Crop multi-link operation progress synchronous monitoring method and system
By constructing a multi-dimensional input system and a real-time monitoring mechanism, the problem of time-series dependence among multiple stages of crop operations has been solved, enabling accurate prediction and dynamic adjustment of the time links of multiple stages of operations, and improving the coordination and timeliness of agricultural machinery operations.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-08
- Publication Date
- 2026-04-07
AI Technical Summary
Existing systems fail to effectively handle the temporal dependencies between multiple stages of crop operations, leading to problems such as broken operation chains, idle resources, or overlapping stages, and are unable to achieve synchronous monitoring and scheduling of multiple stages of operations.
A multi-dimensional input system is constructed, which includes the type of operation link, the characteristic parameters of the target plot, and the characteristic parameters of the equipment. An operation task completion time prediction model is adopted, and the start time of the next link is generated by the connection type. Real-time monitoring and residual feedback mechanisms are introduced to realize the synchronous monitoring and dynamic adjustment of the operation time link of multiple links.
It enables precise time prediction and dynamic adjustment of multi-stage operations, avoiding problems such as time overlap, excessively long intervals, or misordered stages, improving the coordination and timeliness of agricultural machinery operation chains, and ensuring that the operation progress is consistent with the actual progress on site.
Smart Images
Figure CN121809816A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of multi-stage operation progress monitoring technology, specifically to a method and system for synchronous monitoring of multi-stage operation progress in crops. Background Technology
[0002] Agricultural production operations typically involve multiple consecutive steps, such as sowing, fertilizing, spraying pesticides, irrigating, and harvesting, with strict temporal dependencies between each step. For example, covering with soil can only be done after sowing is completely finished, and harvesting must begin after dehydration and drying are completed. The operation cycle, speed, and completion time of these steps are all affected by factors such as plot size, type of machinery, environmental conditions, and operator factors, exhibiting significant dynamic uncertainty.
[0003] With the continuous improvement of mechanization and informatization in modern agriculture, complex operational chains consisting of multiple operational stages have gradually formed in the planting, management, and harvesting of crops, such as tilling, fertilizing, sowing, covering, spraying, irrigation, and harvesting. There are usually strict temporal dependencies between these stages. Delays, advances, or operational anomalies in one stage will significantly impact the execution of subsequent stages, thus affecting overall operational efficiency and agricultural time utilization. Existing systems often use a single operational stage as the control unit and do not model the connection types between multiple stages (such as sequential, parallel, or no connection). This leads to an inability to correctly handle the dependencies between different stages during task scheduling. Once an upstream stage is delayed, the system cannot automatically adjust the start time of subsequent stages, resulting in problems such as operational chain breaks, resource idleness, or stage overlap. Summary of the Invention
[0004] To address the aforementioned problems, this invention provides a method and system for synchronously monitoring the progress of multi-stage crop operations.
[0005] This invention adopts the following technical solution: a method for synchronous monitoring of the progress of multi-stage operations in crop farming, comprising:
[0006] Step S01: Obtain the operation characteristic parameters of the current operation stage, including the operation stage type, the characteristic parameters of the target plot, and the characteristic parameters of the equipment;
[0007] Step S02: Obtain the start time of the current task stage, input the obtained start time of the current task stage and task feature parameters into the task completion time prediction model, and predict the task completion time of the current task stage;
[0008] Step S03: Obtain the connection type between the next work step and the current work step;
[0009] Step S04: Based on the connection type and the task completion time of the current task, generate the start time of the next task, obtain the task feature parameters of the next task, and predict the end time of the next task based on the start time and task feature parameters of the next task.
[0010] Step S05: Derive the time links of multiple operations sequentially through a recursive method to form a time sequence of multiple operations and a progress synchronization table for multiple operations.
[0011] Step S06: After the current task is completed, obtain the actual completion time of the current task, compare and analyze the actual completion time with the predicted completion time of the current task, and determine whether to generate a multi-stage task progress synchronization table update instruction.
[0012] As a further description of the above technical solution: the types of operation steps include turning the soil, fertilizing, sowing, covering with soil, spraying pesticides, and irrigating;
[0013] The characteristic parameters of the target land parcel include land parcel area, land parcel shape complexity, terrain slope, and surface undulation.
[0014] The machine's characteristic parameters include its working width and its rated working speed.
[0015] As a further description of the above technical solution: the method for obtaining the surface relief includes:
[0016] Multiple elevation data points on the target site were obtained through drone-based lidar scanning.
[0017] The surface relief is obtained by calculating the variance of multiple elevation data points within the plot.
[0018] The complexity of the land parcel shape is calculated by taking the perimeter of the land parcel from the set of coordinate points on the parcel boundary, and is obtained based on the ratio of the perimeter of circles with equal areas.
[0019] As a further description of the above technical solution: the connection types include no connection type, sequential connection type, parallel connection type, and condition-triggered connection type;
[0020] Among them, the no-connection type means that there is no time dependency between the next operation and the current operation, that is, the start time of the next operation is independent of the completion time of the current operation.
[0021] The sequential connection type indicates that the next operation can only be started after all the current operations are completed;
[0022] Parallel connection type means that two work steps can be partially overlapped in terms of spatial area or time period, but work interference constraints must be met;
[0023] Condition-triggered connection type, that is, the start of the next link does not depend on the completion time of the current link, but depends on the current link reaching a certain state threshold or characteristic condition.
[0024] As a further description of the above technical solution: the training method of the task completion time prediction model includes:
[0025] Obtain the dataset of each operation stage of the historical target land parcel, the dataset including the start time of the operation stage, operation characteristic parameters, and task completion time of the operation stage;
[0026] Discrete features in the task feature parameters are encoded using one-hot encoding, and continuous features are standardized using Z-score to eliminate dimensional differences. The start time of the task steps and the task feature parameters are used as the input feature vector, and the task completion time of the task steps is used as the output of the model.
[0027] Based on a preset ratio, the dataset is divided into training, validation, and test sets. Gradient boosting regression trees are used as the prediction model. The model is trained using the training set, and mean squared error is used as the loss function to quantify the deviation between the predicted and actual time consumption. Gradient descent is used to update the parameters of each tree based on the negative gradient of the loss function. The mean squared error of the validation set is calculated every 20 trees. Training is stopped when the mean squared error decreases by less than 0.01 for three consecutive times. At the same time, Bayesian optimization is used to tune the hyperparameters.
[0028] The trained model is evaluated using a validation set, and core evaluation metrics are calculated. A model with an overall accuracy of ≥95% is considered to have met the performance evaluation criteria. Once the model has met the performance evaluation criteria, it can be used.
[0029] As a further description of the above technical solution: the method for generating the start time of the next work step based on the connection type and the task completion time of the current work step includes:
[0030] A time dependency graph of task steps is constructed using historical task records. The time dependency graph of task steps consists of nodes. and edge composition, Indicates the first Each step of the operation Indicates work process and The connection relationships are defined, with each edge accompanied by a connection type label. This is used to distinguish between non-connected types, sequential connection types, parallel connection types, and conditionally triggered connection types.
[0031] Based on the time dependency graph of the task steps, the connection type between the current task step and the next task step is obtained. ;
[0032] Obtain the predicted completion time of the current task and the connection type between the current task and the next task. Determine the start time for the next work step.
[0033] As a further description of the above technical solution: the step of obtaining the predicted completion time of the current task and the connection type between the current task and the next task. Methods for determining the start time of the next work step include:
[0034] When the connection type is no connection type, the start time of the next operation is the predetermined start time of that operation in the original operation plan;
[0035] When the connection type is sequential connection type, the start time of the next operation is the end time of the current operation plus the buffer time for the preparation of the next operation.
[0036] When the connection type is parallel connection type, the start time of the next operation is the end time of the current operation minus the allowed operation time overlap time;
[0037] When the connection type is a condition-triggered connection type, the start time of the next operation is the time when the triggering condition is met.
[0038] As a further description of the above technical solution: the method for comparing and analyzing the actual completion time with the predicted completion time of the current task stage to determine whether to generate a multi-stage task progress synchronization table update instruction includes:
[0039] Obtain the residual value between the actual completion time and the predicted completion time of the current task stage. Compare and analyze the residual value with a preset residual threshold. If the residual value is greater than or equal to the preset residual threshold, a multi-stage task progress synchronization table update instruction is generated. If the residual value is less than the preset residual threshold, no multi-stage task progress synchronization table update instruction is generated.
[0040] As a further description of the above technical solution: when the multi-stage operation progress synchronization table update instruction is generated, the actual completion time of the current operation is obtained, the predicted completion time of the current operation stage is updated to the actual completion time, and the multi-stage operation progress synchronization table is updated.
[0041] A system for synchronously monitoring the progress of multi-stage crop operations, used to implement the aforementioned method for synchronously monitoring the progress of multi-stage crop operations, the system comprising:
[0042] The parameter acquisition module acquires the operational characteristic parameters of the current operation stage, including the operation stage type, the target plot's characteristic parameters, and the equipment's characteristic parameters.
[0043] The task time prediction module obtains the start time of the current task stage, inputs the obtained start time of the current task stage and task feature parameters into the task completion time prediction model, and predicts the completion time of the current task stage.
[0044] The connection type acquisition module retrieves the connection type between the next operation stage and the current operation stage.
[0045] The end time acquisition module generates the start time of the next task based on the connection type and the task completion time of the current task, acquires the task feature parameters of the next task, and predicts the end time of the next task based on the start time and task feature parameters of the next task.
[0046] The task progress synchronization table module derives the time chain of multiple tasks in a recursive manner, forming a time sequence of multiple tasks and a task progress synchronization table.
[0047] The update module, after the current task stage is completed, obtains the actual completion time of the current task, compares and analyzes the actual completion time with the predicted completion time of the current task stage, and determines whether to generate a multi-stage task progress synchronization table update instruction.
[0048] Beneficial effects:
[0049] This invention provides a method and system for synchronously monitoring the progress of multi-stage crop operations. By constructing a multi-dimensional input system including operation stage type, target plot characteristic parameters, and machinery characteristic parameters, and combining it with an operation task completion time prediction model, the completion time of each stage can be accurately predicted before the operation begins. Furthermore, based on the connection type, the start time of the next stage is automatically generated, achieving systematic control from single-stage time estimation to multi-stage continuous operation time chain derivation. This mechanism effectively avoids problems such as time overlap, excessively long intervals, or misordered stages that exist in traditional manual scheduling, significantly improving the coordination and timeliness of agricultural machinery operation chains.
[0050] Furthermore, a real-time monitoring and residual feedback mechanism is introduced during the actual execution phase of the operation. When a certain operation step is completed, the system automatically collects the actual completion time and compares it with the predicted value. If the deviation exceeds the threshold, a multi-stage operation progress synchronization table update instruction is generated, triggering the recalculation of the time for subsequent stages. Through this closed-loop dynamic correction mechanism, adaptive updates of the entire time series can be achieved when external environmental changes (such as rainfall, mechanical failures, and fluctuations in operation speed) or model prediction errors occur, ensuring that the overall operation progress table always remains consistent with the actual progress on site. Through the recursive calculation of the multi-stage operation time series and the update of the synchronization table, real-time available operation progress information is provided to the scheduling system, thereby achieving optimal matching and dynamic scheduling of equipment, manpower, and land. Attached Figure Description
[0051] The present invention will be further explained below with reference to the accompanying drawings and embodiments:
[0052] Figure 1 This is a flowchart of a method for synchronous monitoring of the progress of multi-stage operations in crops, provided in Embodiment 1 of the present invention.
[0053] Figure 2 This is a flowchart illustrating the execution process of a method for synchronous monitoring of the progress of multi-stage crop operations, as provided in Embodiment 1 of the present invention.
[0054] Figure 3 A flowchart of the method for obtaining surface relief provided in Embodiment 1 of the present invention;
[0055] Figure 4 This is a module connection diagram of a crop multi-stage operation progress synchronous monitoring system provided in Embodiment 2 of the present invention. Detailed Implementation
[0056] To make the technical means, creative features, objectives, and effects of this invention readily understandable, the invention is further described below with reference to specific illustrations. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.
[0057] Example 1
[0058] Please see Figures 1-3 This invention provides a technical solution: a method for synchronous monitoring of the progress of multi-stage crop operations, comprising:
[0059] Step S01: Obtain the operation characteristic parameters of the current operation stage, including the operation stage type, the characteristic parameters of the target plot, and the characteristic parameters of the equipment;
[0060] The types of work processes include turning the soil, fertilizing, sowing, covering with soil, spraying pesticides, and irrigating;
[0061] The characteristic parameters of the target land parcel include land parcel area, land parcel shape complexity, terrain slope, and surface undulation.
[0062] The area of the land parcel is obtained by calling the land parcel vector boundary data stored in the agricultural land parcel management platform or satellite remote sensing system, and the area attribute is directly extracted by indexing the land parcel ID.
[0063] The plot shape complexity is calculated by taking the plot perimeter from the set of plot boundary coordinate points and obtaining it based on the ratio of the perimeters of circles with equal areas. The plot shape complexity is used to characterize the irregularity of the plot.
[0064] The formula for calculating the complexity of the land parcel shape is: In the formula, For the complexity of the plot shape, The perimeter of the land parcel boundary. This represents the area of the land parcel.
[0065] The terrain slope is obtained in real time by tilt sensors on the operating machinery;
[0066] The method for obtaining the surface relief includes:
[0067] Multiple elevation data points on the target site were obtained through drone-based lidar scanning.
[0068] The surface relief is obtained by calculating the variance of multiple elevation data points within the plot.
[0069] The formula for calculating the surface relief is: In the formula, This refers to the surface relief. For the first plot of land Elevation of each sampling point The average elevation is obtained from UAV lidar scanning, GPS altimetry, or RTK measurement, where n is the total number of elevation sampling points.
[0070] The machine's characteristic parameters include its working width and its rated working speed;
[0071] It should be noted that the machine's operating width is obtained by reading equipment parameters from the machine's control unit or by automatically calculating it through the spacing of historical operating trajectories, and is used to characterize the coverage area capability per unit time.
[0072] The rated operating speed of the machinery is obtained by measuring the equipment model database or the combined GPS and control unit of the machinery, and is used to characterize the machinery's travel efficiency under standard working conditions;
[0073] Step S02: Obtain the start time of the current task stage, input the obtained start time of the current task stage and task feature parameters into the task completion time prediction model, and predict the task completion time of the current task stage;
[0074] It should be noted that the start time of the current operation is the predetermined start time in the original operation plan, that is, the time set manually.
[0075] The training method for the task completion time prediction model includes:
[0076] Obtain the dataset of each operation stage of the historical target land parcel, the dataset including the start time of the operation stage, operation characteristic parameters, and task completion time of the operation stage;
[0077] Discrete features in the task feature parameters, such as task type, are encoded using one-hot encoding. Continuous features, such as plot area, plot shape complexity, terrain slope, and surface relief, are standardized using Z-score to eliminate dimensional differences. The processed task start time and task feature parameters are used to form the input feature vector, and the task completion time is used as the model output.
[0078] Based on a preset ratio, the dataset is divided into a training set, a validation set, and a test set. Optionally, the preset ratio is 7:2:1.
[0079] Gradient boosting regression trees are used as the prediction model. The basic parameters are: 100-300 decision trees, maximum depth of a single tree of 5-8 (to control the risk of overfitting), learning rate of 0.01-0.1 (to balance convergence speed and accuracy), and minimum number of sample splits per tree of 10.
[0080] The model is trained using a training set, with mean squared error as the loss function to quantify the deviation between the predicted time and the actual time. Gradient descent is used to update the parameters of each tree based on the negative gradient of the loss function.
[0081] Meanwhile, the Bayesian optimization method was used to tune the hyperparameters. The hyperparameter search range was: 150-200 decision trees, 6-10 layers for the maximum depth of a single tree, and 2-6 samples for the minimum number of nodes to split. The optimal combination of hyperparameters was selected with the validation set accuracy as the optimization objective.
[0082] Every 20 trees are iterated, the mean squared error of the validation set is calculated; when the mean squared error decreases by less than 0.01 for three consecutive times, training is stopped to avoid the model overfitting the details of the training set, and the model parameters with the minimum mean squared error of the validation set are saved.
[0083] The trained model is evaluated using a validation set, and core evaluation metrics are calculated. A model with an overall accuracy of ≥95% is considered to have met the performance evaluation criteria. Once the model has met the performance evaluation criteria, it can be used.
[0084] Step S03: Obtain the connection type between the next operation and the current operation. The connection type includes no connection type, sequential connection type, parallel connection type, and condition-triggered connection type.
[0085] The "no connection" type indicates that there is no time dependency between the next task and the current task, meaning that the start time of the next task is independent of the completion time of the current task.
[0086] The sequential connection type indicates that the next operation can only be started after all the current operations are completed;
[0087] Parallel connection type means that two operation links can be partially overlapped in terms of spatial area or time period, but operation interference constraints must be met, such as spraying operations needing to avoid areas where fertilizers are being applied;
[0088] Condition-triggered connection type, that is, the start of the next step does not depend on the completion time of the current step, but depends on the current step reaching a certain state threshold or characteristic condition. For example, when the coverage density of the current job reaches a set ratio (such as 80%), the next job can start in advance.
[0089] Step S04: Based on the connection type and the task completion time of the current task, generate the start time of the next task, obtain the task feature parameters of the next task, and predict the end time of the next task based on the start time and task feature parameters of the next task.
[0090] The method for generating the start time of the next task based on the connection type and the task completion time of the current task includes:
[0091] A time dependency graph of task steps is constructed using historical task records. The time dependency graph of task steps consists of nodes. and edge composition, Indicates the first Each step of the operation Indicates work process and The connection relationships are defined, with each edge accompanied by a connection type label. This is used to distinguish between non-connected types, sequential connection types, parallel connection types, and conditionally triggered connection types.
[0092] Based on the time dependency graph of the task steps, the connection type between the current task step and the next task step is obtained. ;
[0093] Obtain the predicted completion time of the current task and the connection type between the current task and the next task. Determine the start time for the next work step;
[0094] The process involves obtaining the predicted completion time of the current task and the connection type between the current task and the next task. Methods for determining the start time of the next work step include:
[0095] When the connection type is no connection type, the start time of the next operation is the predetermined start time of that operation in the original operation plan;
[0096] When the connection type is sequential connection type, the start time of the next operation is the end time of the current operation plus the buffer time for the preparation of the next operation.
[0097] When the connection type is parallel connection type, the start time of the next operation is the end time of the current operation minus the allowed operation time overlap time;
[0098] When the connection type is a condition-triggered connection type, the start time of the next operation is the time when the triggering condition is met.
[0099] Step S05: Derive the time links of multiple operations sequentially through a recursive method to form a time sequence of multiple operations and a progress synchronization table for multiple operations.
[0100] Step S06: After the current task is completed, obtain the actual completion time of the current task, compare and analyze the actual completion time with the predicted completion time of the current task, and determine whether to generate a multi-stage task progress synchronization table update instruction.
[0101] The method for comparing and analyzing the actual completion time with the predicted completion time of the current task to determine whether to generate a multi-stage task progress synchronization table update instruction includes:
[0102] Obtain the residual value between the actual completion time and the predicted completion time of the current task. Compare and analyze the residual value with a preset residual threshold. If the residual value is greater than or equal to the preset residual threshold, generate a multi-stage task progress synchronization table update instruction. If the residual value is less than the preset residual threshold, do not generate a multi-stage task progress synchronization table update instruction.
[0103] When a multi-stage operation progress synchronization table update instruction is generated, the actual completion time of the current operation is obtained, the predicted completion time of the current operation stage is updated to the actual completion time, and the multi-stage operation progress synchronization table is updated.
[0104] Specifically, upon detecting the actual completion of the current stage, the system automatically obtains the residual value between the actual completion time and the predicted completion time of the current task stage. The residual value is then compared and analyzed with a preset residual threshold. When the residual value is greater than or equal to the preset residual threshold, the system automatically triggers a recalculation process for the multi-stage task time series. This ensures that the start and end times of the next task stage are calculated according to the new time benchmark, and the multi-stage task progress synchronization table is dynamically updated. This update process ensures that the entire task plan table is consistent with the actual on-site task progress, preventing time conflicts or resource scheduling imbalances between stages.
[0105] In this embodiment, by constructing a multi-dimensional input system that includes the type of work stage, characteristic parameters of the target plot, and characteristic parameters of the machinery, and combining it with a work task completion time prediction model, the completion time of each stage can be accurately predicted before the start of the work. Furthermore, the start time of the next stage can be automatically generated based on the connection type, achieving systematic control from single-stage time estimation to multi-stage continuous operation time chain derivation. This mechanism effectively avoids problems such as time overlap, excessively long intervals, or misordered stages that exist in traditional manual scheduling, significantly improving the coordination and timeliness of agricultural machinery operation chains.
[0106] Furthermore, a real-time monitoring and residual feedback mechanism is introduced during the actual execution phase of the operation. When a certain operation step is completed, the system automatically collects the actual completion time and compares it with the predicted value. If the deviation exceeds the threshold, a multi-stage operation progress synchronization table update instruction is generated, triggering the recalculation of the time for subsequent stages. Through this closed-loop dynamic correction mechanism, adaptive updates of the entire time series can be achieved when external environmental changes (such as rainfall, mechanical failures, and fluctuations in operation speed) or model prediction errors occur, ensuring that the overall operation progress table always remains consistent with the actual progress on site. Through the recursive calculation of the multi-stage operation time series and the update of the synchronization table, real-time available operation progress information is provided to the scheduling system, thereby achieving optimal matching and dynamic scheduling of equipment, manpower, and land.
[0107] Example 2
[0108] Please see Figure 4 This invention provides a system for synchronously monitoring the progress of multiple stages of crop operations, used to implement a method for synchronously monitoring the progress of multiple stages of crop operations. The system includes:
[0109] The parameter acquisition module acquires the operational characteristic parameters of the current operation stage, including the operation stage type, the target plot's characteristic parameters, and the equipment's characteristic parameters.
[0110] The task time prediction module obtains the start time of the current task stage, inputs the obtained start time of the current task stage and task feature parameters into the task completion time prediction model, and predicts the completion time of the current task stage.
[0111] The connection type acquisition module retrieves the connection type between the next operation stage and the current operation stage.
[0112] The end time acquisition module generates the start time of the next task based on the connection type and the task completion time of the current task, acquires the task feature parameters of the next task, and predicts the end time of the next task based on the start time and task feature parameters of the next task.
[0113] The task progress synchronization table module derives the time chain of multiple tasks in a recursive manner, forming a time sequence of multiple tasks and a task progress synchronization table.
[0114] The update module, after the current task stage is completed, obtains the actual completion time of the current task, compares and analyzes the actual completion time with the predicted completion time of the current task stage, and determines whether to generate a multi-stage task progress synchronization table update instruction.
[0115] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for synchronously monitoring the progress of multi-stage crop operations, characterized in that, include: Step S01: Obtain the operation characteristic parameters of the current operation stage, including the operation stage type, the characteristic parameters of the target plot, and the characteristic parameters of the equipment; Step S02: Obtain the start time of the current task stage, input the obtained start time of the current task stage and task feature parameters into the task completion time prediction model, and predict the task completion time of the current task stage; Step S03: Obtain the connection type between the next work step and the current work step; Step S04: Based on the connection type and the task completion time of the current task, generate the start time of the next task, obtain the task feature parameters of the next task, and predict the end time of the next task based on the start time and task feature parameters of the next task. Step S05: Derive the time links of multiple operations sequentially through a recursive method to form a time sequence of multiple operations and a progress synchronization table for multiple operations. Step S06: After the current task is completed, obtain the actual completion time of the current task, compare and analyze the actual completion time with the predicted completion time of the current task, and determine whether to generate a multi-stage task progress synchronization table update instruction.
2. The method for synchronous monitoring of multi-stage crop operation progress according to claim 1, characterized in that, The types of operational steps include turning the soil, fertilizing, sowing, covering with soil, spraying pesticides, and irrigating; The characteristic parameters of the target land parcel include land parcel area, land parcel shape complexity, terrain slope, and surface undulation. The machine's characteristic parameters include its working width and its rated working speed.
3. The method for synchronous monitoring of multi-stage crop operation progress according to claim 2, characterized in that, The method for obtaining the surface relief includes: Multiple elevation data points on the target site were obtained through drone-based lidar scanning. The surface relief is obtained by calculating the variance of multiple elevation data points within the plot. The complexity of the land parcel shape is calculated by taking the perimeter of the land parcel from the set of coordinate points on the parcel boundary, and is obtained based on the ratio of the perimeter of circles with equal areas.
4. The method for synchronous monitoring of multi-stage crop operation progress according to claim 1, characterized in that, The connection types include no connection type, sequential connection type, parallel connection type, and condition-triggered connection type; Among them, the no-connection type means that there is no time dependency between the next operation and the current operation, that is, the start time of the next operation is independent of the completion time of the current operation. The sequential connection type indicates that the next operation can only be started after all the current operations are completed; Parallel connection type means that two work steps can be partially overlapped in terms of spatial area or time period, but work interference constraints must be met; Condition-triggered connection type, that is, the start of the next link does not depend on the completion time of the current link, but depends on the current link reaching a certain state threshold or characteristic condition.
5. The method for synchronous monitoring of multi-stage crop operation progress according to claim 1, characterized in that, The training method for the task completion time prediction model includes: Obtain the dataset of each operation stage of the historical target land parcel, the dataset including the start time of the operation stage, operation characteristic parameters, and task completion time of the operation stage; Discrete features in the task feature parameters are encoded using one-hot encoding, and continuous features are standardized using Z-score to eliminate dimensional differences. The start time of the task steps and the task feature parameters are used as the input feature vector, and the task completion time of the task steps is used as the output of the model. Based on a preset ratio, the dataset is divided into training, validation, and test sets. Gradient boosting regression trees are used as the prediction model. The model is trained using the training set, and mean squared error is used as the loss function to quantify the deviation between the predicted and actual time consumption. Gradient descent is used to update the parameters of each tree based on the negative gradient of the loss function. The mean squared error of the validation set is calculated every 20 trees. Training is stopped when the mean squared error decreases by less than 0.01 for three consecutive times. At the same time, Bayesian optimization is used to tune the hyperparameters. The trained model is evaluated using a validation set, and core evaluation metrics are calculated. A model with an overall accuracy of ≥95% is considered to have met the performance evaluation criteria. Once the model has met the performance evaluation criteria, it can be used.
6. The method for synchronous monitoring of multi-stage crop operation progress according to claim 1, characterized in that, The method for generating the start time of the next task based on the connection type and the task completion time of the current task includes: A time dependency graph of task steps is constructed using historical task records. The time dependency graph of task steps consists of nodes. and edge composition, Indicates the first Each step of the operation Indicates work process and The connection relationships are defined, with each edge accompanied by a connection type label. This is used to distinguish between non-connected types, sequential connection types, parallel connection types, and conditionally triggered connection types. Based on the time dependency graph of the task steps, the connection type between the current task step and the next task step is obtained. ; Obtain the predicted completion time of the current task and the connection type between the current task and the next task. Determine the start time for the next work step.
7. The method for synchronous monitoring of multi-stage crop operation progress according to claim 6, characterized in that, The process involves obtaining the predicted completion time of the current task and the connection type between the current task and the next task. Methods for determining the start time of the next work step include: When the connection type is no connection type, the start time of the next operation is the predetermined start time of that operation in the original operation plan; When the connection type is sequential connection type, the start time of the next operation is the end time of the current operation plus the buffer time for the preparation of the next operation. When the connection type is parallel connection type, the start time of the next operation is the end time of the current operation minus the allowed operation time overlap time; When the connection type is a condition-triggered connection type, the start time of the next operation is the time when the triggering condition is met.
8. The method for synchronous monitoring of multi-stage crop operation progress according to claim 1, characterized in that, The method for comparing and analyzing the actual completion time with the predicted completion time of the current task to determine whether to generate a multi-stage task progress synchronization table update instruction includes: Obtain the residual value between the actual completion time and the predicted completion time of the current task stage. Compare and analyze the residual value with a preset residual threshold. If the residual value is greater than or equal to the preset residual threshold, a multi-stage task progress synchronization table update instruction is generated. If the residual value is less than the preset residual threshold, no multi-stage task progress synchronization table update instruction is generated.
9. The method for synchronous monitoring of multi-stage crop operation progress according to claim 8, characterized in that, When the multi-stage operation progress synchronization table update instruction is generated, the actual completion time of the current operation is obtained, the predicted completion time of the current operation stage is updated to the actual completion time, and the multi-stage operation progress synchronization table is updated.
10. A system for synchronously monitoring the progress of multi-stage crop operations, used to implement the method for synchronously monitoring the progress of multi-stage crop operations as described in any one of claims 1-9, characterized in that, The system includes: The parameter acquisition module acquires the operational characteristic parameters of the current operation stage, including the operation stage type, the characteristic parameters of the target plot, and the characteristic parameters of the equipment. The task time prediction module obtains the start time of the current task stage, inputs the obtained start time of the current task stage and task feature parameters into the task completion time prediction model, and predicts the completion time of the current task stage. The connection type acquisition module retrieves the connection type between the next operation and the current operation. The end time acquisition module generates the start time of the next task based on the connection type and the task completion time of the current task, acquires the task feature parameters of the next task, and predicts the end time of the next task based on the start time and task feature parameters of the next task. The task progress synchronization table module derives the time chain of multiple tasks in a recursive manner, forming a time sequence of multiple tasks and a task progress synchronization table. The update module, after the current task stage is completed, obtains the actual completion time of the current task, compares and analyzes the actual completion time with the predicted completion time of the current task stage, and determines whether to generate a multi-stage task progress synchronization table update instruction.