Method and system for constructing plant nutrition management model based on causal algorithm

Abstract

The invention provides a causal algorithm-based plant nutrition management model construction method and system, and the method comprises the steps: collecting the historical planting data of a target plant, and determining and screening a plurality of key factors in plant nutrition management; constructing a causal relationship set among the plurality of key factors based on a causal relationship model; and training and correcting the causal relationship model by using the historical planting data to predict the nutritional status of the target plant under different conditions. According to the method, the model is constructed based on the causal relationship model, so that the optimal proportion and time scheme of plant nutrition is calculated, the fertilization guidance suggestions are generated, a user is helped to avoid the phenomenon that the model generation content does not conform to reality in the planting research process, the model illusion is reduced, and meanwhile the generalization ability of the model in a planting scene is improved. The method can be applied to planting industry model training, improves the accuracy of fertilization guidance, and can be used for agricultural machinery design guidance.

Description

Technical Field

[0001] The present invention belongs to the technical field of plant nutrition management, and particularly relates to a method and system for constructing a plant nutrition management model based on a causal algorithm. Background Art

[0002] With the growth of the global population and the increase in food demand, agricultural production faces huge challenges. In order to improve the efficiency and sustainability of agricultural production, precision agriculture technology has received extensive attention and application. Precision agriculture realizes the refined management of farmland, crops, and environmental conditions by using advanced information technology and data analysis methods, thereby improving the utilization rate of special fertilizers, scientifically reducing the application of chemical fertilizers, and increasing crop yield and quality.

[0003] In modern agriculture, precise plant nutrition management is the key to improving crop yield and quality. Traditional fertilization methods often rely on experience or simple soil tests, ignoring the influence of environmental factors and plant growth stages on nutrient requirements. This may lead to low fertilizer utilization rate, increased production costs, and excessive application that damages the soil environment. Therefore, it is particularly important to develop a management model that can accurately predict the nutrient requirements of plants under different conditions. Summary of the Invention

[0004] To alleviate the above problems, the present application provides a method for constructing a plant nutrition management model based on a causal algorithm, including: Collect historical planting data of the target plant, and determine and screen multiple key factors in plant nutrition management; Based on the causal relationship model, construct a set of causal relationships between the multiple key factors; Use the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions.

[0005] Optionally, in the process of collecting historical planting data of the target plant and determining and screening multiple key factors in plant nutrition management, collect the nutritional status factors of the target plant at multiple growth stages and the environmental factors during the growth process of the target plant, and the environmental factors include soil data, water supply data, light data, and temperature data.

[0006] Optionally, in the process of constructing a set of causal relationships between the multiple key factors based on the causal relationship model, it includes: Take the plant variety of the target plant as a node, extract the causal effects of the nutritional status factors and environmental factors at the multiple growth stages on the absorption efficiency of different nutrients by the plant, and construct a Bayesian network model; Determine the conditional probability distribution between multiple nodes through the historical planting data to predict the nutritional status of the target plant under different conditions, where the historical planting data includes experimental data and planting observation data.

[0007] Optionally, taking the plant variety of the target plant as a node, extract the causal effects of the nutritional status factors and environmental factors in the multiple growth stages on the absorption efficiency of different nutrients by the plant. The process of constructing the Bayesian network model includes: Collect detailed data of the target plant at each growth stage for analysis to identify the key factors affecting plant nutrient absorption, select variables that have a significant impact on the nutrient absorption efficiency of the target plant, define the nodes in the network and the latent variables of each node, and determine the causal relationship between the nodes based on the professional knowledge database to construct a causal graph; According to the determined causal relationship, construct the structure of the Bayesian network, where the edges in the network represent the causal relationship between the latent variables of the nodes. Optionally, the process of determining the conditional probability distribution between multiple nodes through the historical planting data to predict the nutritional status of the target plant under different conditions, where the historical planting data includes experimental data and planting observation data, includes: Use the maximum likelihood estimation technique or the Bayesian estimation technique to calculate the conditional probability distribution of each node given its parent nodes, and comprehensively use the mean square error and the coefficient of determination as evaluation indicators to verify the prediction performance of the model.

[0008] Optionally, in the process of using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions, continuously verify and calibrate the Bayesian network model according to the planting observation data, so that the Bayesian network model can identify the nutritional status of the target plant based on the causal relationship set.

[0009] Optionally, the steps after using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions include: Calculate the optimal ratio scheme of plant nutrition to generate guiding suggestions for precise fertilization.

[0010] Optionally, the process of calculating the optimal ratio scheme of plant nutrition to generate guiding suggestions for precise fertilization includes: Use the yield of the target plant as the value of the objective function, define the objective function to quantify the growth performance of the target plant, and determine the decision variables according to the yield of the target plant to determine the parameters to be optimized; Establish the constraint conditions of the objective function according to the resource limitation factors and environmental factors of the target plant, and apply the nonlinear programming algorithm to the objective function and the corresponding constraint conditions to calculate the optimal solution and determine the optimal nutrient ratio scheme for the plant.

[0011] Optionally, the steps before applying the nonlinear programming algorithm to the objective function and the corresponding constraint conditions to calculate the optimal solution include: Taking the yield or fertilizer application amount of the target plant as the value of the objective function, and taking the fertilizer application amounts and application times of more than a dozen preset element types such as nitrogen, phosphorus, and potassium as decision variables, establish a nonlinear function about the fertilizer application amount over time to represent the yield or fertilizer application amount of the target plant. Based on the growth and development characteristics of the target plant and the conditions for applying special plant fertilizers (implementing ground fertilizers and foliar fertilizers) under preset environmental conditions such as soil fertilizer carrying capacity, water source supply, light, and temperature, apply the sequential quadratic programming algorithm or the interior point method to the objective function to calculate the optimal solution of the objective function and determine the fertilizer application amount ratio scheme of nitrogen, phosphorus, and potassium.

[0012] The present application also provides a construction system for a plant nutrition management model based on a causal algorithm, including: A collection and screening module capable of collecting the historical planting data of the target plant and determining and screening multiple key factors in plant nutrition management; A causal relationship construction module capable of constructing a set of causal relationships between the multiple key factors based on a causal relationship model; A calculation module capable of using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions.

[0013] The construction method and system of the plant nutrition management model based on the causal algorithm of the present application are based on the growth and development laws of plants and environmental variables, and can collect the historical planting data of the target plant, determine and screen multiple key factors in plant nutrition management; construct a set of causal relationships between the multiple key factors based on a causal relationship model; use the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions. The present application determines and screens the key factors in plant nutrition management by collecting the historical planting data of the target plant, and then constructs a set of causal relationships between these factors based on a causal relationship model. Use the historical planting data to train and correct the model to predict the nutritional status of the target plant under different conditions, calculate the optimal ratio and time plan of plant nutrition, generate fertilization guidance suggestions, help users reduce the phenomenon of inconsistent model-generated content with reality during the planting research process, reduce "model hallucinations", and improve the generalization ability of the model in planting scenarios. It can be applied to the training of planting models, improve the accuracy of fertilization guidance, and can be used for agricultural machinery design guidance. Brief Description of the Drawings

[0014] The drawings herein are incorporated into and form a part of this specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0015] Figure 1 It is a flowchart of a method for constructing a plant nutrition management model based on a causal algorithm of the present application.

[0016] The implementation, functional features, and advantages of the objectives of the present application will be further described with reference to the embodiments and the accompanying drawings. Through the above-mentioned drawings, specific embodiments of the present application have been shown, and more detailed descriptions will be provided hereinafter. These drawings and written descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. Detailed Description of the Embodiments

[0017] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0018] It should be noted that in this document, the terms "comprising", "including" or any other variation thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or device comprising that element. In addition, components, features, and elements with the same name in different embodiments of the present application may have the same meaning or different meanings, and their specific meanings need to be determined based on their explanations in the specific embodiments or further in combination with the context of the specific embodiments.

[0019] It should be understood that although the terms first, second, third, etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this document, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining". Furthermore, as used in this document, the singular forms "a", "an", and "the" are also intended to include the plural forms unless the context indicates otherwise. It should be further understood that the terms "comprising", "including" indicate the presence of the stated features, steps, operations, elements, components, items, kinds, and / or groups, but do not exclude the presence, occurrence, or addition of one or more other features, steps, operations, elements, components, items, kinds, and / or groups. The terms "or", "and / or", "including at least one of the following", etc. used in this application may be interpreted inclusively, or mean any one or any combination. For example, "including at least one of the following: A, B, C" means "any one of the following: A; B; C; A and B; A and C; B and C; A and B and C", and again, "A, B or C" or "A, B and / or C" means "any one of the following: A; B; C; A and B; A and C; B and C; A and B and C". An exception to this definition only occurs when the combination of elements, functions, steps, or operations is inherently mutually exclusive in some way.

[0020] It should be understood that although the steps in the flowcharts in the embodiments of this application are shown sequentially according to the indication of the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless there is a clear indication in this document, the execution of these steps has no strict order limitation, and they can be executed in other orders. Moreover, at least a part of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0021] Depending on the context, the words "if", "when" as used herein may be interpreted as "when" or "while" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "if determined" or "if detecting (stated condition or event)" may be interpreted as "when determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)".

[0022] It should be noted that in this text, step codes such as S10 and S20 are adopted. The purpose is to more clearly and briefly express the corresponding content, and it does not constitute a substantial limitation in terms of sequence. Those skilled in the art may execute S20 first and then S10 during specific implementation, etc., but all of these should be within the protection scope of this application.

[0023] It should be understood that the specific embodiments described herein are merely used to explain this application and are not used to limit this application. Embodiment

[0024] This application first provides a method for constructing a plant nutrition management model based on a causal algorithm. Figure 1 It is a flowchart of a method for constructing a plant nutrition management model based on a causal algorithm according to this application.

[0025] As Figure 1 shown, in one embodiment, a method for constructing a plant nutrition management model based on a causal algorithm includes: S10: Collect historical planting data of the target plant and determine and screen multiple key factors in plant nutrition management.

[0026] Exemplarily, collect detailed data of the target plant at each growth stage, including environmental factors such as soil nutrient levels, moisture conditions, light intensity, temperature, etc., and the nutritional status of the plant, such as leaf nitrogen content, phosphorus content, etc. Analyze these data during the data analysis process to identify key factors affecting plant nutrient absorption. For example, through statistical analysis, the relationship between soil nitrogen content and plant nitrogen absorption efficiency can be determined.

[0027] S20: Based on the causal relationship model, construct a set of causal relationships between the multiple key factors.

[0028] Exemplarily, variables that have a significant impact on plant nutrient absorption efficiency can be selected, such as soil type, soil nutrient content, environmental temperature, light intensity, etc. A causal diagram can be constructed according to professional knowledge to represent the causal relationships between these variables.

[0029] S30: Use the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions.

[0030] Exemplarily, in a Bayesian network, each variable (such as soil nutrient content, environmental temperature, plant nutrient absorption efficiency) is represented as a node. Then use the historical planting data to estimate the conditional probability distribution of each node. For example, estimate the conditional probability distribution of the plant's nitrogen absorption efficiency under specific soil nutrient content and environmental temperature conditions.

[0031] Exemplarily, "model hallucination" refers to the phenomenon where the content generated by the model does not match the reality. In plant nutrition management, it is necessary to improve the application effect of the model. Therefore, based on the causal relationship in the process of plant nutrition management, this embodiment is established to reduce model hallucination and at the same time improve the generalization ability of the model in the planting scenario.

[0032] In this embodiment, by collecting the historical planting data of the target plant, determining and screening the key factors in plant nutrition management, and then constructing a causal relationship set between these factors based on the causal relationship model. The historical planting data is used to train and correct the model to predict the nutritional status of the target plant under different conditions, and calculate the optimal ratio and time plan of plant nutrition to generate fertilization guidance suggestions to help users reduce the phenomenon of the content generated by the model not matching the reality in the process of planting research, that is, reduce "model hallucination".

[0033] Optionally, in the process of collecting the historical planting data of the target plant and determining and screening multiple key factors in plant nutrition management, the nutritional status factors of the target plant in multiple growth stages and the environmental factors during the growth process of the target plant are collected, and the environmental factors include soil data, water supply data, light data and temperature data.

[0034] Exemplarily, this embodiment first clarifies multiple key factors in the nutrition management of the target plant, including: the nutritional status of the target plant's own growth and development at different stages, and the environmental factors during the growth process of the crop (such as: soil, water supply, light and temperature, etc.).

[0035] Optionally, in the process of constructing the causal relationship set between the multiple key factors based on the causal relationship model, it includes: Taking the plant variety of the target plant as a node, extracting the causal effect or causal logic of the nutritional status factors and environmental factors in the multiple growth stages on the absorption efficiency of different nutrients by the plant, and constructing a Bayesian network model; Determining the conditional probability distribution between multiple nodes through the historical planting data to predict the nutritional status of the target plant under different conditions, where the historical planting data includes experimental data and planting observation data.

[0036] Exemplarily, through the Bayesian model, the causal relationship between multiple key factors is constructed. For example, taking the plant variety as a node, the "growth and development characteristics" and "environmental factors" causally affect the absorption efficiency of different nutrients by the plant. For example: during the emergence period of the crop, it is mainly in the vegetative growth stage (promoting germination, new shoot growth, and leaf growth), so there are certain requirements for the proportion of elements such as nitrogen, phosphorus, and potassium in the fertilizer, and the ionic form will also be different. At the same time, considering environmental factors, such as: light time, periodic temperature, all affect the absorption and utilization of nutrients by the crop.

[0037] Optionally, taking the plant variety of the target plant as a node, the process of extracting the causal effects or causal logics of the nutrient status factors and environmental factors in the multiple growth stages on the absorption efficiency of the plant for different nutrients and constructing a Bayesian network model includes: Collect detailed data of the target plant at each growth stage for analysis to identify the key factors affecting plant nutrient absorption, select variables that have a significant impact on the nutrient absorption efficiency of the target plant, define the nodes in the network and the latent variables of each node, and determine the causal relationships between the nodes based on a professional knowledge database to construct a causal graph; According to the determined causal relationships, construct the structure of the Bayesian network, where the edges in the network represent the causal relationships between the latent variables of the nodes.

[0038] Exemplarily, variables that have a significant impact on the nutrient absorption efficiency of the plant can be selected, such as soil type, soil nutrient content, environmental temperature, light intensity, etc. In the process of constructing the causal graph, construct the causal graph according to professional knowledge to represent the causal relationships between these variables. For example, the soil nutrient content may directly affect the nutrient absorption efficiency of the plant, while the environmental temperature may indirectly affect the nutrient absorption of the plant by affecting the activity of soil microorganisms. Exemplarily, in the Bayesian network, each variable (such as soil nutrient content, environmental temperature, plant nutrient absorption efficiency) is represented as a node. Then use historical planting data to estimate the conditional probability distribution of each node. For example, estimate the conditional probability distribution of the plant's nitrogen absorption efficiency under specific soil nutrient content and environmental temperature conditions.

[0039] Exemplarily, assume that a Bayesian network model for citrus needs to be constructed. The nodes may include: Citrus varieties (A, B, C); morphological and physiological and biochemical data at the growth stages (sprouting stage, flowering stage, full-bloom stage, fruit-setting stage, color-changing and ripening stage); soil type, water supply situation, light intensity, temperature.

[0040] When determining the causal relationships, it is found that citrus variety A has a higher demand for nitrogen during the growth period. The soil type affects the nutrient absorption efficiency of the plant. The light intensity and temperature affect the growth rate of the plant, thereby affecting the nutrient demand.

[0041] When constructing the Bayesian network, these nodes can be connected to represent the causal relationships between them. Then, use historical planting data to estimate the conditional probability distribution of each node. For example, estimate the probability distribution of the nitrogen demand of citrus variety A during the growth period under specific soil type and temperature conditions.

[0042] Optionally, during the process of determining the conditional probability distribution between multiple nodes based on the historical planting data to predict the nutritional status of the target plant under different conditions, where the historical planting data includes experimental data and planting observation data, the process includes: Using the maximum likelihood estimation technique or the Bayesian estimation technique, calculate the conditional probability distribution of each node given its parent nodes, and comprehensively use the mean squared error and the coefficient of determination as evaluation metrics to verify the prediction performance of the model.

[0043] Exemplarily, during the data collection process, data from experiments can be collected. For example, under different soil types, water supply conditions, light intensities, and temperature conditions, the nitrogen, phosphorus, and potassium levels of different citrus varieties at each growth stage can be collected. Observation data from actual plantings can be collected, including the growth conditions and nutrient absorption conditions of citrus under different conditions. Remove outliers and missing values through data cleaning to ensure the quality of the data. Convert the data into a form suitable for the model through data transformation methods, such as converting the soil type into a numerical code.

[0044] During the parameter estimation process, for each node, its conditional probability distribution given its parent nodes can be estimated. For example, the conditional probability distribution of the nitrogen demand of citrus variety A during the growth period given the soil type and temperature conditions can be estimated. Exemplarily, the maximum likelihood estimation (MLE) or the Bayesian estimation method can be used to estimate these parameters.

[0045] Optionally, during the process of using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions, continuously verify and calibrate the Bayesian network model according to the planting observation data, so that the Bayesian network model can identify the nutritional status of the target plant based on the causal relationship set.

[0046] Exemplarily, after accurately constructing the relationship through the causal algorithm, continuously verify and calibrate the model using the actual observation data in the historical planting data. When the prediction of the model is based on reasonable causal relationships rather than spurious associations, the model hallucination can be effectively reduced, thereby providing a more accurate decision-making basis for plant nutrition management.

[0047] Optionally, the steps after using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions include: Calculate the optimal ratio plan for plant nutrition to generate guiding suggestions for precise fertilization.

[0048] Exemplarily, when constructing a Bayesian network, the conditional probability distribution between nodes is determined through a large amount of experimental data and observational data accumulated in the past, so as to predict the nutritional status of plants under different conditions. At the same time, combined with the observational data in the actual planting process, the optimal ratio plan of plant nutrition is calculated. Thus, it is applied to the training of the planting industry model to improve the accuracy of the platform's fertilization guidance.

[0049] Optionally, in the process of calculating the optimal ratio plan of plant nutrition to generate guidance suggestions for precise fertilization, it includes: Using the yield of the target plant as the value of the objective function, defining the objective function to quantify the growth performance of the target plant, and determining decision variables according to the yield of the target plant to determine the parameters to be optimized; Establishing the constraint conditions of the objective function according to the resource limitation factors and environmental factors of the target plant, applying the nonlinear programming algorithm to the objective function and the corresponding constraint conditions to calculate the optimal solution, and determining the optimal ratio plan of the plant nutrition.

[0050] Exemplarily, an objective function is defined to quantify the growth performance of plants. For example, the yield can be used as the objective function, and the goal is to maximize the yield. To ultimately optimize the growth performance of plants, such as the maximum yield or the best quality. Then determine some parameters to be optimized as decision variables, such as the fertilization amounts of nitrogen, phosphorus, and potassium. Establish the constraint conditions of the objective function based on resource limitations and environmental factors.

[0051] Exemplarily, in addition to the nonlinear programming algorithm, optimization algorithms such as linear programming or genetic algorithms can also be used to find the optimal fertilization ratio plan. Apply these algorithms to the objective function and constraint conditions to find the optimal solution of the objective function. In actual planting, the effectiveness of the optimal ratio plan can be continuously verified to adjust the ratio plan according to the actual planting results.

[0052] Exemplarily, in the process of determining the constraint conditions, the factor of the maximum fertilizer carrying capacity in the soil can be considered, and the total annual fertilizer application per mu does not exceed a certain weight. For example, the total annual fertilizer application per mu for early-maturing citrus generally does not exceed 32 kilograms.

[0053] Finally, use the nonlinear programming algorithm to find the optimal fertilization ratio plan. The algorithm will consider the objective function and constraint conditions and give the optimal functional fertilization amount.

[0054] For example, the algorithm may find the following optimal ratio plan: Shoot-promoting fertilizer (ground flushing fertilizer): 6 kg / mu Flower-promoting fertilizer (ground flushing fertilizer): 6 kg / mu Fruit-swelling fertilizer (ground flushing fertilizer): 12 kg / mu Quality fertilizer (ground flushing fertilizer): 8 kg / mu Then, the effectiveness of this ratio plan can be verified in actual cultivation. If the actual yield is consistent with the predicted yield, then this ratio plan can be considered effective. Otherwise, the ratio plan can be adjusted according to the actual cultivation results. In this way, fertilization guidance suggestions can be generated for plants to optimize their growth performance.

[0055] Optionally, before the step of applying the nonlinear programming algorithm to the objective function and the corresponding constraint conditions to calculate the optimal solution, it includes: Taking the yield, fertilizer application amount or cost of the target plant as the value of the objective function, and taking the fertilization amount and fertilization time of at least one preset element type such as nitrogen, phosphorus, and potassium as decision variables, a nonlinear function about the fertilization amount of at least one element such as nitrogen, phosphorus, and potassium over time is established to represent the yield, fertilizer application amount or cost of the target plant; Based on the resource constraint conditions of fertilizer budget limitation and soil nutrient capacity, and the environmental factor constraint condition of nitrogen fertilizer usage amount, the sequential quadratic programming algorithm or the interior point method is applied to the objective function to calculate the optimal solution of the objective function to determine the fertilization amount ratio plan of nitrogen, phosphorus, and potassium; Or, Based on the growth and development characteristics of the target plant and the conditions of applying special fertilizers for plants (implementing ground fertilizers and foliar fertilizers) under preset environmental conditions such as soil fertilizer carrying capacity, water source supply, light, and temperature, the sequential quadratic programming algorithm or the interior point method is applied to the objective function to calculate the optimal solution of the objective function to determine the fertilization amount ratio plan of nitrogen, phosphorus, and potassium.

[0056] Exemplarily, in the process of maximizing citrus yield or minimizing fertilizer application amount, the objective function can be a nonlinear function about the fertilization amounts of nitrogen, phosphorus, and potassium, representing citrus yield or fertilizer application amount. When determining the decision variables, the fertilization amounts of various elements such as nitrogen, phosphorus, and potassium (in kg / mu) can be considered.

[0057] Resource limitations include fertilizer budget limitations, soil nutrient capacity, etc. In addition, minimizing nitrogen fertilizer usage to reduce greenhouse gas emissions is used as an environmental factor constraint condition.

[0058] Exemplarily, the sequential quadratic programming algorithm (SQP) or the interior point method is applied to the objective function and the constraint conditions to find the optimal solution.

[0059] Exemplarily, assume that it is necessary to calculate the optimal fertilization plan for promoting shoot growth, flower bud formation, fruit swelling, and improving quality (including elements such as nitrogen, phosphorus, and potassium) for citrus variety A. Define the following objective function and constraint conditions: Objective function: Maximize citrus yield.

[0060] Decision variables: Promote shoot growth (X 1 ), promote flower bud formation (X 2), Fruit swelling (X 3 ), Quality promotion (X 4 ). The fertilizer application rate (in kg / mu).

[0061] In the constraint conditions, the fertilizer budget: X 1 +X 2 +X 3 +X 4 ≤ 32 (unit: kg / mu). The sequential quadratic programming algorithm (SQP) is used to find the optimal fertilization plan. And the optimal fertilizer application rate and time are given.

[0062] For example, the algorithm may find the following optimal ratio plan: Shoot promotion: X 1 = 6 (unit: kg / mu).

[0063] Flower promotion: X 2 = 6 (unit: kg / mu).

[0064] Fruit swelling: X 3 = 12 (unit: kg / mu).

[0065] Quality promotion: X 4 = 8 (unit: kg / mu).

[0066] In this embodiment, the data-driven causal algorithm in the field of plant nutrition management can integrate historical planting data, including soil nutrient levels, environmental factors (such as temperature, light, water), and the growth status of plants, to construct a complex causal relationship network. By analyzing these data, the algorithm can reveal the influence of nutritional status factors and environmental factors at different growth stages on the nutrient absorption efficiency of plants, so as to provide more accurate fertilization guidance.

[0067] Among them, the Bayesian network model, as a powerful probabilistic graphical model, can effectively represent the conditional dependence relationship between variables and is widely used in prediction and decision analysis. By training and optimizing the Bayesian network model, agricultural experts can predict the different nutritional demands of plant vegetative growth and reproductive growth at different soil and environmental conditions, affect the nutrient absorption efficiency of elements such as nitrogen, phosphorus, and potassium, and then formulate the optimal fertilization plan.

[0068] In addition, the introduction of the non-linear programming algorithm further improves the optimization ability of the fertilization plan. These algorithms can consider various constraint conditions, such as the factor of the maximum fertilizer carrying capacity in the soil, and the total annual fertilizer application per mu does not exceed a certain weight, etc., to ensure that the fertilization plan is both economical and environmentally friendly. Through continuous iteration and optimization, the algorithm can find the optimal solution that maximizes crop yield or minimizes fertilizer application rate.

[0069] In summary, the plant nutrition management technology based on the causal algorithm and Bayesian network model can not only improve nutrient utilization efficiency, reduce production costs, but also reduce the negative impact on the environment, providing strong technical support for the sustainable development of modern agriculture. Embodiment

[0070] The present application also provides a construction system for a plant nutrition management model based on a causal algorithm, including: A collection and screening module, capable of collecting historical planting data of a target plant and determining and screening multiple key factors in plant nutrition management; A causal relationship construction module, capable of constructing a set of causal relationships between the multiple key factors based on a causal relationship model; A calculation module, capable of using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions.

[0071] The construction method and system of the plant nutrition management model based on the causal algorithm of the present application are based on the growth and development laws of plants and environmental variables, and can collect historical planting data of the target plant, determine and screen multiple key factors in plant nutrition management; construct a set of causal relationships between the multiple key factors based on a causal relationship model; use the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions. The present application determines and screens the key factors in plant nutrition management by collecting the historical planting data of the target plant, and then constructs a set of causal relationships between these factors based on a causal relationship model. Use historical planting data to train and correct the model to predict the nutritional status of the target plant under different conditions, calculate the optimal ratio and time plan of plant nutrition, generate fertilization guidance suggestions, help users reduce the phenomenon that the content generated by the model does not match the reality in the process of planting research, reduce "model hallucination", and at the same time improve the generalization ability of the model in the planting scenario. It can be applied to the training of planting models, improve the accuracy of fertilization guidance, and can be used for the guidance of agricultural machinery design.

[0072] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention. The structures, devices, and operation methods not specifically described and explained in the present invention are implemented according to the conventional means in the art without special explanation and limitation.

Claims

1. A method for constructing a plant nutrition management model based on a causal algorithm, characterized in that: include: Collect historical planting data of target plants, identify and screen multiple key factors in plant nutrition management; Based on the causal relationship model, construct a causal relationship set between the multiple key factors; Using the historical planting data, the causal relationship model is trained and corrected to predict the nutritional status of target plants under different conditions.

2. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 1, characterized in that: In the process of collecting historical planting data of target plants and determining and screening multiple key factors in plant nutrition management, the nutritional status factors of the target plants at multiple growth stages and the environmental factors during the growth of the target plants are collected, and the environmental factors include soil data, water supply data, light data and temperature data.

3. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 2, characterized in that: The process of constructing a causal relationship set between the multiple key factors based on the causal relationship model includes: Taking the plant varieties of the target plants as nodes, extracting the causal effects of the nutritional status factors and environmental factors at the multiple growth stages on the absorption efficiency of different nutrients by the plants, and constructing a Bayesian network model; The conditional probability distribution between multiple nodes is determined by the historical planting data to predict the nutritional status of the target plant under different conditions, wherein the historical planting data includes experimental data and planting observation data.

4. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 3, characterized in that: The plant varieties of the target plants are used as nodes, and the causal effects of the nutritional status factors and environmental factors at the multiple growth stages on the absorption efficiency of different nutrients by the plants are extracted. The process of constructing the Bayesian network model includes: Collect detailed data of target plants at various growth stages for analysis to identify key factors affecting plant nutrient absorption, select variables that have a significant impact on the nutrient absorption efficiency of target plants, define nodes in the network and potential variables of each node, and determine the causal relationship between nodes based on the professional knowledge database to construct a causal graph; According to the determined causal relationships, the structure of a Bayesian network is constructed, where the edges in the network represent the causal relationships between the latent variables of the nodes.

5. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 4, characterized in that: The conditional probability distribution between multiple nodes is determined by the historical planting data to predict the nutritional status of the target plant under different conditions, wherein the historical planting data includes experimental data and planting observation data, and the process includes: The maximum likelihood estimation technique or Bayesian estimation technique is used to calculate the conditional probability distribution of each node given its parent node, and the mean square error and determination coefficient are used as evaluation indicators to verify the prediction performance of the model.

6. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 5, characterized in that: In the process of using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions, the Bayesian network model is continuously verified and calibrated according to the planting observation data so that the Bayesian network model can identify the nutritional status of the target plant based on the causal relationship set.

7. A method for constructing a plant nutrition management model based on a causal algorithm according to any one of claims 1 to 6, characterized in that: The steps after using the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions include: Calculate the optimal ratio of plant nutrients to generate guidance for precise fertilization.

8. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 7, characterized in that: Calculate the optimal ratio of plant nutrients to generate guidance for precise fertilization, including: Using the yield of the target plant as the value of the objective function, defining the objective function to quantify the growth performance of the target plant, and determining decision variables based on the yield of the target plant to determine the parameters that need to be optimized; The constraint conditions of the objective function are established according to the resource limiting factors and environmental factors of the target plant, and a nonlinear programming algorithm is applied to the objective function and the corresponding constraint conditions to calculate the optimal solution and determine the optimal ratio scheme of the plant nutrition.

9. The method for constructing a plant nutrition management model based on a causal algorithm according to claim 8, characterized in that: The step of applying a nonlinear programming algorithm to the objective function and corresponding constraints to calculate the optimal solution includes: Taking the yield or fertilizer amount of the target plant as the value of the objective function, taking the fertilizer amount and fertilizer application time of the preset element type as the decision variables, a nonlinear function of fertilizer application time is established to represent the yield or fertilizer application amount of the target plant; Based on the growth and development characteristics of the target plants under preset environmental conditions and the application conditions of plant-specific fertilizers, the sequential quadratic programming algorithm or the interior point method is applied to the objective function to calculate the optimal solution of the objective function to determine the fertilizer ratio plan of nitrogen, phosphorus and potassium.

10. A system for constructing a plant nutrition management model based on a causal algorithm, characterized in that: include: The collection and screening module can collect historical planting data of target plants, identify and screen multiple key factors in plant nutrition management; A causal relationship building module, capable of building a causal relationship set between the multiple key factors based on a causal relationship model; The computing module can use the historical planting data to train and correct the causal relationship model to predict the nutritional status of the target plant under different conditions.

Images