An intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation

Abstract

This invention relates to an intelligent agricultural machinery scheduling method based on multi-source perception and adaptation to hilly terrain, belonging to the field of digital information transmission. The method includes: constructing a three-dimensional digital farmland model, extracting plot slope, fragmentation, and access attributes, and establishing a comprehensive cost function between agricultural machinery and plots; identifying the agricultural machinery's operating status based on RTK-GNSS and IMU data, and calculating an estimated effective operating area based on plot boundaries; scheduling agricultural machinery based on comprehensive cost, service quality, and price factors, and dynamically guiding pricing during the busy farming season based on terrain complexity and supply-demand intensity; and, under weak network conditions, performing compensated uploading, time-series alignment, and area recalculation on the operating data, and writing back and updating scheduling, metering, and settlement data. This invention reduces invalid dispatching and empty runs in hilly terrain, improves the accuracy and verifiability of effective operating area estimation, and reduces settlement disputes.

Description

Technical Field

[0001] This invention relates to the field of digital information transmission, and in particular to an intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation. Background Technology

[0002] Agricultural mechanization in hilly and mountainous areas is significantly limited by undulating terrain, fragmented plots, and road conditions, making operation organization and scheduling difficult and often accompanied by unstable communication coverage. For hilly operation environments, existing technologies use DEM (Digital Image Model) to construct terrain and energy consumption models and optimize operation paths, as well as conduct research on full-coverage path planning under constraints such as plot boundaries, obstacles, and turning methods. Other researchers have proposed DEM-based 3D raster modeling and full-coverage path planning methods for complex hilly farmland, comprehensively considering path length, elevation difference, and turning costs in energy consumption modeling to improve operational feasibility and energy consumption optimization in complex terrain. Still others have constructed energy consumption models considering slope and operation direction angles for full-coverage operations in hilly and mountainous farmland, and adopted improved swarm intelligence optimization strategies to achieve collaborative optimization of single-plot operation direction and multi-plot traversal order, providing a reference for energy consumption quantification and operation sequence optimization in complex terrain. For multi-machine, multi-plot collaborative operations, existing research has constructed multi-objective scheduling models with time window constraints and solved them using an improved ensemble particle swarm optimization algorithm. Other research has introduced learning strategies into variable time window scheduling to improve dynamic response capabilities. While these studies are valuable for terrain modeling, route planning, and scheduling optimization, they typically fail to integrate "hilly accessibility constraints" such as slope, minimum road width, and minimum turning capacity with the feasibility assessment, risk penalties, and economic objectives of multi-machine scheduling in a unified model. This can easily lead to biased scheduling accessibility assessments or invalid dispatches.

[0003] On the other hand, existing research on operational status identification and operational parameter calculation mainly relies on the positioning trajectory of the BeiDou satellite system and the information of land parcel boundaries to distinguish between pure operation, turning, rest, and transfer states, or on trajectory density clustering to identify field operations, empty transfers, and rest states. While these methods provide a basis for status identification, in hilly areas with weak network conditions, the trajectory and status sequences are prone to breaks and time gaps. If the closed-loop business mechanism of "weak network data integrity rate—compensation upload—time sequence alignment—effective operation area recalculation" is not embedded into the scheduling and settlement verification process, inconsistencies in measurement verification and regulatory delays may still occur. Existing reliable data transmission solutions in weak network environments typically employ mechanisms such as local caching, retransmission, and breakpoint resumption to improve transmission success rates, but these mostly remain at the transmission layer reliability level and do not form a closed-loop coupling with agricultural machinery operation status verification, effective area recalculation, and scheduling and settlement processes. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide an intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation, thus solving the deficiencies of the prior art.

[0005] The objective of this invention is achieved through the following technical solution: an intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation, the method comprising:

[0006] Step 1: Construct a three-dimensional digital farmland model and establish a comprehensive cost function; calculate the estimated effective operating area based on the operation status classifier.

[0007] Step 2: Establish a single-objective scheduling optimization model that includes three factors: comprehensive cost function, service quality revenue, and price deviation penalty. Solve the single-objective scheduling optimization model under normal conditions to obtain the scheduling results. During the busy farming season, calculate the dynamic guidance unit price of the land parcel based on terrain complexity and supply and demand intensity. After updating the price deviation penalty to the excess penalty term related to the dynamic guidance unit price, solve the single-objective scheduling optimization model to obtain the scheduling results for the busy farming season.

[0008] Step 3: Under weak network or network outage conditions, trigger compensation upload based on the integrity rate of the work data, perform time-series alignment on the re-uploaded trajectory data and status data, recalculate the estimated effective work area, and update the time parameters, cost parameters, and metering and settlement data in the single-objective scheduling optimization model based on the recalculated estimated effective work area. If the updated result affects the feasibility of the original scheduling, perform rescheduling or consistency correction.

[0009] Step one includes:

[0010] A1. Obtain the vector boundaries of the land parcels, the digital elevation model (DEM), and the network of farm roads to construct a three-dimensional digital farmland model and extract the land parcels. slope Fragmentation And the attributes of the mechanized farming passage, and establish agricultural machinery To the plot The comprehensive cost function is ,in, To reach the total operation time, For energy consumption costs, As a risk penalty item, , and They are respectively , and Weighting coefficients;

[0011] A2. Collect RTK-GNSS positioning data and IMU data to construct a job status classifier. Based on the operation status and plot boundaries Extracting the set of valid trajectory points The effective working area estimate was calculated using the grid cumulative method. .

[0012] In A1, an efficiency attenuation coefficient is introduced to quantify the impact of hilly terrain on operational efficiency: And estimate the total time to reach and complete the task as follows: ,in, For agricultural machinery to move from its current location to the plot of land Arrival time, For the plot of land Estimated effective working area For agricultural machinery Rated operating efficiency and They are respectively and The parameters to be fitted;

[0013] By transforming the safety and traffic constraints of agricultural machinery operations into optimizable penalty costs, the risk penalty term is obtained as follows: ,in, This is the maximum climbing limit for agricultural machinery. For access to the plot Minimum passable width of the road For the width of agricultural machinery, and They are respectively and The penalty coefficient.

[0014] A2 includes:

[0015] Collect RTK-GNSS positioning data and IMU data to construct an operation status classifier. , where the input feature vector Output status In this context, 0 indicates that the agricultural machinery is idle or in motion, and 1 indicates that it is in operation. These represent the accelerations along the x, y, and z axes, respectively, collected by the IMU. These represent the angular velocities along the x, y, and z axes, respectively, collected by the IMU. This indicates the speed of the agricultural machinery as collected by GNSS.

[0016] Based on the status identification results and the land parcel boundaries Extracting the set of valid trajectory points: The grid cumulative method was used to calculate the plots. Estimated effective working area: ,in, For a moment GPS coordinates For a single grid cell, For the plot of land Raster set, For indicator functions, This represents the grid area.

[0017] Step two includes:

[0018] Decision variables Indicates agricultural machinery With the plot of land Based on the allocation relationship, a single-objective multi-factor weighted scheduling optimization model is established: ,in, For the comprehensive cost function, For agricultural machinery Service Plot The benefits of service quality Penalty for price deviation , and They are respectively , and Weighting coefficients;

[0019] Under normal conditions, the single-objective scheduling optimization model is solved by applying unique land allocation constraints, agricultural machinery terrain adaptation constraints, and working time constraints to obtain the scheduling results.

[0020] During the busy farming season, land parcels are calculated based on terrain complexity and supply and demand intensity. Dynamic guidance price per acre: After updating the price deviation penalty to an excess penalty term related to the dynamic guidance unit price, the single-objective scheduling optimization model is solved to obtain the scheduling results for the busy farming season. As the benchmark price, For demand intensity, To ensure sufficient supply and demand, The minimum instruction price, The maximum instruction price, , , and They are respectively , , and The weighting coefficients.

[0021] The aforementioned dynamic guidance unit price Introducing a price deviation penalty term, we obtain the excess penalty for the busy farming season: ,in, For agricultural machinery The price per acre, when When, the penalty is 0; when At that time, the penalty increases with the excess amount.

[0022] Step three includes:

[0023] During network outages, trajectory points, status sequences, and order statuses are written to a local persistent queue. After network recovery, breakpoint resume and compensated uploads are performed. The re-uploaded trajectory points and status sequences are deduplicated and time-sequence aligned according to timestamps, and a new set of valid trajectory points is generated and the estimated effective work area is recalculated. Based on the recalculated estimated effective work area, the arrival and total work duration, comprehensive cost function, and working hours are updated, and the metering and settlement data are updated. If the updated results affect the feasibility of the original scheduling, rescheduling or consistency correction is performed.

[0024] The compensated upload is implemented using a hybrid communication protocol stack based on HTTPS and MQTT, and after network recovery, it performs breakpoint resumption based on an exponential backoff strategy. The retry interval during the network recovery phase meets the following requirements. ,in, For the first The waiting interval for each retry. This is the initial retry interval. For random jitter items, This is the maximum retry interval limit, via The function ensures that the retry interval is within a controllable range, avoiding network congestion caused by high-frequency retries.

[0025] The completeness rate of the operation data includes:

[0026] Calculate data integrity rate: I =1- N loss / N gen When data integrity rate Below the preset threshold When this occurs, a compensation upload mechanism is triggered, in which... This refers to the total amount of data generated during agricultural machinery operations. This represents the amount of data lost.

[0027] Compared with the prior art, the present invention has at least the following beneficial effects: by coupling hilly accessibility constraints, operation status verification, weak network fault tolerance, compensation upload and recalculation write-back and scheduling optimization in a closed loop, it reduces invalid order dispatch and empty runs in hilly scenarios, improves the accuracy and verifiability of effective operation area estimation, reduces settlement disputes, and achieves consistency of scheduling, metering and settlement under weak network conditions, while improving the scheduling response capability in the order pulse scenario of the busy farming season.

[0028] This invention does not simply combine terrain modeling, operation status identification, weak network transmission, and scheduling optimization in parallel. Instead, it forms a coupled closed loop of "metering—scheduling—transmission—recalculation—settlement consistency." It extracts effective trajectory points and calculates the estimated effective operation area through a dual constraint of an operation status classifier and land parcel boundaries. ; will the As the total time for arrival and operation and comprehensive cost function The input is then fed into the objective function and time constraints of the scheduling optimization model; when weak or broken networks cause incomplete data, compensation upload is triggered based on the data integrity rate to complete time-series alignment and recalculation of effective work area, thereby updating... and This will be used to perform consistency correction on scheduling and settlement. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the process of the present invention.

[0030] Figure 2 This is a flowchart illustrating the job status classifier.

[0031] Figure 3 This is a schematic diagram of the solution process for the scheduling optimization model. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application provided below with reference to the accompanying drawings is not intended to limit the scope of protection of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The present invention will be further described below with reference to the accompanying drawings.

[0033] This invention addresses the consistency problem of agricultural machinery scheduling, metering, and settlement in hilly environments, proposing an intelligent scheduling method that couples terrain accessibility modeling, operational status identification, weak network fault tolerance, and scheduling optimization in a closed loop.

[0034] like Figure 1 As shown, the method includes the following steps:

[0035] S1. 3D modeling and comprehensive cost construction of hilly scenes;

[0036] Acquire data on land parcel vector boundaries, digital elevation model (DEM), and farm road network; construct a 3D digital farmland model; and extract land parcels. slope Fragmentation And the attributes of mechanized farming access, to establish agricultural machinery To the plot Comprehensive cost function: ,in, To reach the total operation time, Energy consumption cost is a linear function positively correlated with travel distance and operating area. As a risk penalty item, , and They are respectively , and The weighting coefficients.

[0037] Furthermore, to quantify the impact of hilly terrain on operational efficiency, an efficiency attenuation coefficient is introduced. And estimate the total time to reach and complete the task as follows: ,in, It is the agricultural machinery moving from its current location to the plot of land. Arrival time, For the plot of land Estimated effective working area It is agricultural machinery Rated operating efficiency and They are respectively and The parameters to be fitted.

[0038] Risk penalty items The form of "hard constraints + soft penalties" is then used as follows:

[0039] ,

[0040] When constraints are violated, the penalty cost increases significantly, guiding the optimization algorithm to automatically avoid infeasible solutions. In the formula, This is the maximum climbing limit for agricultural machinery. It leads to the plot of land. Minimum passable width of the road It refers to the width of the agricultural machinery. and They are respectively and The penalty coefficient.

[0041] Through the above modeling, hilly accessibility constraints such as slope and road conditions are uniformly incorporated into the scheduling cost and feasibility assessment process.

[0042] S2, Multi-source fusion state perception and effective working area estimation;

[0043] like Figure 2 As shown, RTK-GNSS positioning data and IMU data are collected, and the location and status of the agricultural machinery are analyzed in real time to construct an operation status classifier as follows:

[0044] ,

[0045] Among them, the input feature vector Output status Where 0 indicates idle or running status, and 1 indicates working status; Three-axis acceleration data collected by the IMU The three-axis angular velocities collected by the IMU This indicates the speed of agricultural machinery as collected by GNSS.

[0046] Based on the status identification results and the land parcel boundaries Extracting the set of valid trajectory points:

[0047] ,

[0048] in, For a moment spatial coordinates, For the target plot Geographical boundaries, This indicates that the agricultural machinery is in operation. This set only contains valid trajectory points of the agricultural machinery in operation within the plot.

[0049] To achieve verifiable estimates of the work area, the grid cumulative method is used to calculate the land parcels. Estimated effective working area: ,in, For a single grid cell, For the plot of land The corresponding raster set, For indicator functions, This refers to the grid area. By using a dual constraint method of "operation status + plot boundary", it is possible to avoid including empty driving, U-turns, and stop trajectories in the effective operating area.

[0050] The effective area estimate As and The input quantity is then fed into the objective function and time constraints of the scheduling optimization model.

[0051] S3, weak network fault tolerance, compensated upload and recalculation writeback;

[0052] Establish a hybrid communication protocol stack based on HTTPS and MQTT; during network outages, write trajectory points, status sequences, and order statuses to a local persistent queue, and execute breakpoint resumption based on an exponential backoff strategy after network recovery. The retry interval satisfies:

[0053] ,

[0054] in, For the first The waiting interval for each retry. This is the initial retry interval. For random jitter items, This is the maximum retry interval limit.

[0055] At the same time, calculate the data integrity rate:

[0056] I =1- N loss / N gen ,

[0057] in, , This refers to the total amount of data generated during agricultural machinery operations. This refers to the amount of data lost. When the data integrity rate... Below the preset threshold At that time, the compensation upload mechanism is triggered.

[0058] After the network is restored and the compensation upload is completed, the retransmitted trajectory points and state sequences are deduplicated and time-series aligned according to timestamps to regenerate a valid set of trajectory points. And recalculate the effective working area The updated version Write back to the order metering and scheduling module for updating. , And the consumption of working hours, and trigger a consistency check of the settlement amount; when the updated If the work time allocation result affects the feasibility of the original scheduling, the affected set of land parcels and the associated agricultural machinery set will be rescheduled.

[0059] S4. Multi-factor weighted scheduling optimization and dynamic pricing during the busy farming season;

[0060] like Figure 3 As shown, under normal circumstances, decision variables Indicates agricultural machinery With the plot of land Based on the allocation relationship, a single-objective multi-factor weighted scheduling optimization model is established:

[0061] ,

[0062] in, For the comprehensive cost function, For agricultural machinery Service Plot The benefits of service quality Penalty for price deviation , and They are respectively , and The weighting coefficients.

[0063] The algorithm applies unique land allocation constraints, agricultural machinery terrain adaptation constraints, and work time constraints, and uses integer programming to solve the problem, yielding the scheduling results. The constraints are as follows:

[0064] ,

[0065] ,

[0066] ,

[0067] in, For agricultural machinery The available working hours window. Through the above constraints, it is ensured that each plot is allocated to only one agricultural machine, and the allocation result satisfies terrain adaptability and working hours constraints.

[0068] During the busy farming season, land parcels are calculated based on terrain complexity and supply and demand intensity. Dynamic guidance price per acre:

[0069] ,

[0070] in, As the benchmark price, For demand intensity, To ensure sufficient supply and demand, and These are the minimum and maximum instruction prices, respectively. , , and They are respectively , , and The weighting coefficients.

[0071] During the busy farming season, penalties will be imposed for price deviations from normal levels. Replace with excess penalty items related to dynamic guidance unit price:

[0072] ,

[0073] in, For agricultural machinery The price is quoted per acre. When When the penalty is 0, At that time, the penalty increases with the excess amount.

[0074] The objective function should be updated accordingly during the busy farming season:

[0075] ,

[0076] During the busy farming season, an approximate solution strategy combining greedy initialization and local search is adopted. First, an initial feasible solution is constructed based on the priority of the plots. Then, within a preset time limit, the allocation result is locally improved through migration and exchange operations, thereby outputting a fast and usable near-optimal scheduling result.

[0077] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and improvements, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for intelligent scheduling of agricultural machinery based on multi-source perception and hilly scene adaptation, characterized in that, The method includes: Step 1: Construct a three-dimensional digital farmland model and establish a comprehensive cost function; calculate the estimated effective operating area based on the operation status classifier. Step 2: Establish a single-objective scheduling optimization model that includes three factors: comprehensive cost function, service quality revenue, and price deviation penalty. Solve the single-objective scheduling optimization model under normal conditions to obtain the scheduling results. During the busy farming season, calculate the dynamic guidance unit price of the land parcel based on terrain complexity and supply and demand intensity. After updating the price deviation penalty to the excess penalty term related to the dynamic guidance unit price, solve the single-objective scheduling optimization model to obtain the scheduling results for the busy farming season. Step 3: Under weak network or network outage conditions, trigger compensation upload based on the integrity rate of the work data, perform time-series alignment on the re-uploaded trajectory data and status data, recalculate the estimated effective work area, and update the time parameters, cost parameters, and metering and settlement data in the single-objective scheduling optimization model based on the recalculated estimated effective work area. If the updated result affects the feasibility of the original scheduling, perform rescheduling or consistency correction.

2. The intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation according to claim 1, characterized in that, Step one includes: A1. Obtain the vector boundaries of the land parcels, the digital elevation model (DEM), and the network of farm roads to construct a three-dimensional digital farmland model and extract the land parcels. slope Fragmentation And the attributes of the mechanized farming passage, and establish agricultural machinery To the plot The comprehensive cost function is ,in, To reach the total operation time, For energy consumption costs, As a risk penalty item, , and They are respectively , and Weighting coefficients; A2. Collect RTK-GNSS positioning data and IMU data to construct a job status classifier. Based on the operation status and plot boundaries Extracting the set of valid trajectory points The effective working area estimate was calculated using the grid cumulative method. ; In A1, to quantify the impact of hilly terrain on operational efficiency, an efficiency attenuation coefficient is introduced: And estimate the total time to reach and complete the task as follows: ,in, For agricultural machinery to move from its current location to the plot of land Arrival time, For the plot of land Estimated effective working area For agricultural machinery Rated operating efficiency and They are respectively and The parameters to be fitted; By transforming the safety and traffic constraints of agricultural machinery operations into optimizable penalty costs, the risk penalty term is obtained as follows: ,in, This is the maximum climbing limit for agricultural machinery. For access to the plot Minimum passable width of the road For the width of agricultural machinery, and They are respectively and The penalty coefficient.

3. The intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation according to claim 2, characterized in that, A2 includes: Collect RTK-GNSS positioning data and IMU data to construct an operation status classifier. , where the input feature vector Output status In this context, 0 indicates that the agricultural machinery is idle or in motion, and 1 indicates that it is in operation. These represent the accelerations along the x, y, and z axes, respectively, collected by the IMU. These represent the angular velocities along the x, y, and z axes, respectively, collected by the IMU. This indicates the speed of the agricultural machinery as collected by GNSS. Based on the status identification results and the land parcel boundaries Extracting the set of valid trajectory points: The grid cumulative method was used to calculate the plots. Estimated effective working area: ,in, For a moment GPS coordinates For a single grid cell, For the plot of land Raster set, For indicator functions, This represents the grid area.

4. The intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation according to claim 1, characterized in that, Step two includes: Decision variables Indicates agricultural machinery With the plot of land Based on the allocation relationship, a single-objective multi-factor weighted scheduling optimization model is established: ,in, For the comprehensive cost function, For agricultural machinery Service Plot The benefits of service quality Penalty for price deviation , and They are respectively , and Weighting coefficients; Under normal conditions, the single-objective scheduling optimization model is solved by applying unique land allocation constraints, agricultural machinery terrain adaptation constraints, and working time constraints to obtain the scheduling results. During the busy farming season, land parcels are calculated based on terrain complexity and supply and demand intensity. Dynamic guidance price per acre: After updating the price deviation penalty to an excess penalty term related to the dynamic guidance unit price, the single-objective scheduling optimization model is solved to obtain the scheduling results for the busy farming season. As the benchmark price, For demand intensity, To ensure sufficient supply and demand, The minimum instruction price, The maximum instruction price, , , and They are respectively , , and The weighting coefficients.

5. The intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation according to claim 4, characterized in that, The aforementioned dynamic guidance unit price Introducing a price deviation penalty term, we obtain the excess penalty for the busy farming season: ,in, For agricultural machinery The price per acre, when When the penalty is 0, the penalty item is 0; when At that time, the penalty increases with the excess amount.

6. A method for intelligent scheduling of agricultural machinery based on multi-source perception and hilly scene adaptation according to any one of claims 1-5, characterized in that, Step three includes: During network outages, trajectory points, status sequences, and order statuses are written to a local persistent queue. After network recovery, breakpoint resume and compensated uploads are performed. The re-uploaded trajectory points and status sequences are deduplicated and time-sequence aligned according to timestamps, and a new set of valid trajectory points is generated and the estimated effective work area is recalculated. Based on the recalculated estimated effective work area, the arrival and total work duration, comprehensive cost function, and working hours are updated, and the metering and settlement data are updated. If the updated results affect the feasibility of the original scheduling, rescheduling or consistency correction is performed.

7. The intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation according to claim 6, characterized in that, The compensated upload is implemented using a hybrid communication protocol stack based on HTTPS and MQTT, and after network recovery, it performs breakpoint resumption based on an exponential backoff strategy. The retry interval during the network recovery phase meets the following requirements. ,in, For the first The waiting interval for each retry. This is the initial retry interval. For random jitter items, This is the maximum retry interval limit, via The function ensures that the retry interval is within a controllable range, avoiding network congestion caused by high-frequency retries.

8. The intelligent scheduling method for agricultural machinery based on multi-source perception and hilly scene adaptation according to claim 6, characterized in that, The completeness rate of the operation data includes: Calculate data integrity rate: I = 1 - N loss / N gen When data integrity rate Below the preset threshold When this occurs, a compensation upload mechanism is triggered, in which... This refers to the total amount of data generated during agricultural machinery operations. This represents the amount of data lost.

Images