Automatic driving agricultural machine marshalling operation control method and system

Abstract

The invention provides an automatic driving agricultural machinery marshalling operation control method and system, and relates to the technical field of agricultural machinery control, and the method comprises the steps: determining the function type and number configuration of needed agricultural machinery according to the farming task type, farmland scale and terrain parameters; based on the agricultural machinery operation capability parameter set and the farmland space features, generating agricultural machinery cooperation groups; planning an operation path containing space-time collision avoidance constraint for each agricultural machine group, and establishing a time sequence scheduling strategy matched with the task load; based on farmland environment data, agricultural machinery working condition data and operation quality data which are acquired in real time, agricultural machinery marshalling topology and operation parameters are dynamically reconstructed, through dynamic grouping, intelligent path planning and real-time state feedback and reconstruction, the flexibility, efficiency and robustness of agricultural machinery collaborative operation are remarkably improved, and the development requirements of precision and intelligence of modern agriculture are met.

Description

Technical Field

[0001] The present invention relates to the field of agricultural machinery control technology, and in particular to a method for controlling automatic driving agricultural machinery grouping operations. Background Art

[0002] With the development of precision agriculture and Internet of Things technology, the application of autonomous agricultural machinery in agricultural production has evolved from single-device automation to multi-machine collaborative operation. Traditional agricultural machinery operation systems mainly rely on single-machine autonomous driving technology (such as GNSS-based tractor automatic navigation, seed drill precision operation, etc.), and achieve basic automation through preset paths and fixed operation parameters. However, faced with complex farmland environments (such as large-scale terraces, multi-obstacle plots) and diversified farming tasks (such as sowing, fertilization, and plant protection multi-link coupling), the single-machine operation mode has significant limitations, including: single-machine operation lacks task parallel processing capabilities and is difficult to fully utilize the advantages of multi-device collaboration; static path planning and parameter configuration cannot cope with real-time changes such as soil moisture gradients and dynamic obstacles (such as wild animals); the complementarity of agricultural machinery functions (such as different operating widths and energy consumption characteristics) is not considered, resulting in a mismatch between group configuration and operation requirements. Although existing group operation methods have attempted to introduce multi-machine collaboration, the following key issues still exist:

[0003] Reliance on a single communication link (such as relying solely on 4G or short-range Bluetooth) results in high latency in real-time posture data transmission, and disconnection between cloud-based task allocation and local obstacle avoidance decisions. The use of a fixed grouping mode (such as grouping by similar model) fails to comprehensively consider the dynamic matching of farmland spatial heterogeneity (such as differences in soil hardness blocks) and agricultural machinery performance parameters (such as operation accuracy and energy consumption curves). The path generated by the A* algorithm based on a global static map lacks spatiotemporal collision avoidance constraints. During operation, the grouping topology cannot be adjusted according to real-time soil moisture conditions, crop growth status, or sudden failures, resulting in fluctuations in operation quality. Summary of the Invention

[0004] Based on the above problems, the present invention provides an automatic driving agricultural machinery grouping operation control method and system, which significantly improves the flexibility, efficiency and robustness of agricultural machinery collaborative operations through dynamic grouping, intelligent path planning, real-time status feedback and reconstruction, and adapts to the development needs of modern agriculture with precision and intelligence.

[0005] In a first aspect, the present application provides a method for controlling an autonomous driving agricultural machinery group operation, the method comprising:

[0006] Determine the functional type and quantity of agricultural machinery required based on the type of farming task, farmland size and terrain parameters;

[0007] Generate agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics;

[0008] Plan an operation path for each agricultural machinery group that includes time and space collision avoidance constraints, and establish a time scheduling strategy that matches the task volume;

[0009] Based on real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data, the agricultural machinery grouping topology and operation parameters are dynamically reconstructed.

[0010] Preferably, the method further comprises:

[0011] A two-layer communication architecture consisting of an execution layer and a decision-making layer is constructed. The execution layer establishes a direct communication link between agricultural machinery and the vehicle controller via data transmission equipment, and the decision-making layer establishes a cloud relay link via wireless communication with the task machine. Agricultural machinery posture data is transmitted in real time via the execution layer direct link, and operation status data is transmitted via the decision-making layer relay link. The ground station interacts with the execution layer and the decision-making layer via data transmission equipment, realizing real-time transmission of agricultural machinery data under the two-layer communication architecture.

[0012] The multimodal sensors installed on agricultural machinery collect operation quality indicators and make feedback adjustments based on preset standards.

[0013] Preferably, the determination of the functional type and quantity configuration of the required agricultural machinery according to the farming task type, farmland size and terrain parameters includes:

[0014] Analyze the characteristics of farming tasks to obtain operation parameters; the operation parameters include operation depth, row spacing requirements, and pesticide dosage;

[0015] Generate constraints on operations based on farmland parameters, including operating area, average slope, soil compaction, field flatness, and obstacle density;

[0016] Modifying the operation parameters according to the constraint conditions;

[0017] Based on the revised operating parameters, the performance parameter range of agricultural machinery that meets the operating requirements is screened out;

[0018] The number of agricultural machinery required is calculated based on the farmland area, working depth, working speed, daily working time and equipment redundancy factor.

[0019] Preferably, the generating of agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics comprises:

[0020] Based on cluster screening of agricultural machinery performance parameters and synergy parameters, candidate groups with complementary functions and stable energy consumption are formed;

[0021] The particle swarm algorithm is used to optimize the matching relationship between groups and farmland sub-areas, and the optimal agricultural machinery grouping and operation area allocation plan is determined by comprehensively considering functional complementarity, coverage synergy and energy efficiency indicators.

[0022] Preferably, the clustering and screening based on the agricultural machinery performance parameters and the synergy parameters to form a candidate group with complementary functions and stable energy consumption includes:

[0023] Establishing a parameter set for the operation capability of each agricultural machine through the performance parameters and coordination parameters of each agricultural machine; the performance parameters include functional modules, operation speed, operation accuracy, width and energy consumption; the coordination parameters include communication delay, positioning accuracy and synchronization error;

[0024] Clustering the agricultural machinery according to the parameter sets of their operating capabilities to obtain initial groups, and calculating the functional complementarity scores in each initial group; determining the first candidate group based on the complementarity scores and the types of operations covered;

[0025] Based on the energy consumption prediction model, the energy consumption of each agricultural machine in the first candidate group is obtained, and the groups whose energy consumption standard deviation is less than a preset first threshold are screened to form a second candidate group.

[0026] Preferably, the particle swarm algorithm is used to optimize the matching relationship between the groups and the farmland sub-areas, comprehensively consider functional complementarity, coverage synergy, and energy efficiency indicators, and determine the optimal agricultural machinery grouping and operation area allocation scheme, including:

[0027] The farmland is divided into initial operation sub-areas, the spatial feature vector of the farmland is constructed, and the similarity between the initial sub-areas is calculated; the spatial features include average slope, soil hardness, obstacle density, and field connectivity;

[0028] Merge adjacent sub-regions whose similarity is greater than a preset threshold according to the similarity between the initial sub-regions and calculate the collaborative coverage area of the merged region;

[0029] The particle swarm algorithm is used to iteratively optimize the random matching matrix of the second candidate grouping and the merged area. The functional complementarity, coverage synergy and energy efficiency of the grouping scheme are evaluated through the fitness function. Finally, the agricultural machinery grouping and operation area allocation are determined by the matrix corresponding to the global optimal solution.

[0030] Preferably, the method of planning an operation path for each agricultural machinery group including time-space collision avoidance constraints and establishing a time sequence scheduling strategy that matches the task volume includes:

[0031] The operation area is divided into multiple path planning units according to the spatial characteristics of the farmland, and a topological map of the nodes accessible to agricultural machinery is constructed;

[0032] Based on the operation sequence constraints and task priorities of the agricultural machinery group, a time window constraint is injected into the path node sequence of each agricultural machinery group;

[0033] Selecting a path generation strategy based on a target tracking mode to generate an initial collision-free path and performing speed adjustment or path replanning on overlapping path segments using a spatiotemporal conflict detection model; the target tracking mode includes a fastest target tracking mode or a neighboring target tracking mode;

[0034] Establish a time series scheduling model based on dynamic time warping, and calculate the operating time window and task segmentation strategy of each agricultural machinery group according to the agricultural machinery group's operating efficiency, remaining task volume and energy endurance threshold;

[0035] The feasibility of scheduling is verified through discrete event simulation, and local path re-optimization and task reallocation mechanisms are triggered when path conflicts or task delays are detected;

[0036] During the operation, the path tracking control parameters are dynamically adjusted based on the real-time posture data of the agricultural machinery and the deviation value of the topology map, and the task status mark in the timing scheduling queue is updated.

[0037] Preferably, the dynamic reconstruction of the agricultural machinery grouping topology and operation parameters based on the real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data includes:

[0038] The soil sensors, visual sensors, and positioning modules on agricultural machinery collect farmland environmental data in real time, including soil moisture gradient, light intensity distribution, dynamic obstacle locations, and crop growth status;

[0039] Obtain operating condition data through the agricultural machinery controller bus, including engine load rate, hydraulic system pressure, remaining energy endurance and fault codes;

[0040] The sensors of the operation actuators collect operation quality data, including standard deviation of sowing depth, fertilization uniformity and pesticide coverage;

[0041] Design an anomaly detection mechanism based on a sliding window to perform time series alignment and feature extraction on the environmental data, working condition data, and quality data. Use preset thresholds and machine learning models to identify agricultural machinery performance degradation, sudden obstacle intrusion, and operational quality deviation events, and generate abnormal event labels and impact range assessment results.

[0042] Trigger the dynamic reconstruction decision-making process and select the group topology adjustment mode based on the type and severity of the abnormal event.

[0043] Preferably, the selecting of the marshaling topology adjustment mode according to the abnormal event type and severity level includes:

[0044] When a single machine failure is detected, a topology reconstruction algorithm based on task migration is used to decompose the faulty agricultural machinery task into subtask sets, which are then redistributed to adjacent groups based on the remaining agricultural machinery load capacity and spatial accessibility, and the spatiotemporal constraints of the collaborative paths between groups are updated.

[0045] When a sudden change in the environment is detected, an adaptive sub-region redivision strategy is adopted to dynamically adjust the operating boundaries of the agricultural machinery group based on obstacle contours and changes in soil moisture conditions. A reinforcement learning model is then used to generate an optimal path compensation plan to bypass the sudden change area.

[0046] When quality deviation is detected, a multi-objective optimization engine is used to adjust the operating speed, operating width, and actuator parameters of the agricultural machinery group online, using the operation quality indicators as constraints. Parameter synchronization is achieved through closed-loop control between agricultural machinery within the group.

[0047] Deploy a dynamic parameter collaborative controller to generate a control sequence based on the reconstructed group topology, including agricultural machinery posture compensation, operating parameter correction, and communication link switching instructions. A distributed consensus algorithm is used to ensure the consistency of agricultural machinery control instructions within the group. The reconstruction scheme is pre-verified through the digital twin system and then issued to the actual agricultural machinery for execution.

[0048] A reconstruction effect feedback loop is established to continuously monitor the marshalling operation stability indicators and task completion progress within a preset time window. If a secondary anomaly or progress lag is detected, incremental local reconstruction is triggered; otherwise, the current marshalling topology and parameters are updated to the baseline configuration.

[0049] In a second aspect, the present application provides an automatic driving agricultural machinery grouping operation control system for implementing any of the methods described in the present application, the system comprising:

[0050] The agricultural machinery configuration module is used to determine the functional type and quantity of the required agricultural machinery according to the farming task type, farmland size and terrain parameters;

[0051] The grouping module is used to generate agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics;

[0052] The planning and scheduling module is used to plan an operation path for each agricultural machinery group, including time and space collision avoidance constraints, and establish a time sequence scheduling strategy that matches the task volume;

[0053] The reconstruction module is used to dynamically reconstruct the agricultural machinery grouping topology and operation parameters based on the real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data.

[0054] Compared with the prior art, the beneficial effects of the present invention include at least the following: first, based on the specific type of farming task (such as sowing, fertilizing, harvesting, etc.), the scale of the farmland (area, shape), and terrain parameters (slope, obstacle distribution, etc.), the functional type of agricultural machinery required (such as tractors, seeders, harvesters, etc.) and its quantity configuration are evaluated and determined to ensure that the type and quantity of agricultural machinery are accurately matched with the operation task to avoid waste or shortage of resources. Based on the operating capacity parameters of each agricultural machine (such as operating speed, operating width, power, etc.) and the spatial characteristics of the farmland (such as plot division, operation area priority, etc.), an algorithm is used to generate an agricultural machinery collaborative grouping scheme. The principle of grouping is to maximize operating efficiency while taking into account the compatibility between agricultural machinery to ensure that each group of agricultural machinery can work efficiently and collaboratively to cover the entire farmland area; a detailed operation path is planned for each agricultural machinery group. The path not only takes into account the terrain and obstacles of the farmland, but also incorporates time and space collision avoidance constraints, that is, to ensure that different agricultural machinery groups or single machines will not have time and space conflicts during the operation process. At the same time, the system establishes a time-series scheduling strategy that matches the workload, rationally arranging the order and time of each group's work to balance the workload and improve overall work efficiency. During the operation process, the system collects real-time farmland environmental data (such as soil moisture and crop growth), agricultural machinery operating data (such as fuel consumption and mechanical wear), and work quality data (such as sowing uniformity and fertilization accuracy). Based on this data, the system dynamically reconstructs the topological structure of the agricultural machinery grouping (such as adjusting the grouping method and increasing or decreasing the number of agricultural machinery) and optimizes operating parameters (such as operating speed and sowing density) to adapt to changing environmental conditions and work requirements, ensuring work quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Schematic diagram of a data-driven, autonomous driving agricultural machinery grouping operation control method according to an embodiment of the present invention;

[0056] Figure 2 is a schematic diagram of a communication architecture according to an embodiment of the present invention;

[0057] Figure 3 is a schematic diagram of a path planning strategy according to an embodiment of the present invention;

[0058] Figure 4 This is a schematic diagram of collaborative operation path planning according to an embodiment of the present invention;

[0059] Figure 5 4 is a schematic diagram of the iterative optimization process of the particle swarm algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In view of the deficiencies in the prior art, the applicant of this case has proposed the technical solution of this application after long-term research and extensive practice. The following will further explain the technical solution, its implementation process and principles, etc. in conjunction with the drawings in the embodiments of this application and specific implementation cases.

[0061] It should be noted that the embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be understood as limiting this application. The embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, this application covers any replacement, modification, equivalent method and scheme made within the spirit, principle and scope of this application defined by the claims. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0062] Furthermore, in order to provide the public with a better understanding of the present application, certain specific details are described in detail in the detailed description of the present application below, and those skilled in the art can fully understand the present application without the description of these specific details.

[0063] Example 1: Refer to the attached Figure 1 The present application provides a method for controlling an automatic driving agricultural machinery grouping operation, the method comprising:

[0064] Determine the functional type and quantity of agricultural machinery required based on the type of farming task, farmland size and terrain parameters;

[0065] Generate agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics;

[0066] Plan an operation path for each agricultural machinery group that includes time and space collision avoidance constraints, and establish a time scheduling strategy that matches the task volume;

[0067] Based on real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data, the agricultural machinery grouping topology and operation parameters are dynamically reconstructed.

[0068] The working principle and effects of the above technical solution are as follows:

[0069] First, based on the specific type of farming task (such as sowing, fertilizing, and harvesting), the size of the farmland (area and shape), and terrain parameters (slope, obstacle distribution, etc.), the functional types of agricultural machinery required (such as tractors, planters, harvesters, etc.) and their quantity configuration are evaluated and determined to ensure that the type and quantity of agricultural machinery are accurately matched to the task, avoiding waste or insufficient resources. Based on the operating capacity parameters of each agricultural machine (such as operating speed, operating width, power, etc.) and the spatial characteristics of the farmland (such as plot division and operating area priority), an algorithm is used to generate a collaborative agricultural machinery grouping scheme. The principle of grouping is to maximize operating efficiency while considering the compatibility between agricultural machines to ensure that each group of agricultural machines can work together efficiently and cover the entire farmland area. Detailed operation paths are planned for each agricultural machine group. The paths not only take into account the farmland terrain and obstacles, but also incorporate temporal and spatial collision avoidance constraints, that is, to ensure that different agricultural machine groups or individual machines do not conflict in time and space during the operation. At the same time, the system establishes a time-series scheduling strategy that matches the workload, rationally arranging the order and time of each group's work to balance the workload and improve overall work efficiency. During the operation process, the system collects real-time farmland environmental data (such as soil moisture and crop growth), agricultural machinery operating data (such as fuel consumption and mechanical wear), and work quality data (such as sowing uniformity and fertilization accuracy). Based on this data, the system dynamically reconstructs the topological structure of the agricultural machinery grouping (such as adjusting the grouping method and increasing or decreasing the number of agricultural machinery) and optimizes operating parameters (such as operating speed and sowing density) to adapt to changing environmental conditions and work requirements, ensuring work quality and efficiency.

[0070] In a possible implementation, the method further includes:

[0071] A two-layer communication architecture consisting of an execution layer and a decision layer is constructed. The execution layer establishes a direct communication link between agricultural machinery through a data transmission device (Xam device) and a VCU (vehicle controller). The decision layer establishes a cloud relay link through a task machine and WiFi. The agricultural machinery posture data is transmitted in real time through the execution layer direct link, and the operation status data is transmitted through the decision layer relay link. The ground station interacts with the execution layer and the decision layer through the data transmission device (Xam device) to realize the real-time transmission of agricultural machinery data under the two-layer communication architecture. Refer to the attached Figure 2 ;

[0072] The multimodal sensors on agricultural machinery collect operation quality indicators and make feedback adjustments based on preset standards;

[0073] After the operation is completed, a detailed operation report is generated, including data such as the operation area, operation quality, energy consumption, etc., providing a basis for subsequent operation optimization.

[0074] The working principle and effects of the above technical solution are as follows:

[0075] Execution layer communication mechanisms include:

[0076] Direct communication link: Agricultural machinery establishes point-to-point communication through dedicated data transmission equipment (such as Xam devices, CAN bus, ZigBee modules), transmitting posture data (position, speed, heading angle) in real time. The execution layer independently processes local path adjustment instructions (such as speed compensation and steering control), reducing cloud dependence and the impact of communication delays on real-time performance.

[0077] The decision-making layer communication mechanism includes:

[0078] The task machine connects to the cloud server via wireless communication networks (such as Wi-Fi, 5G, and LTE) to transmit operation status data (task progress, energy consumption, and fault codes). The cloud integrates data from multiple units to generate optimized grouping plans, timing scheduling strategies, and dynamic reconstruction instructions.

[0079] The ground station hub serves as a human-computer interaction interface, interacting bidirectionally with the execution layer and decision-making layer through data transmission equipment to achieve the distribution of operation parameters, abnormal event alarms and manual intervention.

[0080] Two-layer data fusion:

[0081] Execution layer data is used for local path tracking and real-time collision avoidance; decision layer data is used for global resource scheduling and long-term mission planning;

[0082] The ground station interactively synchronizes two-layer data, supporting real-time monitoring (such as posture trajectory display) and offline analysis (such as energy consumption statistics).

[0083] Multimodal sensor feedback adjustment principle: Operation quality is optimized through multi-source data collection and closed-loop control, as follows: Vision sensors identify crop row spacing and obstacle outlines (such as rocks and weeds); Soil sensors monitor soil moisture, compaction, and nutrient distribution; Actuator sensors measure sowing depth, fertilizer flow rate, and spray pressure; Multi-source data is synchronized by timestamp to eliminate sensor delay errors; Feature extraction calculates operation quality indicators (such as sowing depth standard deviation and pesticide coverage deviation); Real-time quality indicators are compared with preset thresholds (such as sowing depth ±1cm) to trigger different levels of response: Minor deviations are compensated by adjusting operation parameters (such as reducing vehicle speed and increasing spray volume); Severe deviations trigger path replanning or group reconstruction (such as deploying replanting machinery). The execution layer dynamically corrects the agricultural machinery's actions based on feedback signals (such as adjusting the downforce of the seed drill). The decision layer updates the baseline values of operation parameters to ensure consistency of agricultural machinery parameters within the group; Direct communication at the execution layer ensures rapid exchange of posture data, and cloud computing at the decision layer provides a global optimization solution.

[0084] Sensor data is fed back to the agricultural machinery controller via the execution layer, enabling online adjustment of operating parameters. This data is also synchronized to other groups via the decision layer to prevent the spread of quality fluctuations. If a sensor on an agricultural machine fails, the decision layer can infer its operating status from data from other groups to maintain overall mission execution.

[0085] In one possible implementation, determining the functional type and quantity configuration of required agricultural machinery based on the farming task type, farmland size, and terrain parameters includes:

[0086] Analyze the characteristics of farming tasks to obtain operation parameters; the operation parameters include operation depth, row spacing requirements, and pesticide dosage;

[0087] Generate constraints on operations based on farmland parameters, including operating area, average slope, soil compaction, field flatness, and obstacle density;

[0088] Modifying the operation parameters according to the constraint conditions;

[0089] DA 修正 =DA 基础 ×AD 修正

[0090] DA 修正 is the corrected operating parameter, DA 基础 AD is the basic operation parameter parsed according to the task; 修正 is the correction factor;

[0091] Based on the revised operating parameters, the performance parameter range of agricultural machinery that meets the operating requirements is screened out;

[0092] The number of agricultural machinery required is calculated based on the farmland area, working depth, working speed, daily working time and equipment redundancy factor.

[0093] The feature analysis of the farming task to obtain the operation parameters includes:

[0094] Build a farming task database; build a comprehensive farming task database that includes the characteristics of various types of farming tasks such as plowing, sowing, plant protection, and harvesting. These characteristics can be used as the basis for identifying and analyzing specific tasks.

[0095] Based on the specific task type, the preset task parameter library is called. At the same time, the corresponding correction coefficient is applied in combination with the soil type (such as clay, loam, and sand), and the required parameters for the operation are extracted through the task analysis algorithm; such as operating depth, operating speed, and operating width;

[0096] Generate constraints for operations based on farmland parameters, including:

[0097] Collect relevant farmland data, including but not limited to obtaining field boundaries using satellite remote sensing, constructing digital elevation models (DEMs) using LiDAR point cloud data, and obtaining field uniformity indicators using soil conductivity meters. This collected data is processed to calculate key farmland parameters, such as average slope, soil compaction, field flatness, and obstacle density. Based on these parameters, operational constraints are generated. For example, when the average slope exceeds 15°, operating speed is restricted; when soil compaction exceeds 2 MPa, the power requirements for agricultural machinery are increased.

[0098] The number of agricultural machinery required (NZ) is calculated based on the farmland area, working depth, working speed, daily working time and equipment redundancy factor, including:

[0099]

[0100] V (operating speed), De (operating depth), δ i Farmland constraints (various correction coefficients); η is the equipment redundancy coefficient, which is related to the failure rate; t is the daily operating time; floor() is rounded down.

[0101] The working principle of the above technical solution is: integrating characteristic parameters of typical farming tasks such as plowing, sowing, plant protection, and harvesting (such as sowing row spacing and plant protection agent concentration); establishing task-parameter mapping relationships (such as "corn sowing" corresponds to an operating depth of 5-8cm and a row spacing of 60cm); calling the task parameter library: according to the specific task type (such as "wheat harvesting in hilly areas"), extracting basic parameters from the database (operating speed 3-5km / h); clay requires an increase in the operating depth correction coefficient (such as +0.5cm), and sand requires a reduction coefficient (such as -0.3cm). The principle of generating farmland constraints generates agricultural machinery operation constraints through multi-source data fusion and a constraint rule library.

[0102] Obtain field boundaries and obstacle distribution (such as trees and ditches); construct a DEM model through LiDAR point cloud, calculate the average slope, measure soil compaction (unit: MPa) through soil conductivity meter, and evaluate field flatness.

[0103] Establish a constraint rule library: For example, if the slope is greater than 15°, the upper limit of the operating speed is reduced to 70% of the original speed. If the soil compaction is ≥2MPa, the agricultural machinery power is required to be ≥150 horsepower. Obstacle constraints: If the obstacle density is greater than 3 per hectare, the navigation sensor accuracy needs to be increased to ±2cm.

[0104] Operation parameter correction and agricultural machinery performance screening adjust task parameters based on constraints and match the performance range of agricultural machinery. For example, parameter correction logic: if the field flatness difference is greater than 5cm, the allowable error of the operation depth is expanded to ±1.5cm. If the obstacle density is high, the operation width is reduced by 20% to reduce the risk of collision.

[0105] Establish a matrix of agricultural machinery performance parameters (such as power, operating width, and energy consumption);

[0106] Matching rules: For example, the operating depth correction value must meet the minimum / maximum depth range of the agricultural machinery; the upper limit of the operating speed must be ≥ the corrected speed requirement.

[0107] The calculation principle of the number of agricultural machinery is based on the farmland area, operating efficiency and redundancy coefficient, and the number of agricultural machinery configurations is quantified:

[0108] By regularly updating farmland parameters (such as changes in obstacle locations), soil correction coefficients and redundancy coefficients are optimized based on historical operation data to improve calculation accuracy.

[0109] The above technical solution achieves the following: By building a farming task database, standardized parameters for different task types, such as plowing and seeding, are extracted. Combined with soil type correction factors (e.g., an increased operating depth correction factor for clay soils), this ensures that operating parameters are highly aligned with actual needs, improving operational accuracy. For example, in a corn seeding task, the operating depth correction factor is 1.1 in clay soils and 0.9 in sandy soils, preventing over-deep or shallow seeding due to soil variations. Dynamic environmental adaptability is enhanced by utilizing multi-source data such as satellite remote sensing and LiDAR point clouds to generate farmland constraints (e.g., speed limits for slopes > 15°). This allows agricultural machinery formations to automatically adapt to complex terrain, reducing operational accidents or efficiency losses caused by sudden environmental changes. For example, in hilly areas with an average slope of 20°, the operating speed automatically drops to 4 km / h to ensure machine stability. Improved operational quality stability is achieved by dynamically adjusting operating parameters through correction formulas. For example, when field flatness is poor, the operating depth error is relaxed to ±1.5 cm, ensuring consistent operational quality. By optimizing the allocation of agricultural machinery resources and calculating the number of agricultural machinery based on a quantitative formula, combined with a redundancy coefficient (such as a redundancy coefficient of 1.1 when the failure rate is 5%), insufficient equipment or redundant waste can be avoided.

[0110] In one possible implementation, agricultural machinery collaborative grouping is generated based on a set of agricultural machinery operation capability parameters and farmland spatial characteristics, including:

[0111] Based on cluster screening of agricultural machinery performance parameters and synergy parameters, candidate groups with complementary functions and stable energy consumption are formed;

[0112] The particle swarm algorithm is used to optimize the matching relationship between groups and farmland sub-areas, and the optimal agricultural machinery grouping and operation area allocation plan is determined by comprehensively considering functional complementarity, coverage synergy and energy efficiency indicators.

[0113] In a possible implementation, the clustering and screening based on the agricultural machinery performance parameters and the coordination parameters to form candidate groups with complementary functions and stable energy consumption includes:

[0114] Establishing a parameter set for the operation capability of each agricultural machine through the performance parameters and coordination parameters of each agricultural machine; the performance parameters include functional modules, operation speed, operation accuracy, width and energy consumption; the coordination parameters include communication delay, positioning accuracy and synchronization error;

[0115] Clustering the agricultural machinery according to the parameter sets of their operating capabilities to obtain initial groups, and calculating the functional complementarity scores in each initial group; determining the first candidate group based on the complementarity scores and the types of operations covered;

[0116] Based on the energy consumption prediction model, the energy consumption of each agricultural machine in the first candidate group is obtained, and the groups whose energy consumption standard deviation is less than the preset first threshold are screened to form the second candidate group G.

[0117] In one possible implementation, the particle swarm algorithm is used to optimize the matching relationship between groups and farmland sub-areas, and the optimal agricultural machinery grouping and operation area allocation scheme is determined by comprehensively considering functional complementarity, coverage synergy, and energy efficiency indicators, including:

[0118] The farmland is divided into initial operation sub-areas, the spatial feature vector of the farmland is constructed, and the similarity between the initial sub-areas is calculated; the spatial features include average slope, soil hardness, obstacle density, and field connectivity;

[0119] Based on the similarity between the initial sub-regions, the adjacent sub-regions whose similarity is greater than a preset threshold are merged and the collaborative coverage area of the merged region R is calculated;

[0120] The particle swarm algorithm is used to iteratively optimize the random matching matrix of the second candidate grouping and the merged area. The functional complementarity, coverage synergy and energy efficiency of the grouping scheme are evaluated through the fitness function. Finally, the matrix corresponding to the global optimal solution is used to determine the agricultural machinery grouping and operation area allocation. The iterative optimization process of the particle swarm algorithm is shown in the attached figure. Figure 5 .

[0121] In one possible implementation, a fitness function is constructed by functional complementarity score, collaborative operation coverage area, and energy efficiency.

[0122] In a possible implementation, the fitness function is SEI:

[0123]

[0124] F com A is the functional complementarity score, covA is the collaborative operation coverage area; tol is the total operating area; σ eg is the standard deviation of energy consumption within the group; E avg is the average energy consumption within the group; α, β, γ are weight coefficients; N is the number of agricultural machinery in the group;

[0125] At the beginning of the algorithm, the particle swarm randomly generates multiple matrices M to form the initial population; each matrix M represents a different possible scheme for agricultural machinery grouping and regional allocation. Calculate the fitness value. For each matrix M in the initial population, calculate F according to its specific grouping-region allocation information. com 、A cov , σ eg and E avg Parameters such as , and then substitute them into the fitness function SEI to obtain the fitness value corresponding to each matrix M; iterative update In each round of iteration, the particle swarm algorithm will update the position of the particle (generate a new matrix M) based on the current position (matrix M) and speed of the particle. At the same time, the fitness value of the newly generated matrix M is calculated again; during the update process, the position with the maximum fitness value experienced by each particle itself (individual optimal solution pbest) and the position with the maximum fitness value in the entire population (global optimal solution gbest) are recorded. Termination condition judgment When the preset termination condition is met (for example, the number of iterations reaches the upper limit, or the rate of change of the fitness value is less than a certain threshold), the algorithm stops iterating; at this time, the matrix M corresponding to the global optimal solution gbest is the optimal grouping solution found. Determine the final grouping based on the element M in the matrix M corresponding to the global optimal solution gbest ij The value of can determine the specific grouping situation. If M ij =1, then group G i Responsible for area R j , thus completing the grouping and regional allocation of agricultural machinery, if M ij =0, it means irresponsible.

[0126] In a possible implementation, the functional complementarity score F com Obtain it in the following ways:

[0127] Calculate the cosine similarity of the agricultural machinery function vectors within the group:

[0128]

[0129] where ∑ x1＜x2 cosθ x1x2 The sum of the cosine values of the angle between the function vectors of any two agricultural machines (x1, x2) in the group; N is the number of agricultural machines in the initial group;

[0130] In a possible implementation, the collaborative operation coverage area is obtained by:

[0131]

[0132] n is the total number of agricultural machines participating in the collaborative operation, m is the total number of overlapping areas generated by adjacent agricultural machine operation areas (that is, there are m overlapping areas); k is the index identifier of the individual agricultural machine, ranging from 1 to n, which is used to traverse each agricultural machine; h is the index identifier of the individual overlapping area, ranging from 1 to m, which is used to traverse each overlapping area.

[0133] The working principle of the above technical solution is: parameter set construction and cluster analysis:

[0134] First, the performance parameters (such as functional modules, operating speed, operating accuracy, width, and energy consumption) and coordination parameters (such as communication delay, positioning accuracy, and synchronization error) of each agricultural machinery are collected to construct a comprehensive set of agricultural machinery operation capability parameters.

[0135] Based on these parameter sets, a clustering algorithm is used to group agricultural machinery to form initial groups. The clustering algorithm will group together agricultural machinery with similar operating capabilities and collaborative characteristics based on the functional complementarity and collaborative efficiency between agricultural machinery.

[0136] For each initial grouping, its functional complementarity score is calculated; the functional complementarity score is measured by the cosine similarity of the agricultural machinery function vectors within the grouping. The higher the cosine similarity, the stronger the functional complementarity between the agricultural machinery.

[0137] Based on the energy consumption prediction model, the energy consumption of each initial group is evaluated. The energy consumption prediction model will predict the energy consumption of the group during the operation process based on the operating parameters and historical data of the agricultural machinery.

[0138] The groups whose energy consumption standard deviation is less than the preset threshold are screened to form the second candidate group. The smaller the energy consumption standard deviation, the more stable the energy consumption of the group, which is more conducive to improving the overall operation efficiency.

[0139] The farmland is divided into multiple initial operating sub-areas, and a spatial feature vector is constructed for each sub-area, including average slope, soil hardness, obstacle density, field connectivity, etc. The similarity between the initial sub-areas is calculated, and adjacent sub-areas with similarity greater than a preset threshold are merged to form a larger operating area. The merged areas have more similar spatial characteristics, which facilitates the coordinated operation of agricultural machinery.

[0140] A particle swarm algorithm (PSO) randomly matches the second candidate group with the merged operation areas, generating multiple possible group-area assignment matrices. The PSO iteratively optimizes these matrices to find the optimal group-area assignment. During this iteration, the algorithm evaluates each option's functional complementarity, coverage synergy, and energy efficiency based on a fitness function.

[0141] The fitness function comprehensively considers three indicators: functional complementarity score, collaborative operation coverage area and energy efficiency.

[0142] The particle swarm algorithm continuously iterates, updating the positions and velocities of particles to find the grouping-area allocation solution with the highest fitness value. When a preset termination condition is met (such as the number of iterations reaching an upper limit or the rate of change in fitness value falling below a threshold), the algorithm stops iterating and outputs the matrix corresponding to the global optimal solution. This matrix represents the optimal agricultural machinery grouping and work area allocation solution.

[0143] The functional complementarity score measures the functional complementarity between agricultural machinery through cosine similarity. The higher the cosine similarity, the greater the functional difference between the agricultural machinery, the stronger the complementarity, and the more conducive to improving the overall operating efficiency.

[0144] The collaborative operation coverage area is obtained by calculating the sum of the products of the operation width and operation length of the agricultural machinery involved in the collaborative operation, minus the sum of the overlapping areas of the adjacent agricultural machinery operation areas.

[0145] The effects of the above technical solution are as follows: quantifying the complementarity of agricultural machinery function vectors through cosine similarity (e.g., the angle between the function vectors of a seed drill and a fertilizer applicator is small), ensuring the coordination of agricultural machinery functions within the group and reducing duplication of work; screening groups with small energy consumption standard deviations based on energy consumption prediction models to avoid energy waste caused by mixing high-energy-consuming and low-energy-consuming agricultural machinery; using a particle swarm algorithm to dynamically match farmland sub-regions with agricultural machinery groups, reducing the length of cross-regional operation paths; balancing functional complementarity, coverage area, and energy consumption through the fitness function SEI, supporting online adjustment of grouping schemes. When the obstacle density in a certain area suddenly increases, the algorithm automatically deploys a highly flexible group to take over. The sub-region merging strategy reduces the dimension of the optimization variables and improves the convergence speed of the particle swarm algorithm. The coordinated coverage area formula ensures seamless connection of agricultural machinery operations, reducing missed tillage or excessive work; redundant group design (e.g., retaining 20% spare groups) and an energy consumption balancing mechanism ensure task continuity in the event of a single machine failure.

[0146] In a possible implementation, the method includes planning an operation path for each agricultural machinery group that includes time-space collision avoidance constraints and establishing a time sequence scheduling strategy that matches the task volume; including:

[0147] The operation area is divided into multiple path planning units according to the spatial characteristics of the farmland, and a topological map of the nodes accessible to agricultural machinery is constructed;

[0148] Based on the operation sequence constraints and task priorities of the agricultural machinery group, a time window constraint is injected into the path node sequence of each agricultural machinery group; for example, the time window is: [t start +f*Δt,t end -(JD-f)*Δt]; where f is the current node index, Δt is the unit operation time; JD is the total number of path nodes, representing the number of all nodes included in this task. For example, as the node index f increases, the starting time gradually moves forward, forming a dynamic time window, which constrains the agricultural machinery to complete each node task in sequence within the specified time, avoiding time conflicts or task delays.

[0149] Refer to the attached Figure 3-4 , selecting a path generation strategy according to a target tracking mode, generating an initial collision-free path, and performing speed adjustment or path replanning on overlapping path segments through a spatiotemporal conflict detection model; the target tracking mode includes a fastest target tracking mode or a neighboring target tracking mode;

[0150] Establish a time series scheduling model based on dynamic time warping, and calculate the operating time window and task segmentation strategy of each agricultural machinery group according to the agricultural machinery group's operating efficiency, remaining task volume and energy endurance threshold;

[0151] Scheduling feasibility is verified through discrete event simulation. When path conflicts or task delays are detected, local path re-optimization and task reallocation mechanisms are triggered. Dynamic matching of tasks and resources can be achieved based on dynamic time warping (DTW), and the range of the time window required to complete the remaining tasks can be calculated.

[0152] The range of the time window is calculated as follows:

[0153]

[0154] Among them, TW is the time window length, Tb is the energy endurance reserved time (such as 30 minutes), Q r is the remaining task volume (hectares); V is the operating speed; De is the operating depth; η is the equipment redundancy coefficient, and Z is the conversion coefficient, which is used for the comprehensive conversion of parameters such as operating depth and area (related to the operation type).

[0155] During the operation, the path tracking control parameters are dynamically adjusted based on the real-time posture data of the agricultural machinery and the deviation value of the topology map, and the task status mark in the timing scheduling queue is updated.

[0156] The working principle of the above technical solution is:

[0157] Based on farmland DEM data and obstacle distribution, the operation area is divided into grid units of preset area (for example, 50m×50m), and an undirected graph of reachable nodes is constructed. The nodes are the grid centers, and the edges indicate that agricultural machinery can pass (no obstacles and the slope is less than 20°). According to the operation sequence (such as sowing→fertilizing→plant protection) and task priority, a time window is assigned to each path node sequence (for example, the sowing group needs to complete area A between 08:00 and 12:00).

[0158] Select the path generation strategy based on the target tracking mode:

[0159] Fastest target tracking mode: Adopting the time-optimal control model, minimizing the time to reach the target area is the optimization goal, and constructing the Hamiltonian function to solve the optimal acceleration curve;

[0160] Proximal target tracking mode: A virtual spring-damper model is established to maintain relative distance constraints with adjacent agricultural machinery and operating boundaries through potential field functions;

[0161] Based on spatial parameters such as the length, width, turning radius, and distance between agricultural machinery D1 / D2 of the working area, a path planning unit is constructed; combined with the working width and speed of agricultural machinery, the spatiotemporal coverage of the working path is constrained in the timing scheduling strategy to ensure full coverage of the working area and no path overlap; path planning constraints include obstacle avoidance, collaboration between agricultural machinery to avoid collisions, and division of the working area; common path planning algorithms include A* algorithm, Dijkstra algorithm, ant colony algorithm, etc.; for example, the fastest mode uses the A* algorithm to plan the shortest path, which is suitable for flat areas; the neighboring mode generates a local optimal path based on the RRT* algorithm, which is suitable for areas with dense obstacles.

[0162] The Space-Time Prism method can be used to predict the trajectories of agricultural machinery. If two trajectories overlap within a time window, conflict resolution is triggered. For example, the speed of the conflicting agricultural machinery is reduced to 50% of its original speed or the conflicting area is detoured using the RRT* algorithm. If the overlapping area exceeds 10% of the width, the operation sequence of the agricultural machinery is adjusted or the path is shifted.

[0163] Dynamic matching of tasks and resources can be achieved based on dynamic time warping (DTW). The time window starts from the entire task, calculates the time range required to complete the remaining tasks, and plans the total time frame for completing the task based on the remaining task volume, operating parameters, efficiency, and energy endurance reserve, focusing on the overall time rationality of task execution. For example, when energy endurance is less than 30%, the task is segmented into 5 hectares per segment to ensure that the agricultural machinery can return to the charging station.

[0164] Based on the real-time pose deviation (e.g. path tracking error > 5cm), the steering angle and speed can be adjusted by a PID controller.

[0165] AnyLogic software can be used to build discrete event models of agricultural machinery, obstacles, and time windows:

[0166] Model input parameters: number of agricultural machines, operating speed, and path node sequence;

[0167] Conflict detection logic: For example, the positions of agricultural machinery are monitored within a preset time interval (e.g., every 5 seconds). If the distance between two agricultural machinery is less than a safe distance (e.g., 3 meters) and the time windows overlap, a conflict is marked.

[0168] Reassignment strategy: When a conflict is triggered, the conflict area tasks are assigned to the backup group, and the original group is transferred to other areas.

[0169] Each time a path node is completed, the task status is updated to "Completed" and the time window for subsequent nodes is released.

[0170] The effects of the above technical solution are:

[0171] Based on farmland DEM data and obstacle distribution, grid cells are divided and a topological map is constructed to clearly define the range of agricultural machinery passage. This provides a precise spatial model for path planning, reduces ineffective planning, and improves path planning efficiency and accuracy. Time conflicts are effectively avoided by injecting time window constraints. By dynamically adjusting node time ranges, agricultural machinery is constrained to complete tasks according to the order and priority of operations, avoiding time conflicts among multiple units, reducing the risk of operation delays, and improving time coordination efficiency. Enhanced adaptability to complex scenarios combines the fastest target tracking mode (the A algorithm plans the shortest path, suitable for flat areas) with the neighboring target tracking mode (the RRT algorithm generates the local optimal path, suitable for areas with dense obstacles) to adapt to different farmland environments. This ensures that path planning is both efficient and responsive to actual terrain, enhancing the flexibility of agricultural machinery operations. To ensure safe and reliable operations, the space-time prism method is used to predict agricultural machinery trajectories. Conflict resolution (such as speed reduction, detours, and adjustment of operation sequence) is triggered for overlapping trajectories within the time window, reducing the risk of collisions. For example, if the overlap exceeds 10% of the width, the path is adjusted to minimize equipment damage and task interruptions, ensuring operational safety. Dynamic optimization matching of tasks and resources is based on dynamic time warping (DTW) calculation time window, combined with the remaining task volume and energy endurance segmented tasks (such as 5 hectares / segment when energy <30%), to achieve dynamic matching of task volume and agricultural machinery resources (operation efficiency, energy), avoid energy waste or task interruption, improve resource utilization, and improve real-time path tracking accuracy based on real-time posture deviation (such as path tracking error > 5cm). The steering angle and speed are dynamically adjusted through the PID controller to ensure that agricultural machinery accurately tracks the planned path, improve operation accuracy (such as sowing and fertilization uniformity), and ensure the quality of farmland operations. Scheduling feasibility verification and optimization: Discrete event simulation (e.g., AnyLogic model) inputs parameters such as the number of agricultural machines and operating speed. Real-time monitoring of the distance between agricultural machines (<3m) and the overlap of time windows triggers task reallocation (conflicting areas are assigned to backup groups). This allows for early detection and optimization of scheduling issues, ensuring a smooth workflow and improving task completion rates. Task status is managed in real time. Each time a path node is completed, the task status is updated to "Completed" and the time window for subsequent nodes is released. Real-time feedback on the progress of the task is provided, providing accurate status information for subsequent scheduling, enhancing the real-time responsiveness of the scheduling system, and optimizing overall workflow management.

[0172] In one possible implementation, the dynamic reconstruction of the agricultural machinery grouping topology and operation parameters based on the real-time collected farmland environment data, agricultural machinery operating condition data, and operation quality data includes:

[0173] The soil sensors, visual sensors, and positioning modules on agricultural machinery collect farmland environmental data in real time, including soil moisture gradient, light intensity distribution, dynamic obstacle locations, and crop growth status;

[0174] Obtain operating condition data through the agricultural machinery controller bus, including engine load rate, hydraulic system pressure, remaining energy endurance and fault codes;

[0175] The sensors of the operation actuators collect operation quality data, including standard deviation of sowing depth, fertilization uniformity and pesticide coverage;

[0176] Design an anomaly detection mechanism based on a sliding window to perform time series alignment and feature extraction on the environmental data, working condition data, and quality data. Use preset thresholds and machine learning models to identify agricultural machinery performance degradation, sudden obstacle intrusion, and operational quality deviation events, and generate abnormal event labels and impact range assessment results.

[0177] Trigger the dynamic reconstruction decision-making process and select the group topology adjustment mode based on the type and severity of the abnormal event.

[0178] In a possible implementation, selecting a marshaling topology adjustment mode based on the abnormal event type and severity level includes:

[0179] When a single machine failure is detected, a topology reconstruction algorithm based on task migration is used to decompose the faulty agricultural machinery task into subtask sets, which are then redistributed to adjacent groups based on the remaining agricultural machinery load capacity and spatial accessibility, and the spatiotemporal constraints of the collaborative paths between groups are updated.

[0180] When a sudden change in the environment is detected, an adaptive sub-region redivision strategy is adopted to dynamically adjust the operating boundaries of the agricultural machinery group based on obstacle contours and changes in soil moisture conditions. A reinforcement learning model is then used to generate an optimal path compensation plan to bypass the sudden change area.

[0181] When quality deviation is detected, a multi-objective optimization engine is used to adjust the operating speed, operating width, and actuator parameters of the agricultural machinery group online, using the operation quality indicators as constraints. Parameter synchronization is achieved through closed-loop control between agricultural machinery within the group.

[0182] Deploy a dynamic parameter collaborative controller to generate a control sequence based on the reconstructed group topology, including agricultural machinery posture compensation, operating parameter correction, and communication link switching instructions. A distributed consensus algorithm is used to ensure the consistency of agricultural machinery control instructions within the group. The reconstruction scheme is pre-verified through the digital twin system and then issued to the actual agricultural machinery for execution.

[0183] Build a hierarchical fusion decision-making system for multiple concurrent abnormal scenarios;

[0184] A reconstruction effect feedback loop is established to continuously monitor the marshalling operation stability indicators and task completion progress within a preset time window. If a secondary anomaly or progress lag is detected, incremental local reconstruction is triggered; otherwise, the current marshalling topology and parameters are updated to the baseline configuration.

[0185] For multiple concurrent exception scenarios, a hierarchical fusion decision-making system is built, including:

[0186] A fuzzy membership function is established based on the three dimensions of safety (level of threat from sudden environmental changes), task impact (assessment of progress delays caused by faults / quality issues), and repair complexity (difficulty of troubleshooting, cost of quality correction). For example, for the safety of obstacle intrusion, the membership is divided into "high threat", "medium threat", and "low threat" according to the distance from the agricultural machinery, and the priority of event handling is determined through weighted calculation.

[0187] Use historical data to train causal analysis models and explore potential connections between events; for example, discover the causal chain of "obstacle approaching agricultural machinery → agricultural machinery turning sharply to avoid obstacle → sudden change in hydraulic system pressure → component failure", monitor the risk of other events induced by environmental mutations in real time, and adjust operating parameters in advance (such as reducing obstacle avoidance speed) to prevent derivative failures.

[0188] Define resource pool constraints, including: remaining operational capacity of agricultural machinery (functional module availability, energy endurance), load threshold (no more than 80% of rated load), and movement cost (based on energy consumption predictions for new paths). Build an optimization model: With the goal of minimizing task completion time and balancing resource consumption, a hybrid algorithm of "reinforcement learning + integer programming" is used to dynamically assign tasks from faulty agricultural machinery to remaining machinery, while also taking into account the impact of path changes after obstacle avoidance.

[0189] Set up a hierarchical processing mechanism:

[0190] Emergency processing layer: Prioritizes responding to sudden environmental changes, generates obstacle avoidance paths through reinforcement learning, verifies path safety through digital twin rehearsals, and simultaneously updates grouping space constraints.

[0191] Task maintenance layer: Within the new operation area, based on resource pool scheduling results, a task migration algorithm is used to split the tasks of the faulty agricultural machinery. For example, a seeding task is split into "ditching + reseeding" subtasks and assigned to different agricultural machinery to ensure uninterrupted operation.

[0192] Quality Optimization Layer: To address quality deviations, a "operation parameter-quality indicator" mapping model is constructed using new topologies and paths as constraints. For example, if obstacle avoidance causes a change in operating speed, actuator parameters are dynamically adjusted through multi-objective optimization (taking into account speed, seeding depth, and fertilization uniformity), and closed-loop control within the group is used to achieve precise parameter synchronization.

[0193] A knowledge graph is constructed that encompasses agricultural machinery performance, historical handling cases, and environmental response strategies. When a complex combination of events occurs, the graph is used to retrieve similar scenarios (e.g., historical cases of "a certain type of obstacle + a certain component failure + quality deviation"), and initial handling solutions are generated through analogical reasoning. Simultaneously, new handling experiences are fed back into the graph to update the "event-strategy" knowledge association, enhancing the system's intelligent decision-making capabilities for complex scenarios.

[0194] The working principle and effectiveness of this technical solution are as follows: Using soil sensors, vision sensors, and positioning modules on agricultural machinery, real-time data is collected on soil moisture gradients, light intensity distribution, dynamic obstacle locations, and crop growth status. Operating condition data such as engine load factor, hydraulic system pressure, remaining energy endurance, and fault codes are acquired via the agricultural machinery controller bus. Sensors on the actuators collect operational quality data such as seeding depth standard deviation, fertilization uniformity, and pesticide coverage.

[0195] The collected data is time-series aligned to ensure temporal consistency of data from different sources, and features useful for anomaly detection and decision-making are extracted from the raw data.

[0196] A sliding window-based anomaly detection mechanism is designed to monitor preprocessed data in real time. Using preset thresholds and machine learning models (such as classification and clustering algorithms), it identifies abnormal events such as machinery performance degradation, sudden obstacle intrusion, and deviations in operational quality. Once an abnormal event is detected, an abnormal event label and impact assessment are immediately generated. Based on the type and severity of the abnormal event, a corresponding dynamic reconstruction decision process is triggered.

[0197] Marshalling topology adjustment mode selection:

[0198] In the event of a single machine failure, a topology reconstruction algorithm based on task migration is used to decompose the tasks of the failed machine into subtasks. These subtasks are then redistributed to adjacent groups based on the load capacity and spatial accessibility of the remaining machines. Simultaneously, the spatiotemporal constraints of the collaborative paths between groups are updated.

[0199] Environmental Sudden Changes: Adaptive sub-area redivision strategies are used to dynamically adjust the operating boundaries of agricultural machinery groups based on obstacle contours and soil moisture changes. A reinforcement learning model is also used to generate optimal path compensation solutions to bypass sudden change areas.

[0200] Quality Deviation: Using a multi-objective optimization engine, we use operational quality indicators as constraints to adjust the operating speed, operating width, and actuator parameters of the agricultural machinery group online. We also achieve precise synchronization of these parameters through closed-loop control between agricultural machinery within the group.

[0201] A dynamic parameter collaborative controller is deployed to generate a control sequence including agricultural machinery posture compensation, operating parameter correction and communication link switching instructions based on the reconstructed grouping topology.

[0202] A distributed consensus algorithm is used to ensure the consistency of agricultural machinery control instructions within the group.

[0203] The reconstruction plan is pre-verified through the digital twin system to ensure the feasibility and safety of the plan, and then the control sequence is sent to the actual agricultural machinery for execution.

[0204] Hierarchical fusion decision-making system:

[0205] Fuzzy membership functions are established to evaluate and prioritize abnormal events based on safety, mission impact, and repair complexity. Causal analysis models are trained using historical data to uncover potential connections between events and predict potential derivative failures or quality deviations.

[0206] Resource pool constraint and optimization model:

[0207] Define resource pool constraints, including remaining operating capacity of agricultural machinery, load threshold, movement cost, etc.

[0208] An optimization model is constructed with the goal of minimizing task completion time and balancing resource consumption. A hybrid algorithm combining reinforcement learning and integer programming is used to dynamically allocate tasks from faulty agricultural machinery to the remaining agricultural machinery.

[0209] Layered processing mechanism:

[0210] Emergency processing layer: Prioritizes responding to emergency events such as sudden environmental changes, generates obstacle avoidance paths through reinforcement learning, and verifies the safety of the paths through digital twin system rehearsals.

[0211] Task maintenance layer: In the new operation area, based on the resource pool scheduling results, the task migration algorithm is used to split the tasks of the faulty agricultural machinery to ensure the continuity of the operation process.

[0212] Quality optimization layer: To address the problem of operation quality deviation, a mapping model between operation parameters and quality indicators is constructed with new topology and path as constraints, and the actuator parameters are dynamically adjusted through a multi-objective optimization algorithm.

[0213] Knowledge graph and intelligent decision-making:

[0214] Build a knowledge graph that includes agricultural machinery performance, historical processing cases, and environmental response strategies.

[0215] When a complex combination of events occurs, solutions for similar scenarios are retrieved through the knowledge graph, and analogical reasoning is performed to generate an initial solution.

[0216] Feedback new processing experience into the knowledge graph, update the relationship between events and strategies, and enhance the system's intelligent decision-making capabilities for complex scenarios.

[0217] To summarize, the working principle of the autonomous driving agricultural machinery grouping operation control method is an intelligent process that integrates real-time data collection, anomaly detection and decision-making, dynamic reconstruction, and collaborative control. Its core lies in the use of multi-source data fusion, machine learning models, and optimization algorithms to achieve flexible adjustment of agricultural machinery groupings and dynamic optimization of operation parameters, thereby improving agricultural production efficiency and quality and reducing labor costs and resource consumption.

[0218] In Example 2, the present application provides an automatic driving agricultural machinery grouping operation control system for implementing any of the methods described in the present application, the system comprising:

[0219] The agricultural machinery configuration module is used to determine the functional type and quantity of the required agricultural machinery according to the farming task type, farmland size and terrain parameters;

[0220] The grouping module is used to generate agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics;

[0221] The planning and scheduling module is used to plan an operation path for each agricultural machinery group, including time and space collision avoidance constraints, and establish a time sequence scheduling strategy that matches the task volume;

[0222] The reconstruction module is used to dynamically reconstruct the agricultural machinery grouping topology and operation parameters based on the real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data.

[0223] In a possible implementation, the system further includes:

[0224] A communication construction module is used to construct a two-layer communication architecture consisting of an execution layer and a decision-making layer. The execution layer establishes a direct communication link between agricultural machinery and the vehicle controller via data transmission equipment, and the decision-making layer establishes a cloud relay link via wireless communication with the task machine. The agricultural machinery posture data is transmitted in real time via the execution layer direct link, and the operation status data is transmitted via the decision-making layer relay link. The ground station interacts with the execution layer and the decision-making layer via the data transmission equipment, realizing real-time transmission of agricultural machinery data under the two-layer communication architecture.

[0225] Feedback adjustment module, used to collect operation quality indicators through multimodal sensors on agricultural machinery and conduct feedback adjustment based on preset standards;

[0226] The report generation module is used to generate a detailed operation report after the operation is completed, including data such as the operation area, operation quality, energy consumption, etc., to provide a basis for subsequent operation optimization.

[0227] The working principle and effect of the above technical solution are the same as those in the embodiment of the method of this application, and will not be repeated here.

[0228] An embodiment of the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the embodiments of the present application when executing the computer program.

[0229] An embodiment of the present application also provides a computer-readable storage medium, which is used to store a computer program. When the computer program is executed, the steps of any method described in the embodiments of the present application are implemented. The specific implementation method is consistent with the implementation method and the technical effect achieved in the above-mentioned method embodiments, and some contents will not be repeated here.

[0230] In the present application, a readable storage medium can be any tangible medium that contains or stores a program that can be used by or in combination with an instruction execution system, device or device. A program product can use any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0231] A computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, or any suitable combination thereof. The program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java and C++, as well as conventional procedural programming languages such as C or similar programming languages. The program code may be executed entirely on the user computing device, partially on an associated device, as a standalone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. Where a remote computing device is involved, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0232] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limiting the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the invention without departing from the principles and purpose of the present invention. All such changes shall fall within the scope of protection of the claims of the present invention.

Claims

1. A method for controlling automatic driving agricultural machinery grouping operations, characterized in that: The method comprises: Determine the functional type and quantity of agricultural machinery required based on the type of farming task, farmland size and terrain parameters; Generate agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics; Plan an operation path for each agricultural machinery group that includes time and space collision avoidance constraints, and establish a time scheduling strategy that matches the task volume; Based on real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data, the agricultural machinery grouping topology and operation parameters are dynamically reconstructed.

2. The automatic driving agricultural machinery grouping operation control method according to claim 1 is characterized in that: The method further comprises: A two-layer communication architecture consisting of an execution layer and a decision-making layer is constructed. The execution layer establishes a direct communication link between agricultural machinery and the vehicle controller through data transmission equipment, and the decision-making layer establishes a cloud relay link through wireless communication with the task machine. Agricultural machinery posture data is transmitted in real time through the execution layer direct link, and operation status data is transmitted through the decision-making layer relay link. The ground station interacts with the execution layer and the decision-making layer through data transmission equipment, realizing real-time transmission of agricultural machinery data under the two-layer communication architecture. The multimodal sensors installed on agricultural machinery collect operation quality indicators and make feedback adjustments based on preset standards.

3. The automatic driving agricultural machinery grouping operation control method according to claim 1 is characterized in that: The functional type and quantity configuration of the required agricultural machinery are determined based on the farming task type, farmland size and terrain parameters, including: Analyze the characteristics of farming tasks to obtain operation parameters; the operation parameters include operation depth, row spacing requirements, and pesticide dosage; Generate constraints on operations based on farmland parameters, including operating area, average slope, soil compaction, field flatness, and obstacle density; Modifying the operation parameters according to the constraint conditions; Based on the revised operating parameters, the performance parameter range of agricultural machinery that meets the operating requirements is screened out; The number of agricultural machinery required is calculated based on the farmland area, working depth, working speed, daily working time and equipment redundancy factor.

4. The automatic driving agricultural machinery grouping operation control method according to claim 1, characterized in that: The method of generating agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics includes: Based on cluster screening of agricultural machinery performance parameters and synergy parameters, candidate groups with complementary functions and stable energy consumption are formed; The particle swarm algorithm is used to optimize the matching relationship between groups and farmland sub-areas, and the optimal agricultural machinery grouping and operation area allocation plan is determined by comprehensively considering functional complementarity, coverage synergy and energy efficiency indicators.

5. The automatic driving agricultural machinery grouping operation control method according to claim 4 is characterized in that: The clustering and screening based on the agricultural machinery performance parameters and the synergy parameters to form candidate groups with complementary functions and stable energy consumption includes: Establishing a parameter set for the operation capability of each agricultural machine through the performance parameters and coordination parameters of each agricultural machine; the performance parameters include functional modules, operation speed, operation accuracy, width and energy consumption; the coordination parameters include communication delay, positioning accuracy and synchronization error; Clustering the agricultural machinery according to the operational capability parameter sets of each agricultural machinery to obtain initial groups, and calculating the functional complementarity scores in each initial group; determining the first candidate group based on the complementarity scores and the types of operations covered; Based on the energy consumption prediction model, the energy consumption of each agricultural machine in the first candidate group is obtained, and the groups whose energy consumption standard deviation is less than a preset first threshold are screened to form a second candidate group.

6. The automatic driving agricultural machinery grouping operation control method according to claim 5, characterized in that: The particle swarm algorithm is used to optimize the matching relationship between the groups and the farmland sub-areas, and the optimal agricultural machinery grouping and operation area allocation scheme is determined by comprehensively considering functional complementarity, coverage synergy, and energy efficiency indicators, including: The farmland is divided into initial operation sub-areas, the spatial feature vector of the farmland is constructed, and the similarity between the initial sub-areas is calculated; the spatial features include average slope, soil hardness, obstacle density, and field connectivity; Merge adjacent sub-regions whose similarity is greater than a preset threshold according to the similarity between the initial sub-regions and calculate the collaborative coverage area of the merged region; The particle swarm algorithm is used to iteratively optimize the random matching matrix of the second candidate grouping and the merged area. The functional complementarity, coverage synergy and energy efficiency of the grouping scheme are evaluated through the fitness function. Finally, the agricultural machinery grouping and operation area allocation are determined by the matrix corresponding to the global optimal solution.

7. The automatic driving agricultural machinery grouping operation control method according to claim 1, characterized in that: The method includes planning an operation path for each agricultural machinery group including time and space collision avoidance constraints and establishing a time sequence scheduling strategy that matches the task volume; including: The operation area is divided into multiple path planning units according to the spatial characteristics of the farmland, and a topological map of the nodes accessible to agricultural machinery is constructed; Based on the operation sequence constraints and task priorities of the agricultural machinery group, a time window constraint is injected into the path node sequence of each agricultural machinery group; Selecting a path generation strategy based on a target tracking mode to generate an initial collision-free path and performing speed adjustment or path replanning on overlapping path segments using a spatiotemporal conflict detection model; the target tracking mode includes a fastest target tracking mode or a neighboring target tracking mode; Establish a time series scheduling model based on dynamic time warping, and calculate the operating time window and task segmentation strategy of each agricultural machinery group according to the agricultural machinery group's operating efficiency, remaining task volume and energy endurance threshold; The feasibility of scheduling is verified through discrete event simulation, and local path re-optimization and task reallocation mechanisms are triggered when path conflicts or task delays are detected; During the operation, the path tracking control parameters are dynamically adjusted based on the real-time posture data of the agricultural machinery and the deviation value of the topology map, and the task status mark in the timing scheduling queue is updated.

8. The automatic driving agricultural machinery grouping operation control method according to claim 1, characterized in that: The method of dynamically reconstructing the agricultural machinery grouping topology and operation parameters based on the real-time collected farmland environment data, agricultural machinery operating condition data, and operation quality data includes: The soil sensors, visual sensors, and positioning modules on agricultural machinery collect farmland environmental data in real time, including soil moisture gradient, light intensity distribution, dynamic obstacle locations, and crop growth status; Obtain operating condition data through the agricultural machinery controller bus, including engine load rate, hydraulic system pressure, remaining energy endurance and fault codes; The sensors of the operation actuators collect operation quality data, including standard deviation of sowing depth, fertilization uniformity and pesticide coverage; Design an anomaly detection mechanism based on a sliding window to perform time series alignment and feature extraction on the environmental data, working condition data, and quality data. Use preset thresholds and machine learning models to identify agricultural machinery performance degradation, sudden obstacle intrusion, and operational quality deviation events, and generate abnormal event labels and impact range assessment results. Trigger the dynamic reconstruction decision-making process and select the group topology adjustment mode based on the type and severity of the abnormal event.

9. The automatic driving agricultural machinery grouping operation control method according to claim 8, characterized in that: The selection of a marshaling topology adjustment mode based on the type and severity of the abnormal event includes: When a single machine failure is detected, a topology reconstruction algorithm based on task migration is used to decompose the faulty agricultural machinery task into subtask sets, which are then redistributed to adjacent groups based on the remaining agricultural machinery load capacity and spatial accessibility, and the spatiotemporal constraints of the collaborative paths between groups are updated. When a sudden change in the environment is detected, an adaptive sub-region redivision strategy is adopted to dynamically adjust the operating boundaries of the agricultural machinery group based on obstacle contours and changes in soil moisture conditions. A reinforcement learning model is then used to generate an optimal path compensation plan to bypass the sudden change area. When quality deviation is detected, a multi-objective optimization engine is used to adjust the operating speed, operating width, and actuator parameters of the agricultural machinery group online, using the operation quality indicators as constraints. Parameter synchronization is achieved through closed-loop control between agricultural machinery within the group. Deploy a dynamic parameter collaborative controller to generate a control sequence based on the reconstructed group topology, including agricultural machinery posture compensation, operating parameter correction, and communication link switching instructions. A distributed consensus algorithm is used to ensure the consistency of agricultural machinery control instructions within the group. The reconstruction scheme is pre-verified through the digital twin system and then issued to the actual agricultural machinery for execution. A reconstruction effect feedback loop is established to continuously monitor the marshalling operation stability indicators and task completion progress within a preset time window. If a secondary anomaly or progress lag is detected, incremental local reconstruction is triggered; otherwise, the current marshalling topology and parameters are updated to the baseline configuration.

10. An automatic driving agricultural machinery grouping operation control system, used to implement the method described in any one of claims 1 to 9, characterized in that: The system comprises: The agricultural machinery configuration module is used to determine the functional type and quantity of the required agricultural machinery according to the farming task type, farmland size and terrain parameters; The grouping module is used to generate agricultural machinery collaborative groups based on the agricultural machinery operation capability parameter set and farmland spatial characteristics; The planning and scheduling module is used to plan an operation path for each agricultural machinery group, including time and space collision avoidance constraints, and establish a time sequence scheduling strategy that matches the task volume; The reconstruction module is used to dynamically reconstruct the agricultural machinery grouping topology and operation parameters based on the real-time collected farmland environment data, agricultural machinery operating condition data and operation quality data.

Images