency was increased by 28%, and the cutting quality and efficiency of the potato cutting machine were greatly improved.

Abstract:Room temperature processed cheese products, referring to the sterilized cheese, can be stored at room temperature after being treated by ultra-high temperature treated (UHT). It is a hot spot in China’s dairy industry. However, processed cheese is prone to instability after UHT treatment, resulting in gel structure reconstruction, water precipitation and other texture problems during storage, which affect the product quality. It was intended to clarify the effect of protein raw materials on the water precipitation of room temperature processed cheese products and its mechanism. Three different protein raw materials: membrane filtration casein micelle, concentrated milk protein and rennet casein were screened, and their protein content and composition as well as their physical and chemical properties such as particle size, potential, hydration and calcium ion distribution were analyzed. Secondly, the effects of protein raw materials on the water precipitation and texture properties of room temperature processed cheese products were analyzed. The mechanism of the effect of protein raw materials on the water precipitation of room temperature processed cheese products was analyzed from the aspects of water distribution and microstructure. The results showed that there were significant differences in protein composition among the three protein materials. The whey protein content of concentrated milk protein was 7.81 percentage points higher than that of membrane filtration casein micelles, while rennet casein did not contain whey protein. In terms of hydration, concentrated milk protein and membrane filtered casein micelles were similar and significantly higher than rennet casein. Protein raw materials significantly affected the water precipitation of room temperature processed cheese products. When stored for 90d, the water precipitation rate of room temperature processed cheese products prepared from rennet casein was the highest, while that of room temperature processed cheese products prepared from concentrated milk protein was the lowest, and its surface water precipitation rate and centrifugal water precipitation rate were 0.42% and 1.10%, respectively. The results of water distribution and microstructure showed that the protein raw material affected the water precipitation of room temperature processed cheese products by affecting the existing state and three-dimensional network spatial structure of water in processed cheese. The effect and mechanism of different protein raw materials on the water precipitation of room temperature processed cheese products was analyzed, which provided a theoretical basis for improving the water precipitation of room temperature processed cheese products.

Abstract:To investigate the effect of cavitation jet pretreatment on soybean protein isolate-resveratrol (SPI-RES) complexes, SPI was subjected to cavitation jet pretreatment (0min, 2min, 4min, 6min, 8min and 10min) before non-covalent binding with RES to form complexes. The binding of SPI to RES was studied through loading amount and encapsulation efficacy. The interaction and binding mechanism between SPI and RES were studied by using fluorescence spectroscopy, Fourier transform infrared spectroscopy, and molecular docking techniques. In addition, the physical and chemical properties and functional characteristics of the SPI-RES complexes were investigated, including ζ-potential, surface hydrophobicity, and antioxidant activity. The results indicated that after a certain period of cavitation jet pretreatment, SPI significantly increased the encapsulation efficacy and loading amount of RES, and the particle size and ζ-potential of complexes was decreased and increased respectively. The fluorescence spectrum indicated that the quenching of SPI by RES was static quenching, and the reaction was spontaneous. Fourier transform infrared spectroscopy indicated that appropriate cavitation jet pretreatment promoted the transition of SPI from ordered to disordered structures, thereby combining more RES. The thermodynamic parameters and molecular docking results indicated that hydrophobic interactions were the main forces and involved hydrogen bonds. In addition, after appropriate cavitation jet pretreatment, the surface hydrophobicity and antioxidant activity of SPI-RES complexes were increased. The research result can provide a preliminary theoretical basis for the exploration of the application field of cavitation jet pretreatment of soybean protein isolate and the development of fat-soluble active substances in health food.

Abstract:The full supply chain of fruits and vegetables not only has the characteristics of complex business processes, heterogeneous data sources, and numerous interest roles in the general food supply chain, but also has the characteristics of complex risk evolution trends, strong timeliness requirements, and the presence of sudden new risks. To effectively address the pain points of traditional fruit and vegetable supply chain supervision solutions, such as insufficient regulatory coverage, long traceability effect cycles, and poor collaborative effects among various entities, a blockchain based full-process and information management model for the entire process of fruits and vegetables was constructed, and a systematic acceptance was conducted. Firstly, based on the analysis of holographic information throughout full-process and information of fruits and vegetables, a multi-chain-based holographic information management model for fruits and vegetables was constructed. Then a fast retrieval method for fruit and vegetable information based on block comprehensive index was proposed, and a cross chain secure interaction mechanism for fruit and vegetable information based on notary chain was designed. Secondly, a prototype system for the full-process and information management of fruits and vegetables was designed and developed based on the open-source blockchain platform of chainMaker. Finally, experimental verification and system case application analysis were conducted. The results showed that the average time for public data uplink in this system was 589.03ms, the average time for private data uplink was 708.59ms, the average time for public data queries was 26.87ms, and the average time for private data queries was 30.67ms. The multi-chain-based full-process and information management model and system designed not only met the needs of different users with different permission informations for uploading and querying, but also achieved penetrating supervision of full-process and information of fruits and vegetables, and met the needs of enterprises for permission control and secure sharing of private data, improved the efficiency of fruit and vegetable information retrieval, and provided reference for the development of full-process and information management model.

Abstract:The terrain in hilly and mountainous areas is complex and diverse，and the operating environment of agricultural machinery equipment is mainly inclined slopes. Traditional agricultural machinery equipment has low efficiency and poor stability when operating in complex terrain in hilly and mountainous areas，and may even cause serious accidents such as tilting and overturning. A wheel-legged robot platform that can autonomously achieve pose leveling control in complex terrain of hills and mountains was proposed from the perspective of biomimetic mechanical design，improving the stability and safety of operations in complex terrain. Firstly，taking the insect hind foot as the biomimetic mechanical design object，and combining the principle of multi link mechanism，the overall architecture design of the new variable stroke wheel-legged mechanism and robot platform was completed. The kinematics characteristics of the wheel leg mechanism were analyzed by using the D-H parameter method. The results showed that the maximum adjustment of the ground clearance of the wheel-legged robot was 574mm，and it had a strong ability to surmount obstacles. Then the spatial attitude parameters of the wheel-legged robot were defined in the spatial coordinate system，and the spatial attitude model between the fuselage attitude angle and the wheel leg extension was derived. An NSGA-Ⅱ based algorithm for inverse solution of the fuselage spatial attitude was designed. Based on the spatial attitude inverse solution algorithm，an omnidirectional attitude control system for wheel-legged robots was constructed, which included a vehicle body leveling controller，a “virtual leg” compensation controller, and a centroid height controller. It can control the spatial attitude parameters of wheel-legged robots such as pitch angle，roll angle，grounding force，and centroid height when driving in complex terrain. The algorithm simulation verification of the pose control system was completed through the ADAMS-Matlab joint simulation model of the wheel-legged robot. Finally，experiments on automatic adjustment of ground clearance and omnidirectional posture adjustment of the vehicle body were conducted on the robot prototype. The results showed that the maximum adjustment amount of the ground clearance of the test prototype was 574mm，and the omnidirectional automatic leveling of the vehicle body posture could be achieved in complex terrain. The average leveling time was about 1.2s，and the average leveling error was 0.8°. The response speed and leveling accuracy of the posture control could meet the actual work requirements.

Abstract:In order to improve the stability of the tractors in hilly and mountainous areas, a motion control method was designed based on improved genetic algorithm, which can achieve real-time leveling control according to terrain conditions. Firstly, according to the kinematic relationship of the tractor attitude adjustment mechanism, a kinematic model representing the relationship between the wheel center position and the body attitude parameters was established, and a numerical example was given to solve the problem to verify the correctness and accuracy of the kinematic model. Then, with the aim of improving the stability of tractor body, a tractor motion control method based on improved genetic algorithm was designed while considering the kinematics model. Finally, the simulation results showed that the motion control algorithm can effectively reduce the body attitude angle, the maximum lateral slope angle was reduced by 13.3°, and the maximum pitch angle of the longitudinal slope was reduced by 4.3°. The comprehensive adjustment on the road with both slopes can reduce the maximum side angle and the maximum pitch angle by 13.8° and 4°, respectively, which greatly improved the stability of the body. At the same time, the improved genetic algorithm was compared with the traditional genetic algorithm. The results showed that the improved genetic algorithm was superior to the traditional genetic algorithm both in response time and control accuracy. The response time of the improved genetic algorithm was reduced by 63.93% compared with that of the traditional genetic algorithm, and the efficiency of the algorithm was greatly improved.

Abstract:Based on the widely used advantages of low degree of freedom parallel mechanisms (PM), a new-type three translation one rotation (3T1R) parallel mechanism was proposed, which had the characteristics of simple configuration, symmetrical structure, and high positioning accuracy. It can be applied to small-scale precision operations, as well as large-scale transportation, sorting, and spraying major industrial fields. Based on the topology analysis theory of orientation characteristic equation, the azimuth feature set of position and orientation characteristic (POC) was obtained, and the type and number of degrees of freedom (DOF) of the parallel mechanism were analyzed and verified. A kinematic model was established based on the closed-loop vector method, and the rationality of kinematics was verified through a positional forward and inverse solution example and the results indicated that the kinematic model was completely correct. The reachable workspace’s three-dimensional point cloud picture of the 3T1R parallel mechanism was analyzed by using the limit boundary search method based on the position inverse solution equation. A velocity Jacobian matrix was established through velocity analysis, and based on it, the positioning accuracy and operability performance indicators of the mechanism were analyzed. The PCA/PSO algorithm was used to optimize the design of the reachable workspace, positioning accuracy and operability as three performance indicators, and the optimization results were compared and analyzed. The final optimized reachable workspace volume was increased from 0.2933m3 to 0.4231m3, the positioning accuracy error amplification factor was decreased from 15.5044 to 4.4308, and the operability index was decreased from 9.7027 to 1.3996.

Abstract:The 6-DOF vibration calibration system based on direct drive parallel mechanisms can achieve high precision and multi-degree-of-freedom motion simulation and calibration, showing promising application prospects. To address the issue of low tracking accuracy of slider positions in the driven joint of the platform, a dynamic torque feedforward compensation analysis was conducted on this mechanism. Firstly, the dynamic model of the mechanism was established by using the principle of virtual work. Then, traditional servo control was done based on motion controller. A control strategy was proposed that combined a fundamental servo algorithm with dynamic feedforward compensation. Experimental torque compensation was conducted on the prototype, verifying the effectiveness of dynamic torque feedforward compensation in enhancing the position tracking accuracy of each driven joint slider and accelerating the response speed of the driven joint sliders. Experimental results demonstrated that by incorporating torque feedforward compensation into the open-loop servo program, the tracking errors of the mechanism during motion can be reduced. Specifically, during single-degree-of-freedom sinusoidal motions with different amplitudes, the tracking errors were decreased by 40.32%, 39.04%, 43.24%, and 48.19%, respectively. Furthermore, performance testing experiments were carried out on the vibration platform. A laser measurement system and data acquisition module were set up on the platform to perform sensor calibration and performance analysis.

Abstract:The topology design, kinematics and dynamics analysis of parallel mechanism with redundant branches but no parasitic motion were studied, and finally a conceptual application scenario was put forward based on the advantages of the mechanism itself. Firstly, a parallel robot with redundant branches but no parasitic motion was designed by using the type synthesis theory based on position and direction characteristics. The robot can realize translation in two directions and rotation in one direction. After topological analysis of the mechanism, it was found that the output motion of the mechanism met the expected design requirements, and it had partial motion decoupling characteristics, so it was easy to control the output of the moving platform through the driving pair. Secondly, the forward kinematics and inverse kinematics of the mechanism were obtained by using the kinematics modeling method based on topological characteristics. The singularity of the mechanism was analyzed by inverse solution, and the workspace was determined by positive position solution. In addition, the ordered single open chain method based on virtual work principle was used to model the dynamics of the mechanism. By solving the model, the driving force of driving joints on the static platform and the binding force at the joints connected by two sub-kinematic chains (SKC) were obtained. The simulation values of these two forces were obtained in the simulation software and compared. The results showed that the error between the simulation values and the theoretical values was small, so it was considered that the dynamic model established by the ordered single open chain method was correct. Finally, combined with the existing improvement of the moving target in shooting training, the conceptual design of parallel mechanism for military training application scenarios was given.

Abstract:Linear conjugated internal gear pumps are widely used in eletro-hydrostatic actuating systems as power elements with their efficient and low-noise performance. A linear conjugated internal gear pump simulation model was established by the lumped parameter method, the fluid domain within a linear conjugated internal gear pump can be specified as a finite number of volumes where the fluid properties (e.g, pressure, density, and modulus of elasticity, etc.) can be considered as concentrated variables. A test platform was constructed to measure the suction and discharge pressure pulsations of the linear conjugated internal gear pump. The excitation sources, including the pressure pulsation in the inlet and outlet ports, the pressure in the tooth cavity, and the radial force of the gear and tooth ring in the x and y directions were calculated and analyzed. The results demonstrated that the lumped parameter model of linear conjugate internal gear pump established had good accuracy and reliability. The magnitude of the radial force on the gear and tooth ring varied periodically with the deflection angle. In the x direction, the radial force acting on the tooth ring was pointed to the low-pressure side, and the radial force acting on the gear was pointed to the high-pressure side. In the y direction, both theradial forces acting on the gear and ring were fluctuated between positive and negative values. The radial forces in the x direction were smaller than those in the y direction at the fundamental harmonic;the phenomenon of oil trapping led to a slight increase in the pressure in the tooth cavities. The research result can provide a reference for the optimized design and vibration noise analysis of linear conjugated internal gear pumps.

Abstract:Considering the electro-hydraulic composite braking system of electric loader, a hydraulic braking system based on hydraulic brake valve with regenerative braking free stroke was proposed to meet the braking requirements of multi-working conditions and ensure braking safety. The stroke of hydraulic brake valve was divided into motor regenerative brake free stroke and hydraulic brake stroke, which was one of the judgment conditions for hydraulic brake system intervention. Based on the ideal curves of front and rear wheel braking force distribution for electric loader and the braking strength obtained from the braking intention identification, a distribution curve between braking intensity and the vehicle torque demand was developed. To further improve the coordination of distribution for regenerative braking force and hydraulic braking force while taking into account energy recovery efficiency, a collaborative control strategy of electro-hydraulic composite braking was proposed. When hydraulic braking system was involved, the regenerative braking torque was changed with the hydraulic braking torque, reducing the fluctuation of vehicle total braking torque and ensuring smoothness during the braking mode switching process. Finally, the co-simulation model based on AMESim-Matlab/Simulink was studied, and an experimental prototype was built to verify the feasibility of collaborative control strategy of composite braking for electric loader. The results showed that the energy recovery efficiency was up to 71.6%, the recovery rate of braking was up to 44.5%, and the energy saving was up to 7.6% in one working cycle. The proposed control strategy had good braking performance and energy recovery efficiency.

Abstract:In response to the problems of low recognition rate and unstable response in traditional gesture recognition systems, a flexible strain sensor data glove gesture recognition system was developed, which included flexible sensors, signal acquisition systems, and gesture recognition algorithms. The system can accurately capture the motion information of each finger joint, and had the characteristics of high degree of freedom, low cost and high recognition rate. Carbon black (CB) and carbon nanotubes (CNTs) were doped into soft silica gel, and a resistive sensor with good linearity and high sensitivity was designed by extension technology. The experimental results showed that the sensor had good static and dynamic response characteristics, and the sensor calibration was completed. Using multiple flexible sensors to prepare data gloves and build a signal acquisition system, a gesture recognition method combining BP neural network and template matching technology was further proposed to improve the recognition rate of similar gestures, and the recognition rate of the algorithm was 98.5%. Gesture recognition experiments were carried out for different groups of people. The results showed that the accuracy of the gesture recognition system reached 92.8%, and the response time was about 40ms. The data glove had good application potential.