Tobacco pest pesticide matching method and system

Abstract

The present application relates to the technical field of tobacco information processing, and particularly relates to a tobacco pest and disease pesticide matching method and system.A kind of tobacco pest and disease pesticide matching method includes: according to the pest and disease identified and the pesticide candidate generation first set, the first set includes multiple tensor sets, each tensor set includes pesticide procurement decision tensor, pesticide mixing decision tensor;Each tensor set in the first set is input into first computing device for processing, to obtain the target value set corresponding to each tensor set;According to the tensor set corresponding to the best target value set in the first set, recommend pesticide matching scheme.The present application realizes intelligent pesticide matching, and the pesticide matching scheme has higher treatment effect on pest and disease, saves pesticide matching time, and reduces cost.

Description

Technical Field

[0001] This invention relates to the field of tobacco information processing technology, specifically to a method and system for matching pesticides to tobacco pests and diseases. Background Technology

[0002] Pesticide application is a crucial aspect of tobacco cultivation, and its scientific and rational application is of great significance for pest and disease control and a bountiful tobacco harvest. Currently, guidance on pesticide application for tobacco pests and diseases relies primarily on farmers' experience and pharmacists' guidance, which is insufficient for large-scale, intelligent tobacco cultivation and lacks a systematic and comprehensive assessment of pesticide application. While there is considerable research both domestically and internationally on precision pesticide application, which aims to reduce pesticide usage as much as possible without diminishing its effectiveness, thereby reducing pesticide pollution, research mainly focuses on related equipment, systems, and platforms. However, research on pesticide matching schemes is relatively limited. Therefore, further exploration of pesticide matching schemes is of great importance. Existing research on tobacco pests and diseases focuses on identifying pests and diseases and searching for suitable pesticides, lacking recommendations for scientific and rational pesticide matching schemes, including recommendations on the types and dosages of pesticides to be purchased. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention discloses a method and system for matching pesticides to tobacco pests and diseases.

[0004] In a first aspect, embodiments of the present invention provide a method for matching pesticides to tobacco pests and diseases, the method comprising:

[0005] A first set is generated based on the identified pests and diseases and candidate pesticides. The first set includes multiple tensor sets, each of which includes a pesticide procurement decision tensor and a pesticide mixing decision tensor.

[0006] Each tensor set in the first set is input into the first computing device for processing to obtain a target value set corresponding to each tensor set. The operations performed by the first computing device on the input tensor sets include: obtaining a first target value based on the pesticide procurement decision tensor and cost information; obtaining a second target value based on the pesticide procurement decision tensor, pesticide treatment efficacy coefficient, and infection area coefficient; and obtaining the pesticide mixed spraying cost C for treating each pest and disease. i :

[0007]

[0008] Among them, C i Let M represent the cost of pesticide mixture application for treating the i-th pest or disease, M be the set of pesticide types, |M| be the pesticide type coefficient, and d be the pesticide mixture decision tensor. ijmThe pesticide candidate characterization tensor x represents whether pesticide type j and pesticide m used to treat pest i can be mixed. ij The third objective value is obtained by summing the costs of pesticide mixture application for all types of pests and diseases, indicating whether pesticide j is used to treat pests and diseases i;

[0009] Based on the tensor set corresponding to the best target value set in the first set, a pesticide matching scheme is recommended.

[0010] Preferably, the pesticide procurement decision tensor includes a first procurement decision tensor and a second procurement decision tensor; the first procurement decision tensor represents the quantity of pesticides purchased from pesticide vendors for treating pests and diseases, and the dimensions of the first procurement decision tensor include: pest and disease type dimension and pesticide type dimension; the second procurement decision tensor represents the quantity of pesticides purchased from other farmers for treating pests and diseases, and the dimensions of the second procurement decision tensor include: pest and disease type dimension, farmer dimension, and pesticide type dimension.

[0011] Preferably, generating the first set based on the identified pests and diseases and candidate pesticides includes:

[0012] For each other farmer, obtain the amount of various pesticides held to treat each type of pest and disease. Based on the amount of pesticides held and the therapeutic efficacy coefficient of the pesticides, obtain the initial therapeutic efficacy for each type of pest and disease. Generate a correction coefficient based on the pesticide mixing and spraying cost for each type of pest and disease. Use the correction coefficient to correct the initial therapeutic efficacy to obtain the corrected therapeutic efficacy for each type of pest and disease.

[0013] The degree of correlation among other farmers is calculated based on the corrective treatment efficacy of other farmers for each type of pest and disease.

[0014] The set of other farmers is classified according to the degree of association between them, resulting in different categories of other farmer subsets.

[0015] Select farmers from other farmer subsets of different categories, and generate a second procurement decision tensor based on the pesticide holdings of the selected farmers.

[0016] Preferably, the step of generating a first set based on the identified pests and diseases and candidate pesticides further includes:

[0017] Collect images of tobacco planting areas;

[0018] Pest and disease identification in images of tobacco growing areas;

[0019] Select pesticide candidates for the identified pests and diseases.

[0020] Preferably, the step of recommending a pesticide matching scheme based on the tensor set corresponding to the optimal target value set in the first set further includes:

[0021] The first set is input into the second computing device for processing: the second computing device performs crossover and mutation operations on the first set with a set probability, and after repair, obtains the second set; each tensor set in the second set is input into the first computing device, and the second result set is obtained according to the output of the first computing device; all results in the first result set and the second result set are sorted, and the first set is modified according to the sorting result;

[0022] Repeat the step of inputting the first set into the second computing device for processing until the termination condition is met.

[0023] Preferably, before the step of inputting each tensor set in the first set into the first computing device for processing and obtaining the first result set based on the output of the first computing device, the method further includes: generating a constraint set; the input of the first computing device also includes the constraint set.

[0024] Preferably, the constraint set includes:

[0025] Constraints between the pesticide mixing decision tensor and the pesticide mixing characterization tensor:

[0026]

[0027] Where, d ijm It is the pesticide mixing decision tensor, k jm It is a pesticide mixture characterization tensor, where N is the set of pest and disease types, M is the set of pesticide types, and i, j, and m are indices;

[0028] Constraints on the quantity of pesticides purchased:

[0029]

[0030]

[0031] Among them, y ij It is the pesticide procurement decision tensor, g ij h represents the area coefficient of pests and diseases i that can be treated by a unit of pesticide j. i Let a be the area coefficient of infection of pest i. ij It is an auxiliary variable;

[0032] Pesticide candidate constraints:

[0033]

[0034] Where, x ij It is the pesticide candidate characterization tensor, e ij This indicates whether pesticide j can treat pests and diseases i;

[0035] Constraints between the pesticide candidate characterization tensor and the pesticide procurement decision tensor:

[0036]

[0037] Among them, y ij It is the pesticide procurement decision tensor;

[0038] Pesticide application restrictions:

[0039]

[0040] Where N is the set of pests and diseases.

[0041] Secondly, embodiments of the present invention also provide a tobacco pest and disease pesticide matching system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a tobacco pest and disease pesticide matching method.

[0042] The technical solution provided by the embodiments of the present invention has at least the following advantages compared with the prior art:

[0043] This invention provides a method and system for matching pesticides to tobacco pests and diseases. It can perform unmanned autonomous image acquisition of tobacco pests and diseases over a large area and obtain a scientific and reasonable pesticide matching scheme for tobacco pests and diseases through multi-objective optimization reasoning calculation, including procurement structure, pesticide mixing and spraying instructions, etc., which can effectively improve the treatment effect of tobacco pests and diseases, improve pesticide spraying efficiency, and reduce costs.

[0044] Specifically, existing pesticide matching schemes only consider purchasing pesticides from pesticide retailers. Compared to existing technologies, this invention, based on sharing theory, considers purchasing pesticides from other farmers and obtains a scientifically sound pesticide matching scheme through a scientifically reasonable constraint set and solution method, achieving rational resource utilization and effectively reducing the cost of pesticide matching schemes while improving resource utilization. Furthermore, this invention considers the cost of pesticide mixing and spraying, and obtains the cost of pesticide mixing and spraying through an effective and accurate calculation method, reducing the cost of pesticide mixing and spraying while ensuring the efficacy of pest and disease treatment. Even further, this invention obtains the correlation between farmers by acquiring the treatment utility vector of the amount of pesticide held by each farmer, improving the quality of the initial tensor set, thereby improving the efficiency of obtaining a scientifically sound pesticide matching scheme. Attached Figure Description

[0045] Figure 1 This is an exemplary flowchart of a pesticide matching method for tobacco pests and diseases according to the first embodiment of the present invention. Detailed Implementation

[0046] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0047] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0048] Furthermore, to better illustrate this disclosure, numerous specific details are provided in the following detailed embodiments. Those skilled in the art should understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure. An embodiment of the present invention provides an in-vehicle image processing method.

[0049] Figure 1 A flowchart illustrating a method for matching pesticides to tobacco pests and diseases according to a first embodiment of the present invention is shown. Figure 1 As shown, the method includes steps 101, 102, and 103.

[0050] Step 101: Generate a first set based on the identified pests and diseases and candidate pesticides. The first set includes multiple tensor sets, each of which includes a procurement decision tensor and a pesticide mixing decision tensor.

[0051] Step 102: Input each tensor set in the first set into the first computing device for processing to obtain the target value set corresponding to each tensor set; The operations performed by the first computing device on the input tensor sets include: obtaining a first target value based on the procurement decision tensor and cost information; obtaining a second target value based on the procurement decision tensor, treatment utility coefficient and infection area coefficient; and calculating the pesticide mixing cost based on the pesticide mixing decision tensor, pesticide type coefficient and pesticide candidate characterization tensor, as a third target value.

[0052] Step 103: Recommend pesticide matching schemes based on the tensor set corresponding to the best target value set in the first set.

[0053] This embodiment considers the cost of pesticide mixing and spraying, and obtains the cost of pesticide mixing and spraying through an effective and accurate method, thereby reducing the cost of pesticide mixing and spraying while ensuring the efficacy of pest and disease treatment.

[0054] Optionally, based on the first embodiment, the method may further include the following step prior to step 101:

[0055] Step 100a: Collect images of the tobacco planting area.

[0056] Drones equipped with cameras are used to automatically collect aerial images of tobacco growing areas.

[0057] First, the tobacco planting area image is divided into at least one grid, with each grid representing an area that can be covered by the drone's overhead view. Then, the drone's flight path is generated based on the grid's center point, capturing one image per grid. The flight path can be generated by sequentially connecting the center points of adjacent grids, ensuring coverage of each grid. A serpentine antenna connection method can be used for this connection.

[0058] Step 100b: Use a pre-trained neural network for pest and disease identification to identify pests and diseases in images of tobacco growing areas.

[0059] Pests and diseases encompass both diseases and pests, serving as a collective term. A pre-trained tobacco pest and disease identification neural network is used to identify pests and diseases in collected images of tobacco-growing areas. Specifically, a dataset of tobacco-growing area images is first collected, and the collected image data is labeled, including the types of pests and diseases and whether any are diseased. One implementation method uses a convolutional neural network (CNN) structure, including an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The tobacco-growing area images used for training are input into the network, passing through the input layer, convolutional layer, pooling layer, fully connected layer, and output layer to obtain pest and disease prediction results. The network is trained and optimized based on the difference between the prediction results and the labeled results until convergence, using a cross-entropy loss function. Another implementation method is to build a tobacco pest and disease identification neural network based on a residual network, which introduces residual blocks and has deeper layers. After step S3, an infection area coefficient is needed for pesticide matching scheme inference. One implementation method is to count the number of tobacco planting area images infected with a certain pest or disease, and then multiply the number by the area covered by the grid to obtain the infected area of ​​that pest or disease.

[0060] Step 100c: Use the tobacco pest and disease knowledge graph to select pesticide candidates for the pests and diseases identified by the neural network.

[0061] Knowledge graphs are used to represent entities and relationships in the real world, aggregating large amounts of knowledge to achieve rapid knowledge response and reasoning. First, data is crawled from tobacco pest and disease data sources and preprocessed to obtain a plain text corpus. Entity recognition is performed on the plain text corpus, knowledge is extracted and integrated, and a knowledge representation model is constructed, ultimately resulting in a tobacco pest and disease knowledge graph. Commonly used graph databases include Neo4j. The tobacco pest and disease knowledge graph includes at least the pests and diseases controlled by different pesticides, pesticide attributes, and precautions. Pesticide attributes include the treatment area coefficient, which represents the infected area that a unit of pesticide can treat. Precautions include: which pesticides should not be mixed with it during use. Based on the pests and diseases identified by the neural network, the tobacco pest and disease knowledge graph outputs the corresponding treatment pesticide, pesticide attributes, and precautions.

[0062] Next, the tobacco pest and disease inference model can be used to match the pests and diseases identified by the neural network with the pesticide candidates from the knowledge graph and determine the dosage, thus inferring the application parameters. The specific implementation method is as follows: steps 101-103.

[0063] A method for matching pesticides to tobacco pests and diseases according to a second embodiment of the present invention includes the following steps.

[0064] Step 201: Generate a first set based on the identified pests and diseases and candidate pesticides. The first set includes multiple tensor sets, and each tensor set includes a first procurement decision tensor (y_s). ij ), Second Procurement Decision Tensor (y_f) rij ), Third Procurement Decision Tensor (y) ij ), pesticide mixing decision tensor (d ijm ), First intermediate tensor (δ_s), Second intermediate tensor (δ_f) r ).

[0065] The first set is a set of tensor sets, meaning that each element in the first set is a tensor set. Each tensor set also includes: the pesticide candidate representation tensor (x... ij Each element in the first set is of the form (y_s). ij ,y_f rij ,y_ ij ,d ijm ,δ_s,δ_f r ,x ij ).

[0066] In existing technologies, pesticide matching schemes only consider the scenario of purchasing pesticides from pesticide retailers. However, in reality, other tobacco farmers may have unused pesticides from previous harvests or large stockpiles of pesticides. Pesticides have a limited shelf life; long-term storage inevitably reduces their effectiveness and wastes resources. This invention, based on the sharing theory, considers obtaining pesticides from other tobacco farmers, which helps conserve social resources and improve social efficiency.

[0067] First Procurement Decision Tensor y_s ij This represents the quantity of pesticide type j purchased from the pesticide retailer for treating pest type i. The elements in the first procurement decision tensor should be non-negative, preferably non-negative integers. The dimensions of the first procurement decision tensor include: pest type dimension and pesticide dimension. The second procurement decision tensor y_f rij The second procurement decision tensor represents the quantity of pesticide type j purchased from other farmers r to treat corresponding pests i. The dimensions of the second procurement decision tensor include: pest type dimension, farmer dimension, and pesticide type dimension. Elements in the second procurement decision tensor should be non-negative, preferably non-negative integers. The third procurement decision tensor y_ ij The pesticide procurement decision tensor (i.e., the quantity of pesticide type j used to treat corresponding pest i) represents the total procurement quantity of each pesticide used to treat each pest. Dimensions include: pesticide type dimension and pest dimension. Elements in the third procurement decision tensor should be non-negative, preferably non-negative integers. The first intermediate tensor represents whether to purchase pesticides from a pesticide vendor. Elements in this tensor are binary, preferably 0 or 1. The second intermediate tensor represents whether to purchase pesticides from farmer r. Elements in this tensor are binary, preferably 0 or 1. The relationship between the first, second, and third procurement decision tensors is as follows:

[0068]

[0069] Where N is the set of pest and disease types, M is the set of pesticide types, R represents the set of other farmers, and r is the index of the farmer.

[0070] Pesticide candidate characterization tensor x ij This tensor represents whether pesticide j is used to treat pest i. The elements within this tensor are binary, preferably 0 or 1. The elements of each tensor set in the first generated set are randomly generated within their respective value ranges. Pesticide mixing decision tensor d ijmThis tensor characterizes whether pesticide type j used to treat pest i can be mixed with pesticide m. The elements within this tensor are binary, preferably 0 and 1. The pesticide mixing decision tensor has three dimensions: pest type dimension, first pesticide dimension, and second pesticide dimension. This tensor characterizes whether pesticides in the first pesticide dimension can be mixed with pesticides in the second pesticide dimension when applying pesticides to pests. The pesticide mixing spraying cost characterizes the lowest mixing spraying cost when treating pests; in this embodiment, it is the number of mixing sprays. For example, if five pesticides A, B, C, D, and E are used to treat pest i, and B and C can be mixed, the lowest mixing spraying cost is 4, meaning A, D, and E are sprayed separately, and B and C are sprayed together.

[0071] Step 202: Input each tensor set in the first set into the first computing device for processing, and obtain the first result set based on the output of the first computing device.

[0072] The operations performed by the first computing device on the input tensor set include: obtaining the first pesticide procurement cost based on the first procurement decision tensor and the cost coefficient; obtaining the first pesticide delivery cost based on the first intermediate tensor and the first delivery fee; obtaining the second pesticide procurement cost based on the second procurement decision tensor, the cost coefficient, and the discount coefficient; obtaining the second pesticide delivery cost based on the second intermediate tensor and the second delivery fee; obtaining a first target value based on the first pesticide procurement cost, the second pesticide procurement cost, the first pesticide delivery cost, and the second pesticide delivery cost; obtaining a second target value based on the third procurement decision tensor, the treatment utility coefficient, the treatment area coefficient, and the infection area coefficient; outputting the first target value and the second target value; calculating the pesticide mixing cost based on the pesticide mixing decision tensor, the pesticide type coefficient, and the pesticide candidate characterization tensor, and outputting it as the third target value; and outputting the first target value, the second target value, and the third target value.

[0073] The second objective value f2 is as follows:

[0074]

[0075] Among them, w ij The therapeutic efficacy coefficient of pesticide j against pests and diseases is g. ij h represents the treatment area coefficient per unit of pesticide j for pest i, i.e., how much area can be treated per unit of pesticide. i denoted as the infection area coefficient of pest i.

[0076] Based on the considerations provided by the knowledge graph, it is determined which pesticides can be mixed and sprayed, and which cannot, when treating pests and diseases. Using pesticides that can be mixed and sprayed can reduce spraying costs. Therefore, this embodiment uses a third objective value to obtain the spraying cost of pesticides. The third objective value is the sum of the mixed spraying costs for each type of pest and disease, and the mixed spraying costs for each type of pest and disease are as follows:

[0077]

[0078] Where |M| is the number of pesticide types, C i This represents the cost of mixing pesticides to treat the i-th type of pest or disease. The third objective value is obtained by summing the costs of mixing pesticides for all types of pests and diseases. The purpose of this invention is to obtain the optimal pesticide matching scheme, which has the minimum first objective value, the minimum third objective value, and the maximum second objective value.

[0079] Step 203: Input the first set into the second computing device for processing. The processing includes: the second computing device performing crossover and mutation operations on the first set with a set probability, and then repairing it to obtain the second set; inputting each tensor set in the second set into the first computing device, and obtaining the second result set according to the output of the first computing device; sorting all the results in the first result set and the second result set, and modifying the first set according to the sorting result.

[0080] The second computing device is a computing device based on a genetic algorithm. Preferably, a non-dominated sorting genetic algorithm II is used. For the crossover operation, two elements in the set are randomly selected with a certain probability, and their genetic material is exchanged, resulting in two offspring that are different from either parent, i.e., two new elements in the second set. This completes one crossover operation. Preferably, the probability of crossover is 0.7. For the mutation operation, a mutation point is selected with a certain probability for mutation. Let x... ij For example, if the selected x ij If the value is 1, the mutation is 0; if the value is 0, the mutation is 1. If the mutation point is a continuous variable, then a feasible value can be selected. Preferably, the probability of mutation is 0.7. The repair operation mainly depends on the problem and is used to ensure that the algorithm only searches in the feasible space. There are certain logical relationships between the variables. The logical relationships include: the first, second, and third procurement decision tensors should satisfy the aforementioned summation relationship, the amount of pesticides purchased should be the sum of the amount of pesticides purchased from the pesticide sales outlet and the farmers; when the element x in the pesticide candidate representation tensor ij When y is 0, the relevant y ij y_s ij y_f rij d ijm Both should be 0; y ijThe quantity should not exceed the quantity required to treat pests and diseases using pesticide j alone, etc. Logical relationships are used to constrain the inherent relationships between variables, ensuring that the search always takes place within the feasible space. When the solution is inconsistent with the actual situation, or does not meet the logical requirements that the actual situation should satisfy, a repair operation can be used for repair. For example, when x 12 When y = 0, it means that the second drug cannot treat the first disease, therefore y 12 The corresponding value must be 0, meaning no second drug was purchased to treat the first disease. Adding a repair operation can accelerate the algorithm's convergence speed and improve its robustness.

[0081] The second result set is the result set corresponding to each element in the second set. The result set corresponding to each element includes the first target value, the second target value, and the third target value.

[0082] Step 204: Repeat the step of inputting the first set into the second computing device for processing until the termination condition is met. Based on the tensor set corresponding to the best result in the first set, recommend a pesticide matching scheme.

[0083] The termination condition includes executing step S203 a set number of times. After execution, the Pareto solution set is obtained. Those skilled in the art can undoubtedly determine that the input to the second computing device in this step is the first set modified after the previous execution. The result corresponding to each element in the first set includes a first target value, a second target value, and a third target value. The first target value, the second target value, and the third target value are weighted and summed, and the summation results are sorted. The best result is selected based on the sorting. Specifically, the first target value, the second target value, and the third target value are first normalized, and then the normalized results are weighted and summed. Those skilled in the art can adjust the weights of the weighted summation according to the dimensions used in implementation and the emphasized objective. Preferably, the weights are 0.1818, 0.7273, and 0.0909, respectively. It should be noted that the optimization objective is to minimize the first target value and the third target value, and maximize the second target value. Therefore, when normalizing the second target value, the reverse process needs to be performed. For example, a common normalization process is: x - min / (max - min), where x is the value to be normalized, and max and min are the maximum and minimum values, respectively. Then, when processing the second target value, the following inverse process is needed: 1 - [(x - min) / (max - min)]. The set with the lowest summation is the optimal target value set. Based on the tensor set corresponding to the optimal result, the pesticide matching scheme can be obtained. The pesticide matching scheme includes: the quantity of a certain type of pesticide for a specific pest type to be purchased from pesticide retailers and farmers respectively (i.e., the first purchase decision tensor and the second purchase decision tensor), and how to apply the pesticide when spraying for a specific type of pest (pesticide mixing decision tensor).

[0084] This embodiment considers the cost of pesticide mixing and spraying, and obtains the cost of pesticide mixing and spraying through an effective and accurate method, thereby reducing the cost of pesticide mixing and spraying while ensuring the efficacy of pest and disease treatment.

[0085] Based on the above embodiments, optionally, generating a first set based on the identified pests and diseases and candidate pesticides includes the following steps.

[0086] Step 200a: For each other farmer, obtain the amount of various pesticides held for treating each type of pest and disease, and obtain the initial therapeutic efficacy for each type of pest and disease based on the amount of pesticides held and the therapeutic efficacy coefficient of the pesticides; generate a correction coefficient based on the pesticide mixing and spraying cost for each type of pest and disease, and use the correction coefficient to correct the initial therapeutic efficacy to obtain the corrected therapeutic efficacy for each type of pest and disease.

[0087] Specifically, the initial therapeutic efficacy is calculated as follows:

[0088]

[0089] in, This indicates the initial therapeutic efficacy of the amount of pesticide held by other farmers r against pest / disease type i. Let e ​​represent the set of pesticide types owned by farmer r. ij A value of 1 indicates that pesticide j can treat pest i, while a value of 0 indicates that it cannot treat it; w is a binary variable. ij The therapeutic efficacy coefficient of pesticide j against pests and diseases is g. ij hav_f represents the area treated by a unit of pesticide j for pest i. rj This indicates the quantity of pesticide type j held by farmer r. ij With x ij The difference lies in the fact that eij is generated based on whether pesticides can treat pests and diseases, given by the knowledge graph, while x... ij This indicates whether pesticide type j is used to treat pest type i.

[0090] Based on whether the pesticides held by farmer r for treating pest type i can be mixed and sprayed, the COST of pesticide mixing and spraying for pest type i held by farmer r is obtained. ri COST ri This does not involve decision tensors; the value can be obtained simply by checking whether the pesticides held can be mixed for application. For example, consider the cost. riThe acquisition method. If five pesticides A, B, C, D, and E are used to treat pest i, and B and C can be mixed, the minimum cost of mixed spraying is 4, that is, A, D, and E are sprayed separately, and B and C are mixed and sprayed. If B and C can be mixed, and D and E can be mixed, then the cost of mixed spraying for the specific pest is 3.

[0091] For each type of pest or disease i, obtain the COST of mixed spraying for all other farmers. ri To obtain the corrected therapeutic efficacy for each farmer for each type of pest and disease, reverse normalization is performed. The cost of pesticide mixing and spraying after reverse normalization is used as a correction coefficient and multiplied by the initial therapeutic efficacy. A common normalization method is: x - min / (max - min), where x is the value to be normalized, and max and min are the maximum and minimum values, respectively. Therefore, reverse normalization requires: 1 - [(x - min) / (max - min)].

[0092] Step 200b: Calculate the degree of correlation among other farmers based on the corrective treatment efficacy of other farmers for each type of pest and disease.

[0093] The corrected treatment utility for all pest and disease types of another farmer is represented by a vector, called the corrected treatment utility vector. Each other farmer has its own corrected treatment utility vector. The similarity between the corrected treatment utility vectors of any two other farmers is calculated as the degree of association. Optionally, cosine similarity can be used as a measure.

[0094] Step 200c: Classify the set of other farmers according to the degree of association between them to obtain different categories of other farmer subsets.

[0095] All other farmers are categorized based on the degree of correlation between each pair of farmers. Optionally, a clustering algorithm can be used for classification, including density-based clustering and k-means clustering. The implementer can choose according to the specific implementation scenario. Optionally, when using the k-means clustering method, k is selected as... After clustering, all other farmers are divided into multiple subsets of other farmers. Optionally, outliers can be ignored after clustering.

[0096] Step 200d: Select farmers from other farmer subsets of different categories, and generate a second procurement decision tensor based on the pesticide holdings of the selected farmers.

[0097] There are many methods for selecting farmers from subsets of other farmers in different categories, and the selection needs to cover the identified pests and diseases. One implementation method is to select representative farmers from subsets of other farmers in different categories. That is, to select the core points of different clusters obtained after clustering. Considering delivery costs, one implementation method is to select the farmers closest to the farmer who needs to purchase pesticides from subsets of other farmers in different categories. Then, the second purchasing decision tensor is initialized based on the selected farmers. Optionally, all pesticides related to pests and diseases held by the selected farmers can be purchased. The first purchasing decision tensor can be randomly initialized within its feasible region. Thus, the first, second, and third purchasing decision tensors are obtained.

[0098] This invention improves the quality of the initial tensor set by obtaining the therapeutic utility vector of the amount of pesticide held by each farmer to obtain the correlation between farmers, thereby improving the efficiency and quality of obtaining a scientific and reasonable pesticide matching scheme.

[0099] The third embodiment of the present invention is given below.

[0100] Step 301: Generate a first set based on the identified pests and diseases and candidate pesticides. The first set includes multiple tensor sets, each of which includes a first procurement decision tensor, a second procurement decision tensor, a third procurement decision tensor, a pesticide mixing decision tensor, a first intermediate tensor, and a second intermediate tensor.

[0101] Step 302: Generate a constraint set; the input to the first computing device also includes the constraint set. The constraint set includes: the constraint that the quantity of pesticides purchased from farmers does not exceed the quantity of pesticides they possess; the constraint that the elements in the third purchasing decision tensor are the sum of the relevant elements in the first purchasing decision tensor and the second purchasing decision tensor; the constraint that each pest or disease must be treated with at least one pesticide; the constraint that only non-incompatible pesticides can be mixed for spraying; and the constraint that only pesticides used can be mixed for spraying.

[0102] The restriction that the quantity of pesticides purchased from farmers cannot exceed the quantity of pesticides they possess is as follows:

[0103]

[0104] Among them, hav_f rj Let r be the quantity of pesticides j owned by farmer r.

[0105] The constraint that the elements in the third procurement decision tensor are the sum of relevant elements in the first and second procurement decision tensors is as follows:

[0106]

[0107] Among them, y ijFor the third procurement decision tensor, y_s ij It is the first procurement decision tensor, y_f rij It is the second procurement decision tensor.

[0108] There is a constraint that each pest or disease requires at least one pesticide treatment:

[0109]

[0110] Where, x ij Indicates whether pesticides j are used to treat pests and diseases i.

[0111] Restrictions on the mixing and spraying of pesticides:

[0112]

[0113] Where, k jm This is a pesticide mixing representation tensor, generated based on the considerations from the knowledge graph output regarding whether pesticides can be mixed. It represents whether pesticide j and pesticide m can be mixed, with elements being binary variables. Preferably, 1 represents "cannot mix," and 0 represents "can mix." The difference between the pesticide mixing representation tensor and the pesticide mixing decision representation tensor lies in whether it is used for decision-making and whether it includes the pest / disease type dimension. The pesticide mixing representation tensor is based on knowledge obtained from the knowledge graph and does not represent whether the two pesticides will ultimately be used to treat pests / diseases. The pesticide mixing decision representation tensor, on the other hand, represents whether the two pesticides will ultimately be mixed and sprayed to treat pests / diseases.

[0114] Constraints on the amount of pesticides purchased:

[0115]

[0116]

[0117] Among them, g ij h represents the area coefficient of pests and diseases i that can be treated by a unit of pesticide j. i Let be the infection area coefficient of pest i, and aij be an auxiliary variable.

[0118] Pesticide candidate constraints:

[0119]

[0120] Among them, e ij A value of 1 indicates that pesticide j can treat pest i, while a value of 0 indicates that it cannot treat the pest. This is a binary variable.

[0121] Constraints between the pesticide candidate characterization tensor and the first procurement decision tensor:

[0122]

[0123] The above constraint is used to ensure that when one is 0, the other is also 0.

[0124] Pesticide application restrictions:

[0125]

[0126] The above constraints are to ensure that pesticides have been sprayed on all infected areas.

[0127] Step 303: Input each tensor set in the first set into the first computing device for processing, and obtain the target value set based on the output of the first computing device. The operations performed by the first computing device on the input tensor sets include: obtaining the first pesticide purchase cost based on the first procurement decision tensor and the cost coefficient; obtaining the first pesticide delivery cost based on the first intermediate tensor and the first delivery cost; obtaining the second pesticide purchase cost based on the second procurement decision tensor, the cost coefficient, and the discount coefficient; obtaining the second pesticide delivery cost based on the second intermediate tensor and the second delivery cost; obtaining the first target value based on the first pesticide purchase cost, the second pesticide purchase cost, the first pesticide delivery cost, and the second pesticide delivery cost; obtaining the second target value based on the third procurement decision tensor, the treatment utility coefficient, the treatment area coefficient, and the infection area coefficient; and outputting the first target value and the second target value.

[0128] The first set is based on a mathematical optimization solver, such as the Gurobi solver. The optimization is performed within the constraint set, aiming to minimize the first objective value and maximize the second objective value.

[0129] Step 304: Recommend pesticide matching schemes based on the tensor set corresponding to the best target value set in the first set.

[0130] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for matching pesticides to control tobacco pests and diseases, characterized in that, The method specifically includes: A first set is generated based on the identified pests and diseases and candidate pesticides. The first set includes multiple tensor sets, each of which includes a pesticide procurement decision tensor and a pesticide mixing decision tensor. Each tensor set in the first set is input into the first computing device for processing to obtain a target value set corresponding to each tensor set. The operations performed by the first computing device on the input tensor sets include: obtaining a first target value based on the pesticide procurement decision tensor and cost information; obtaining a second target value based on the pesticide procurement decision tensor, pesticide treatment efficacy coefficient, and infection area coefficient; and obtaining the pesticide mixed spraying cost C for treating each pest and disease. i : Among them, C i Let M represent the cost of pesticide mixture application for treating the i-th pest or disease, M be the set of pesticide types, |M| be the pesticide type coefficient, and d be the pesticide mixture decision tensor. ijm The pesticide candidate characterization tensor x represents whether pesticide type j and pesticide m used to treat pest i can be mixed. ij The third objective value is obtained by summing the costs of pesticide mixture application for all types of pests and diseases, indicating whether pesticide j is used to treat pests and diseases i; Based on the tensor set corresponding to the best target value set in the first set, a pesticide matching scheme is recommended; The pesticide procurement decision tensor includes a first procurement decision tensor and a second procurement decision tensor. The first procurement decision tensor represents the quantity of pesticides purchased from pesticide vendors for treating pests and diseases, and its dimensions include: pest and disease type dimension and pesticide type dimension. The second procurement decision tensor represents the quantity of pesticides purchased from other farmers for treating pests and diseases, and its dimensions include: pest and disease type dimension, farmer dimension, and pesticide type dimension. The generation of the first set based on the identified pests and diseases and candidate pesticides includes: For each other farmer, obtain the amount of various pesticides held to treat each type of pest and disease. Based on the amount of pesticides held and the therapeutic efficacy coefficient of the pesticides, obtain the initial therapeutic efficacy for each type of pest and disease. Generate a correction coefficient based on the pesticide mixing and spraying cost for each type of pest and disease. Use the correction coefficient to correct the initial therapeutic efficacy to obtain the corrected therapeutic efficacy for each type of pest and disease. The degree of correlation among other farmers is calculated based on the corrective treatment efficacy of other farmers for each type of pest and disease. The set of other farmers is classified according to the degree of association between them, resulting in different categories of other farmer subsets. Select farmers from other subsets of farmers of different categories, and generate a second procurement decision tensor based on the pesticide holdings of the selected farmers; The step of recommending pesticide matching schemes based on the tensor set corresponding to the best target value set in the first set also includes: The first set is input into the second computing device for processing: the second computing device performs crossover and mutation operations on the first set with a set probability, and after repair, obtains the second set; each tensor set in the second set is input into the first computing device, and the second result set is obtained according to the output of the first computing device; all results in the first result set and the second result set are sorted, and the first set is modified according to the sorting result; Repeat the step of inputting the first set into the second computing device for processing until the termination condition is met.

2. The method for matching pesticides to tobacco pests and diseases as described in claim 1, characterized in that, The step of generating a first set based on identified pests and diseases and candidate pesticides also includes the following steps before: Collect images of tobacco planting areas; Pest and disease identification in images of tobacco growing areas; Select pesticide candidates for the identified pests and diseases.

3. The method for matching pesticides to tobacco pests and diseases as described in claim 1, characterized in that, Before the step of inputting each tensor set in the first set into the first computing device for processing and obtaining the first result set based on the output of the first computing device, the method further includes: generating a constraint set; the input of the first computing device also includes the constraint set.

4. The method for matching pesticides to tobacco pests and diseases as described in claim 3, characterized in that, The constraint set includes: Constraints between the pesticide mixing decision tensor and the pesticide mixing characterization tensor: Where, d ijm It is the pesticide mixing decision tensor, k jm It is a pesticide mixture characterization tensor, where N is the set of identified pests and diseases, and M is the set of candidate pesticides. Constraints on the quantity of pesticides purchased: Among them, y ij It is the pesticide procurement decision tensor, g ij h represents the area coefficient of pests and diseases i that can be treated by a unit of pesticide j. i Let a be the area coefficient of infection of pest i. ij It is an auxiliary variable; Pesticide candidate constraints: Where, x ij It is the pesticide candidate characterization tensor, e ij This indicates whether pesticide j can treat pests and diseases i; Constraints between the pesticide candidate characterization tensor and the pesticide procurement decision tensor: Where, x ij It is the pesticide candidate characterization tensor, y ij It is the pesticide procurement decision tensor; Pesticide application restrictions: Among them, y ij It is the pesticide procurement decision tensor, g ij The area coefficient representing the number of pests i that can be treated by a unit of pesticide j is N, where N is the set of pest types.

5. A pesticide matching system for tobacco pests and diseases, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a method for matching pesticides to tobacco pests and diseases as described in any one of claims 1-4.

Images