Intelligent operation task scheduling method and system for agricultural equipment

Abstract

The invention discloses an agricultural equipment intelligent operation task scheduling method and system, and the method comprises the steps: building a multi-target scheduling function according to an agricultural equipment operation task type, equipment energy consumption cost, path planning cost, operation time constraint and a resource utilization rate; inputting the real-time environment data and the agricultural equipment state data into a deep neural network for learning and training to obtain an agricultural equipment dynamic collaborative optimization model for outputting a task demand predicted value; dynamically updating model parameters of the dynamic collaborative optimization model according to the farmland operation progress and the sudden working condition; and solving an optimal solution of the multi-target scheduling function by utilizing the updated dynamic collaborative optimization model, and generating a global optimal job task scheduling instruction. According to the method, the problem of intelligent agricultural equipment operation task scheduling under the multi-target condition is solved, and the real-time performance, the accuracy and the overall optimization effect of agricultural equipment scheduling are improved through a deep learning prediction model, a dynamic optimization algorithm and real-time environment data processing.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural machinery automation and intelligent scheduling control, and particularly to an intelligent operation task scheduling method and system for agricultural equipment. Background Art

[0002] With the rapid development of Internet of Things, artificial intelligence and big data technologies, the intelligent operation task scheduling of agricultural equipment has gradually become an important means to improve agricultural production efficiency and reduce production costs. At present, the research on intelligent operation task scheduling of agricultural equipment mostly focuses on the optimization scheduling under single-objective or static conditions, such as the allocation of mechanical equipment based on fixed paths or the optimization planning with the single goal of minimizing energy consumption. However, the environmental conditions faced in actual agricultural operations are more complex. For example, the farmland operation tasks are diversified, the energy consumption cost of agricultural machinery is continuously increasing, the path complexity caused by the irregular layout of farmland plots, and the rigid constraints of operation time. The traditional agricultural equipment task scheduling methods are difficult to take into account energy consumption cost, path optimization, operation time and resource utilization rate under multi-objective conditions, and there are problems such as single scheduling scheme, poor flexibility and inability to quickly respond to the changes in the actual operation environment.

[0003] In addition, the realization of the current agricultural equipment intelligent scheduling technology lacks the in-depth mining and utilization of real-time environmental data and the state data of the equipment itself. Usually, it only relies on static historical data and empirical models, and cannot respond in time to sudden working conditions and the deviation of farmland operation progress. Especially when facing unpredictable events such as sudden failures of agricultural machinery and sudden weather changes, the existing technologies are difficult to make quick decisions and effective scheduling adjustments, resulting in reduced operation efficiency, waste of resources, and even the production plan cannot be completed as scheduled. In addition, the lack of accurate prediction and effective utilization of the real-time state of agricultural machinery further leads to the disconnection between the operation plan and the actual situation, seriously affecting the reliability and efficiency of agricultural production.

[0004] Therefore, the existing agricultural equipment intelligent operation scheduling technology has problems such as poor multi-objective scheduling effect, insufficient real-time response ability, and lack of effective analysis and utilization of dynamic environmental data. The content described in the present invention precisely solves the problem of how to achieve intelligent operation task scheduling of agricultural equipment under multi-objective conditions, and improves the real-time performance, accuracy and overall optimization effect of agricultural equipment scheduling through deep learning prediction models, dynamic optimization algorithms and real-time processing of environmental data. Summary of the Invention

[0005] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this part, the abstract of the specification and the title of the invention of the present application to avoid obscuring the purpose of this part, the abstract of the specification and the title of the invention, and such simplifications or omissions shall not be used to limit the scope of the present invention.

[0006] In view of the above existing problems, the present invention is proposed.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: Establish a multi-objective scheduling function according to the agricultural equipment operation task type, equipment energy consumption cost, path planning cost, operation time constraint and resource utilization rate;

[0008] Input the real-time environmental data and agricultural equipment status data into a deep neural network for learning and training to obtain an agricultural equipment dynamic collaborative optimization model that outputs task demand prediction values;

[0009] Dynamically update the model parameters of the dynamic collaborative optimization model according to the farmland operation progress and sudden working conditions;

[0010] Use the updated dynamic collaborative optimization model to solve the optimal solution of the multi-objective scheduling function and generate a global optimal operation task scheduling instruction.

[0011] As a preferred solution of the intelligent operation task scheduling method for agricultural equipment according to the present invention, the establishment of the multi-objective scheduling function includes:

[0012] Let the decision variable set be x, and define the multi-objective scheduling function as:

[0013] F(x) = w1C e (x) + w2C p (x) + w3T(x) + w4R(x)

[0014] Wherein, C e (x) is the equipment energy consumption cost, C p (x) is the path planning cost, T(x) is the objective function value of the operation time constraint, R(x) is the negative utility function value of the resource utilization rate, w1, w2, w3, w4 are respectively the weight coefficients of the equipment energy consumption cost, path planning cost, operation time constraint, and resource utilization rate, and x represents the operation arrangement, path selection and resource allocation plan of multiple agricultural equipment in different time periods.

[0015] As a preferred solution of the intelligent operation task scheduling method for agricultural equipment according to the present invention, the agricultural equipment operation task type includes at least tilling, sowing, fertilizing, spraying pesticides, harvesting and transporting;

[0016] The equipment energy consumption cost includes, defining E i as the average energy consumption of the i-th equipment per unit time, t i as the duration of this equipment participating in this round of operation, p i as the energy consumption price or conversion factor, then:

[0017]

[0018] The path planning cost includes defining d i as the total distance traveled by the i-th piece of equipment on the operation path, and β as the path cost per unit distance. Then:

[0019]

[0020] The operation time constraint includes that if t max is the maximum allowable operation duration for this stage, and t use is the total time used for all assigned operations. Define the operation time penalty term as:

[0021]

[0022] where γ is the additional penalty coefficient after exceeding the specified time;

[0023] The resource utilization rate includes regarding the resource utilization rate as a positive optimization goal, defining the resource utilization rate as U(x), and letting R(x) = -U(x). Among them, U(x) is expressed as:

[0024]

[0025] When U(x) is higher, it indicates that the resource utilization situation is better, and the corresponding value of R(x) is lower, that is, the negative utility is smaller.

[0026] As a preferred solution of the intelligent operation task scheduling method for agricultural equipment described in the present invention, the agricultural equipment dynamic collaborative optimization model that inputs real-time environmental data and agricultural equipment status data into a deep neural network for learning and training to obtain an output task demand prediction value includes:

[0027] Clean and denoise the real-time collected environmental data and agricultural equipment status data, and perform batch segmentation according to the time stamp;

[0028] Perform normalization processing on data of different dimensions to ensure that the magnitudes of inputs in each dimension are the same;

[0029] Let the input sequence be Q = (q1, q2,..., q T ), where q t is the feature vector at time t, including the preprocessed real-time environmental data and agricultural equipment status data;

[0030] Let the hidden layer state sequence be h t , and the output sequence be Adopt a recurrent neural network structure, and its state update and output prediction are expressed as:

[0031] h t = σ(Wqh ·q t +W hh ·h t-1 +b h )

[0032]

[0033] wherein, W qh , W hh , W hy are respectively the weight vectors input to the hidden layer, from the hidden layer to the hidden layer, and from the hidden layer to the output, and b h , b y are respectively the bias vectors of the hidden layer and the output layer, σ(·) is the activation function, and h t-1 is the state of the hidden layer at the previous moment, is the predicted output at time t;

[0034] Define the loss function wherein is the model output sequence, and y is the actual value of the corresponding task requirement;

[0035] Use the optimization algorithm based on stochastic gradient descent to update W qh , W hh , W hy , b h , b y parameters;

[0036] When the maximum number of training rounds is reached and the error converges, the current recurrent neural network model is output, that is, the finally obtained agricultural equipment dynamic collaborative optimization model is expressed as:

[0037]

[0038] wherein, θ represents all the model parameters obtained after training, is the predicted value of the task requirement made at each time t.

[0039] As a preferred scheme of the agricultural equipment intelligent operation task scheduling method described in the present invention, the real-time environmental data at least includes meteorological elements, soil and plot information, and pest and disaster monitoring information;

[0040] Among them, the meteorological elements include temperature, humidity, wind speed, precipitation, light intensity, and weather prediction information, the soil and plot information includes soil temperature, soil humidity, soil nutrient components, farmland plot boundaries, and terrain change data, and the pest and disaster monitoring information includes the distribution of pest occurrence areas, pest density, and disaster warning information;

[0041] The agricultural equipment status data at least includes position and motion information, energy / fuel information, mechanical health, and operation load information;

[0042] Among them, the position and motion information includes GPS coordinates, current operation speed, and moving direction. The energy / fuel information includes fuel reserve, remaining battery power, and fuel / battery consumption rate. The mechanical health includes component operating temperature, vibration amplitude, cumulative operating duration, and fault diagnosis results. The operation load information includes current operation type, actual working time, carried tools, status of attached equipment, and operation load rate.

[0043] As a preferred solution of the intelligent operation task scheduling method for agricultural equipment according to the present invention, the model parameters of the dynamic collaborative optimization model are dynamically updated according to the farm operation progress and sudden working conditions, including:

[0044] Define η as the completion ratio of the target in this stage. If |η - η arget |≥ω, it indicates that there is a large deviation between the operation progress and the target value, and the online update of model parameters is immediately triggered;

[0045] Among them, η arget is the expected completion ratio, and ω is the set threshold;

[0046] When one of the conditions that the weather condition critical value exceeds the safety threshold or the failure rate of the key parts of the equipment exceeds the set upper limit is met, it is regarded as a sudden working condition;

[0047] Adopt an incremental learning strategy, input the newly collected environmental data, equipment status data, and progress deviation information into the dynamic collaborative optimization model, and adaptively adjust the network weights and learning rate;

[0048] If the prediction accuracy of the updated model is better than that of the model before update, the update result is retained; otherwise, roll back to the original model and record relevant information.

[0049] As a preferred solution of the intelligent operation task scheduling method for agricultural equipment according to the present invention, the failure rate of the key parts includes but is not limited to equipment power system failure and GPS malfunction.

[0050] As a preferred solution of the intelligent operation task scheduling method for agricultural equipment according to the present invention, use the updated dynamic collaborative optimization model to solve the optimal solution of the multi-objective scheduling function and generate a global optimal operation task scheduling instruction, including:

[0051] Iteratively solve the following objectives through the ant colony algorithm:

[0052] F(x) = w1C e (x) + w2C p(x) + w3T(x) + w4R(x)

[0053] Determine whether the objective function value of the solution obtained in the current iteration is better than the historical optimal value F best , if F(x) < F best , then update F best and record the corresponding solution;

[0054] If F(x) ≥ F best , then continue the iteration until the maximum number of iterations is reached;

[0055] After obtaining the optimal solution, organize the operation task types, operation paths, energy consumption plans, and resource allocation schemes of each piece of equipment during the corresponding time period into a scheduling instruction and send it to the equipment terminal.

[0056] As a preferred embodiment of the intelligent operation task scheduling system for agricultural equipment described in the present invention, it includes: one or more processors;

[0057] A memory that stores instructions that can be operated, and when the instructions are executed by the one or more processors, the one or more processors are caused to perform operations, and the operations include the processes of the intelligent operation task scheduling method for agricultural equipment as described above.

[0058] As a preferred embodiment of a computer-readable medium storing software described in the present invention, the software includes instructions that can be executed by one or more computers, and when the instructions are executed in this way, the one or more computers are caused to perform operations, and the operations include the processes of the intelligent operation task scheduling method for agricultural equipment as described above.

[0059] Advantages of the present invention:

[0060] 1. By establishing a multi-objective scheduling function, the specific indicators and optimization directions of scheduling optimization are clarified, so that the scheduling model better fits the actual needs of agricultural production, provides a clear quantitative basis for subsequent optimization calculations, avoids local optimization problems caused by single-objective optimization, and takes into account cost control, time management, and effective resource allocation under multiple constraint conditions;

[0061] 2. Using deep learning algorithms to real-time mine and analyze the internal laws between environmental characteristics and equipment state changes in agricultural production, providing accurate and reliable prediction information, avoiding the risks of information lag or misjudgment caused by traditional experience-based predictions, and improving the rationality of agricultural equipment task allocation and the accuracy of operation predictions;

[0062] 3. By monitoring and identifying unexpected situations or production progress deviations that occur in the actual agricultural production process in real time, and dynamically adjusting the parameters of the scheduling model in a timely manner, the scheduling decision of agricultural equipment can quickly adapt to the changing environment and requirements, thereby avoiding problems such as resource waste, production plan delays, or execution errors caused by the inability to adjust the operation plan in a timely manner, ensuring the effective implementation of the production plan, and improving the anti-interference ability of agricultural production;

[0063] 4. By solving the optimal solution of the multi-objective scheduling function through the updated dynamic collaborative optimization model, the generation of globally optimal agricultural equipment operation task scheduling instructions is realized. Using the accurate prediction and dynamic model parameter update results obtained in the previous steps, it effectively solves the problems of task arrangement conflicts, improper use of equipment resources, and chaotic operation order that may occur in actual production, thereby providing a scheduling decision-making plan with both a global perspective and detailed accuracy, improving the overall efficiency of the agricultural production process, reducing operation energy consumption and costs, optimizing resource allocation, and enhancing the executability and reliability of the operation plan. Brief Description of the Drawings

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. Among them:

[0065] Figure 1 It is a schematic flow chart of the intelligent operation task scheduling method for agricultural equipment shown in the present invention. Detailed Embodiments

[0066] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments.

[0067] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0068] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0069] According to the embodiments of the present invention, in combination with Figure 1The flow chart shown, an intelligent operation task scheduling method for agricultural equipment, specifically includes the following steps:

[0070] S1. Establish a multi-objective scheduling function according to the agricultural equipment operation task type, equipment energy consumption cost, path planning cost, operation time constraint, and resource utilization rate. Among them, it should be noted in this step that:

[0071] The agricultural equipment operation task type includes at least tilling, sowing, fertilizing, spraying pesticides, harvesting, and transportation;

[0072] The equipment energy consumption cost includes, define E i as the average energy consumption per unit time of the i-th equipment, t i as the duration of this equipment participating in this round of operation, p i as the energy consumption price or conversion factor, then:

[0073]

[0074] The path planning cost includes, define d i as the total distance traveled by the i-th equipment on the operation path, β as the path cost per unit distance, then:

[0075]

[0076] The operation time constraint includes, if t max is the maximum allowable operation duration in this stage, t use is the total time used for all allocated operations, define the operation time penalty term as:

[0077]

[0078] Among them, γ is the additional penalty coefficient after exceeding the specified time;

[0079] The resource utilization rate includes, regarding the resource utilization rate as a positive optimization goal, define the resource utilization rate as U(x), and let R(x) = -U(x), where U(x) is expressed as:

[0080]

[0081] When U(x) is higher, it means the resource utilization situation is better, and the corresponding R(x) value is lower, that is, the negative utility is smaller;

[0082] As an example, let the decision variable set be x, and define the multi-objective scheduling function as:

[0083] F(x) = w1C e (x) + w2C p (x) + w3T(x) + w4R(x)

[0084] Among them, C e (x) is the equipment energy consumption cost, C p (x) is the path planning cost, T(x) is the objective function value of the operation time constraint, R(x) is the negative utility function value of the resource utilization rate, w1, w2, w3, w4 are the weight coefficients of the equipment energy consumption cost, path planning cost, operation time constraint, and resource utilization rate respectively, and x represents the operation arrangement, path selection, and resource allocation scheme of multiple agricultural equipment in different time periods.

[0085] Preferably, in this embodiment, by establishing a multi-objective scheduling function, the specific indicators and optimization directions of scheduling optimization are clarified, so that the scheduling model better fits the actual needs of agricultural production, provides a clear quantitative basis for subsequent optimization calculations, avoids local optimization problems caused by single-objective optimization, and takes into account cost control, time management, and effective resource allocation under multiple constraint conditions.

[0086] S2. Input the real-time environmental data and agricultural equipment status data into the deep neural network for learning and training to obtain an agricultural equipment dynamic collaborative optimization model that outputs the predicted task demand value. It should be noted in this step that:

[0087] The real-time environmental data at least includes meteorological elements, soil and plot information, and pest and disaster monitoring information;

[0088] Exemplarily, the meteorological elements include temperature, humidity, wind speed, precipitation, light intensity, and weather prediction information, the soil and plot information includes soil temperature, soil humidity, soil nutrient content, farmland plot boundary, and terrain change data, and the pest and disaster monitoring information includes the distribution area of pest occurrence, pest density, and disaster warning information;

[0089] The agricultural equipment status data at least includes position and movement information, energy / fuel information, mechanical health, and operation load information;

[0090] Exemplarily, the position and movement information includes GPS coordinates, current operation speed, and moving direction, the energy / fuel information includes fuel reserve, remaining battery power, and fuel / battery consumption rate, the mechanical health includes component operating temperature, vibration amplitude, cumulative operating duration, and fault diagnosis results, and the operation load information includes the current operation type, actual working time, carried tools, and status of attached equipment and operation load rate;

[0091] In an alternative embodiment, inputting the real-time environmental data and agricultural equipment status data into the deep neural network for learning and training to obtain an agricultural equipment dynamic collaborative optimization model that outputs the predicted task demand value includes:

[0092] Clean and denoise the real-time collected environmental data and agricultural equipment status data, and batch-segment them according to timestamps;

[0093] Perform normalization processing on data of different dimensions to ensure that the magnitudes of inputs of each dimension are the same;

[0094] Let the input sequence be Q = (q1, q2, …, q T ), where q t is the feature vector at time t, including the preprocessed real-time environmental data and agricultural equipment status data;

[0095] Let the hidden layer state sequence be h t , and the output sequence be Adopt a recurrent neural network structure, and its state update and output prediction are expressed as:

[0096] h t = σ(W qh ·q t + W hh ·h t-1 + b h )

[0097]

[0098] Among them, W qh , W hh , W hy are the weight vectors from the input to the hidden layer, from the hidden layer to the hidden layer, and from the hidden layer to the output respectively, b h , b y are the bias vectors of the hidden layer and the output layer respectively, σ(·) is the activation function, h t-1 is the hidden layer state at the previous moment, is the predicted output at time t;

[0099] Define the loss function where is the model output sequence, and y is the actual value of the corresponding task requirement;

[0100] Adopt an optimization algorithm based on stochastic gradient descent to update the W qh , W hh , W hy , b h , b y parameters in reverse;

[0101] When the maximum number of training rounds is reached and the error converges, output the current recurrent neural network model, that is, the finally obtained dynamic collaborative optimization model of agricultural equipment is expressed as:

[0102]

[0103] Among them, θ represents all the model parameters obtained after training. Then it is the predicted value of the task requirement made at each moment t.

[0104] It should be noted that by inputting real-time environmental data and agricultural equipment status data into a deep neural network for learning and training, an agricultural equipment dynamic collaborative optimization model that outputs the predicted value of task requirements is constructed, realizing the accurate prediction of the agricultural operation environment and equipment operation status. In this step, deep learning algorithms are used to real-time mine and analyze the internal laws between environmental characteristics and equipment status changes in agricultural production, providing accurate and reliable prediction information, avoiding the risks of information lag or misjudgment that may be brought by traditional experience-based prediction, and improving the rationality of agricultural equipment task allocation and the accuracy of operation prediction.

[0105] S3. Dynamically update the model parameters of the dynamic collaborative optimization model according to the farm operation progress and sudden working conditions. Among them, it should be noted in this step that:

[0106] Define η as the completion ratio of the target in this stage (such as the ratio of the cumulative cultivated area completed to the cultivated area to be completed);

[0107] If |η - η arget | ≥ ω, it means that there is a large deviation between the operation progress and the target value, and the online update of model parameters is immediately triggered;

[0108] If |η - η arget | < ω, no parameter update is made in this stage;

[0109] Among them, η arget is the expected completion ratio, and ω is the set threshold;

[0110] When one of the conditions that the weather condition critical value exceeds the safety threshold or the failure rate of the key parts of the equipment exceeds the set upper limit is met, it is regarded as a sudden working condition;

[0111] By real-time monitoring of meteorological data, farmland environment data and equipment status data, the above conditions are monitored, and if any one meets the standard, it is determined that a sudden working condition has occurred;

[0112] Exemplarily, the failure rate of key parts includes but is not limited to the failure of the equipment power system and the GPS malfunction.

[0113] In an alternative embodiment, the newly collected environmental data and equipment status data are input into the trained dynamic collaborative optimization model together, and the incremental learning process is started, where:

[0114] Let θ represent the parameter vector of the model, including the weight matrix, bias term and learning rate in the neural network structure, and k represent the current iteration or update round;

[0115] When a new data set D ew (including environmental data and equipment status data) is monitored, the model parameters θ are updated online, and the update formula is:

[0116]

[0117] where θ (k) represents the parameter vector of the k-th round model, and θ (k+1) represents the parameter vector of the (k + 1)-th round model after update. α is the learning rate, represents the gradient of the loss function L calculated on the new data set D ew with respect to θ (k) ;

[0118] Define a prediction error metric (such as MSE) to measure the performance of the model on the new data set. If the prediction error of the updated model θ (k+1) is lower than that of the previous θ (k) error, it means that the updated model has higher accuracy;

[0119] If the prediction accuracy of the updated model is higher than that of the previous model, the update result is retained, and θ (k+1) is used as the current latest parameter vector;

[0120] If the prediction accuracy of the updated model is lower than that of the previous model, roll back to the original model θ (k) and record this update attempt to avoid model instability in unknown sudden working conditions or overfitting situations.

[0121] Preferably, through the above dynamic update mechanism, the model is adaptively adjusted at the first time when a sudden working condition in farmland operations occurs, ensuring that the scheduling decision matches the actual agricultural production requirements. When the meteorological conditions change suddenly or the equipment components fail, the model can quickly identify and update the parameters, avoiding the scheduling delay problem easily generated by traditional static models.

[0122] S4. Use the updated dynamic collaborative optimization model to solve the optimal solution of the multi-objective scheduling function and generate a global optimal job task scheduling instruction. Among them, it should be noted in this step that:

[0123] The updated dynamic collaborative optimization model from step S3, containing the latest parameter θ (k) , can accurately predict energy consumption, job load, and resource utilization rate during the evaluation of each candidate solution, and thus provide a more accurate calculation basis for the objective function F(x);

[0124] Define the number of ants as M and the maximum number of iterations as G max, discretize the agricultural equipment operation scenarios to be searched, and initialize the pheromone matrix τ on this basis;

[0125] In the g-th generation of iteration, each ant gradually selects the next "agricultural equipment - task - time period" combination from the starting point (i.e., the initial scheduling state) to form a complete operation scheduling plan;

[0126] When evaluating each step of the "equipment - task" decision, call the updated dynamic collaborative optimization model to predict the current energy consumption of the equipment, the possible operation duration, and the deviation of resource utilization rate, so as to timely correct the weights of the candidate paths;

[0127] Conduct a feasibility check on the scheduling plan x constructed by each ant, such as determining whether it violates the time window, field access restrictions, and equipment load limit conditions. If the check fails, re-plan the path;

[0128] Calculate the multi-objective scheduling function F(x) for the feasible solution:

[0129] F(x) = w1C e (x) + w2C p (x) + w3T(x) + w4R(x)

[0130] When calling the dynamic collaborative optimization model to calculate the above parts, obtain an evaluation result that is more in line with the current real-time data and equipment status, and ensure the accuracy of F(x);

[0131] Let F best represent the objective function value of the global optimal solution. If the current ant constructs a new solution x new that satisfies F(x new ) < F best , then update F best ←F(x new ), and save the corresponding solution x best ←x new ;

[0132] After completing the construction and evaluation of each generation of ants, enhance the pheromone for the paths of high-quality solutions and volatilize the pheromone for the paths of low-quality solutions to guide the ants in the next iteration to be more inclined to find high-quality scheduling plans;

[0133] If the current iteration number g has reached the maximum iteration number G max , or no new solution better than F best has been generated for multiple consecutive generations, stop the iteration; otherwise, continue to the next generation of iteration;

[0134] The x best at the termination moment is the global optimal operation task scheduling plan, and its corresponding objective function value F best is the best;

[0135] As an example, after obtaining the global optimal solution x best , according to the resource and time sequence allocation information of "equipment - task - time period" in the solution, the corresponding instruction set is sorted out and generated; for example:

[0136] The allocation of operation task types includes tillage, sowing, spraying, and transportation;

[0137] The operation path planning includes driving routes, approach and departure paths;

[0138] The arrangement of operation time and sequence includes the starting operation time, estimated completion time, and workable time limit;

[0139] The energy consumption plan and resource allocation include fuel supply plan, battery replacement period, and load limit;

[0140] The above scheduling instructions are sent to each piece of equipment through the agricultural equipment control terminal. The instructions are sent using a wireless communication protocol to ensure that the equipment accurately executes the assigned tasks and paths during operation;

[0141] During the operation process, new environmental data and equipment status data are collected in real time and are linked with step S3: when a sudden working condition occurs, the model parameters are updated online again and the ant colony algorithm is called again for optimization, so that the overall scheduling has continuous and dynamic adaptability.

[0142] Exemplarily, if three types of tasks, namely plowing, fertilizing, and transportation, need to be arranged simultaneously in a farmland during a certain period, the equipment includes 3 plowing machines, 2 fertilizing machines, and 1 transport vehicle, and the plot complexity is relatively high, then the ants gradually generate the scheduling sequence of "equipment - task - time period" during the solution, and the updated dynamic collaborative optimization model accurately evaluates its energy consumption and time cost. Finally, the scheduling plan with the highest overall resource utilization rate and the smallest time limit penalty is selected and sent to the equipment control terminal in the form of "machine tool number - operation task - starting time - path ID - energy consumption plan" to ensure that multiple machine tools do not interfere with each other and achieve efficient allocation of time and resources.

[0143] The aforementioned preprocessing of real-time environmental data and agricultural equipment status data and the learning and training method of the recurrent neural network can be carried out by means and methods in the prior art, and will not be elaborated herein.

[0144] Under the application of the above embodiments, some other aspects disclosed in the embodiments of the present invention also propose an intelligent operation task scheduling system for agricultural equipment, including: one or more processors and a memory.

[0145] The memory is used to store instructions that can be operated, and when these instructions are executed by the one or more processors, they cause the one or more processors to perform operations, including the processes of the intelligent operation task scheduling method for agricultural equipment in the foregoing embodiments, especially Figure 1 the process of the method shown.

[0146] Some other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium storing software, and the software includes instructions that can be executed by one or more computers. When these instructions are executed in such a way, they cause the one or more computers to perform operations, including the processes of the intelligent operation task scheduling method for agricultural equipment in the foregoing embodiments, especially Figure 1 the process of the method shown.

[0147] It should be recognized that the embodiments of the present invention can be implemented or carried out by computer hardware, a combination of hardware and software, or computer instructions stored in a non-transitory computer-readable memory.

[0148] The method can be implemented in a computer program using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program, wherein the storage medium so configured causes the computer to operate in a specific and predefined manner.

[0149] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with a computer system. However, if necessary, the program can be implemented in assembly or machine language.

[0150] In any case, the language can be a compiled or interpreted language.

[0151] In addition, for this purpose, the program can run on a dedicated integrated circuit programmed for this purpose.

[0152] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) executed commonly on one or more processors, by hardware, or a combination thereof. The computer program includes multiple instructions executable by one or more processors.

[0153] Furthermore, the method can be implemented in any type of computing platform operably connected, including but not limited to personal computers, minicomputers, mainframes, workstations, network or distributed computing environments, separate or integrated computer platforms, or communicating with charged particle tools or other imaging devices.

[0154] Aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, an optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer and can be used to configure and operate the computer to perform the processes described herein when the storage medium or device is read by the computer.

[0155] In addition, the machine-readable code, or portions thereof, can be transmitted via a wired or wireless network.

[0156] The present invention includes these and other different types of non-transitory computer-readable storage media when such media include instructions or programs that, when combined with a microprocessor or other data processor, implement the steps described above.

[0157] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An intelligent operation task scheduling method for agricultural equipment, characterized in that, Including: Establish a multi-objective scheduling function according to the agricultural equipment operation task type, equipment energy consumption cost, path planning cost, operation time constraint, and resource utilization rate; Input the real-time environmental data and agricultural equipment status data into a deep neural network for learning and training to obtain an agricultural equipment dynamic collaborative optimization model that outputs the predicted value of the task demand; Dynamically update the model parameters of the dynamic collaborative optimization model according to the farm operation progress and sudden working conditions; Use the updated dynamic collaborative optimization model to solve the optimal solution of the multi-objective scheduling function and generate a global optimal operation task scheduling instruction.

2. The intelligent operation task scheduling method for agricultural equipment according to claim 1, wherein, The establishment of the multi-objective scheduling function includes: Let the decision variable set be x, and define the multi-objective scheduling function as: F(x) = w1C e (x) + w2C p (x) + w3T(x) + w4R(x) Among them, C e (x) is the equipment energy consumption cost, C p (x) is the path planning cost, T(x) is the objective function value of the operation time constraint, R(x) is the negative utility function value of the resource utilization rate, w1, w2, w3, and w4 are the weight coefficients of the equipment energy consumption cost, path planning cost, operation time constraint, and resource utilization rate respectively, and x represents the operation arrangement, path selection, and resource allocation plan of multiple agricultural equipment in different time periods.

3. The intelligent operation task scheduling method for agricultural equipment according to claim 2, wherein The agricultural equipment operation task type at least includes tilling, sowing, fertilizing, spraying pesticides, harvesting, and transporting; The equipment energy consumption cost includes, defining E i as the average energy consumption of the i-th equipment per unit time, t i as the duration of this equipment's participation in this round of operations, p i as the energy consumption price or conversion factor, then: The path planning cost includes defining d i as the total distance traveled by the i-th piece of equipment on the operation path, and β as the path cost per unit distance, then: The operation time constraint includes that if t max is the maximum allowable operation duration for this stage, and t use is the total time used for all assigned operations, the operation time penalty term is defined as: Where γ is a penalty coefficient that is additionally increased after exceeding the specified time; The resource utilization rate includes regarding the resource utilization rate as a positive optimization objective, defining the resource utilization rate as U(x), and letting R(x)=-U(x), where U(x) is expressed as: When U(x) is higher, it means that the resource utilization situation is better, and the corresponding value of R(x) is lower, that is, the negative utility is smaller.

4. The intelligent operation task scheduling method for agricultural equipment according to claim 1, wherein The agricultural equipment dynamic collaborative optimization model that inputs the real-time environmental data and agricultural equipment status data into a deep neural network for learning and training to obtain the predicted value of the output task demand includes: Clean and denoise the real-time collected environmental data and agricultural equipment status data, and perform batch segmentation according to the time stamp; Perform normalization processing on data of different dimensions to ensure that the magnitudes of inputs of each dimension are the same; Let the input sequence be Q = (q1, q2, …, q T ), where q t is the feature vector at time t, including preprocessed real-time environmental data and agricultural equipment status data; Let the hidden layer state sequence be h t , and the output sequence be Adopt a recurrent neural network structure, and its state update and output prediction are expressed as: h t = σ(W qh ·q t + W hh ·h t-1 + b h ) Among them, W qh , W hh , W hy are the weight vectors input to the hidden layer, from the hidden layer to the hidden layer, and from the hidden layer to the output respectively. b h , b y are the bias vectors of the hidden layer and the output layer respectively. σ(·) is the activation function, and h t-1 is the state of the hidden layer at the previous moment, and is the predicted output at time t; Define the loss function where is the model output sequence, and y is the actual value of the corresponding task requirement; Update W backward using an optimization algorithm based on stochastic gradient descent qh ,W hh ,W hy ,b h ,b y parameters; When the maximum number of training rounds is reached and the error converges, that is, output the current recurrent neural network model, that is, the finally obtained agricultural equipment dynamic collaborative optimization model is expressed as: where θ represents all the model parameters obtained after training, which is the predicted value of the task requirement at each time t.

5. The intelligent operation task scheduling method for agricultural equipment according to claim 1 or 4, characterized in that The real-time environmental data at least includes meteorological elements, soil and plot information, and pest and disaster monitoring information; Among them, the meteorological elements include temperature, humidity, wind speed, precipitation, light intensity, and weather forecast information, the soil and plot information includes soil temperature, soil humidity, soil nutrient content, farmland plot boundary, and terrain change data, and the pest and disaster monitoring information includes the distribution area of pest occurrence, pest density, and disaster warning information; The agricultural equipment status data at least includes position and movement information, energy / fuel information, mechanical health, and operation load information; Among them, the position and movement information includes GPS coordinates, current operation speed, and moving direction, the energy / fuel information includes fuel remaining, battery remaining power, and fuel / battery consumption rate, the mechanical health includes component operation temperature, vibration amplitude, cumulative operation duration, and fault diagnosis results, and the operation load information includes the current operation type, actual working time, tools and auxiliary equipment status carried, and operation load rate.

6. The intelligent operation task scheduling method for agricultural equipment according to claim 4, characterized in that, Dynamically update the model parameters of the dynamic collaborative optimization model according to the farm operation progress and sudden working conditions, including: Define η as the completion ratio of the target in this stage. If |η - η arget | ≥ ω, it indicates that there is a large deviation between the operation progress and the target value, and the online update of the model parameters is immediately triggered; where η arget is the expected completion ratio, and ω is the set threshold value; When either the condition that the weather condition critical value exceeds the safety threshold or the failure rate of the key parts of the equipment exceeds the set upper limit is met, it is regarded as a sudden working condition; Adopt an incremental learning strategy, input the newly collected environmental data, equipment status data, and progress deviation information into the dynamic collaborative optimization model, and adaptively adjust the network weights and learning rates; If the prediction accuracy of the updated model is better than that of the model before the update, retain the update result; otherwise, roll back to the original model and record relevant information.

7. The intelligent operation task scheduling method for agricultural equipment according to claim 6, wherein, The failure rates of the key parts include, but are not limited to, equipment power system failures and GPS malfunctions.

8. The intelligent operation task scheduling method for agricultural equipment according to claim 6, characterized in that Use the updated dynamic collaborative optimization model to solve the optimal solution of the multi-objective scheduling function and generate a global optimal job task scheduling instruction, including: Iteratively solve the following objectives through the ant colony algorithm: F(x) = w1C e (x) + w2C p (x) + w3T(x) + w4R(x) Determine whether the objective function value of the solution obtained in the current iteration is better than the historical optimal value F best , if F(x) < F best , then update F best and record the corresponding solution; If F(x) ≥ F best , continue the iteration until the maximum number of iterations is reached; After obtaining the optimal solution, organize the job task types, job paths, energy consumption plans, and resource allocation schemes of each equipment within the corresponding time period into a scheduling instruction and send it to the equipment terminal.

9. An intelligent operation task scheduling system for agricultural equipment, characterized in that, Including: One or more processors; A memory storing executable instructions, which when executed by the one or more processors cause the one or more processors to perform operations, and the operations include the processes of the intelligent operation task scheduling method for agricultural equipment as described in any one of claims 1 to 8.

10. A computer-readable medium storing software, characterized in that: The software includes instructions executable by one or more computers, and the execution of such instructions causes the one or more computers to perform operations, and the operations include the processes of the intelligent operation task scheduling method for agricultural equipment as described in any one of claims 1 to 8.

Images