Credit strategy determination method and device, program product and storage medium

Abstract

The embodiment of the invention discloses a credit strategy determination method and device, a program product and a storage medium. The method comprises the following steps: performing causal analysis on agricultural production data by using a causal artificial intelligence technology to obtain a causal analysis result; simulating an agricultural production process corresponding to the agricultural production data by using a digital twin technology to obtain a simulation result; and determining a target credit strategy based on the causal analysis result and the simulation result. Therefore, the causal artificial intelligence technology and the digital twinborn technology are combined, the causal relationship in agricultural production can be deeply analyzed, and the change of the agricultural production environment can be reflected in real time, so that more accurate and dynamic support can be provided for credit decision, and an adaptive credit strategy can be provided for an agricultural production unit.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more specifically, to a credit strategy determination method, device, program product and storage medium. Background Art

[0002] With the development of agricultural modernization, agricultural production units need to increase financial support and expand the scale of agricultural production, so the demand for agricultural credit is increasing.

[0003] At present, the main method is to rely on historical agricultural production data and empirical judgment to provide corresponding credit strategies for agricultural production units. However, in the current environment, agricultural production is facing the challenges of multiple uncertain factors such as climate change, pests and diseases, which directly affect the output and quality of agricultural products. As a result, relying on historical agricultural production data and empirical judgment, it is impossible to provide agricultural production units with adaptive credit strategies. Summary of the invention

[0004] In view of this, the embodiments of the present application disclose a credit strategy determination method, device, program product and storage medium to provide an adaptive credit strategy for agricultural production units.

[0005] The technical solutions provided by the embodiments of this application are as follows:

[0006] In a first aspect, an embodiment of the present application provides a method for determining a credit strategy, the method comprising:

[0007] Use causal artificial intelligence technology to conduct causal analysis on agricultural production data and obtain causal analysis results;

[0008] Using digital twin technology to simulate the agricultural production process corresponding to the agricultural production data to obtain simulation results;

[0009] Based on the causal analysis results and the simulation results, a target credit strategy is determined.

[0010] In combination with the first aspect, in a possible implementation of the first aspect, the agricultural production data includes: meteorological data, pest data, historical yield data and agricultural policy data;

[0011] The causal analysis of agricultural production data using causal artificial intelligence technology to obtain causal analysis results includes:

[0012] A structural causal model is used to perform causal analysis on agricultural production data to obtain causal analysis results; the causal analysis results include the causal relationship between at least two of the following: meteorological conditions, pest occurrence, yield, agricultural policy, and unobserved factors.

[0013] In combination with the first aspect, in a possible implementation of the first aspect, the agricultural production data is analyzed using a structural causal model to obtain a causal analysis result, including:

[0014] Based on the agricultural production data, test whether the causal effect in the structural causal model can be identified;

[0015] If the causal effect can be identified, determining a parameter corresponding to the causal effect;

[0016] The causal analysis result is determined based on the parameter.

[0017] In combination with the first aspect, in a possible implementation of the first aspect, simulating the agricultural production process corresponding to the agricultural production data using the digital twin technology to obtain a simulation result includes:

[0018] Using digital twin technology to create a virtual environment corresponding to the agricultural production data;

[0019] The agricultural production process is simulated by using the virtual environment to obtain simulation results. The agricultural production process includes: crop growth process, soil changes, and climate changes.

[0020] In combination with the first aspect, in a possible implementation manner of the first aspect, the agricultural production data includes: first sensor data and first monitoring data, and the method further includes:

[0021] Acquire second sensor data and second monitoring data, where the second sensor data is collected after the first sensor data, and the second monitoring data is collected after the first monitoring data;

[0022] Based on the second sensor data and the second monitoring data, the virtual environment is adjusted.

[0023] In combination with the first aspect, in a possible implementation of the first aspect, determining a target credit strategy based on the causal analysis result and the simulation result includes:

[0024] Determining the credit risks and credit returns corresponding to a plurality of candidate credit strategies based on the causal analysis results and the simulation results;

[0025] The target credit strategy is determined from the plurality of candidate credit strategies based on the credit risk and the credit return.

[0026] In combination with the first aspect, in a possible implementation manner of the first aspect, the method further includes:

[0027] Based on the simulation results, agricultural management adjustment strategies are determined.

[0028] In a second aspect, an embodiment of the present application provides a credit strategy determination system, the system comprising:

[0029] A causal artificial intelligence module is used to perform causal analysis on agricultural production data using causal artificial intelligence technology to obtain causal analysis results;

[0030] A digital twin module, used to simulate the agricultural production process corresponding to the agricultural production data using digital twin technology to obtain simulation results;

[0031] A decision support module is used to determine a target credit strategy based on the causal analysis result and the simulation result.

[0032] The credit strategy determination system has the function of implementing the credit strategy determination method in the first aspect or any possible implementation of the first aspect. The function can be implemented by hardware or by hardware executing corresponding software, and the hardware or software includes one or more modules corresponding to the above functions.

[0033] In a third aspect, an embodiment of the present application provides a credit strategy determination device, which may include at least one processor, and the processor is used to call computer instructions in a memory so that the credit device executes the credit strategy determination method in the above-mentioned first aspect or any optional embodiment of the first aspect.

[0034] In combination with the third aspect, in a possible implementation of the third aspect, the credit strategy determination device may further include a memory.

[0035] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which may include instructions. When the instructions are executed on a computer, the computer executes the credit strategy determination method in the above-mentioned first aspect or any optional embodiment of the first aspect.

[0036] In a fifth aspect, an embodiment of the present application provides a computer program product, which may include instructions. When the instructions are executed on a computer, the computer executes the credit strategy determination method in the above-mentioned first aspect or any optional embodiment of the first aspect.

[0037] In a sixth aspect, an embodiment of the present application provides a chip system, which includes a processor for supporting the device to implement the functions involved in the above aspects, for example, sending or processing the data and / or information involved in the above methods. In one possible design, the chip system also includes a memory, which is used to store program instructions and data necessary for the device. The chip system can be composed of chips, or it can include chips and other discrete devices.

[0038] In the seventh aspect, an embodiment of the present application provides a chip, comprising one or more interface circuits and one or more processors; the interface circuit is used to receive signals from a memory of an electronic device and send signals to the processor, the signals including computer instructions stored in the memory; when the processor executes the computer instructions, the electronic device executes the credit strategy determination method in the first aspect or any optional implementation scheme in the first aspect.

[0039] Based on the above technical solution, this application has the following beneficial effects:

[0040] The embodiments of the present application disclose a method, device, program product and storage medium for determining a credit strategy. The method includes: using causal artificial intelligence technology to perform causal analysis on agricultural production data to obtain causal analysis results; using digital twin technology to simulate the agricultural production process corresponding to the agricultural production data to obtain simulation results; and determining the target credit strategy based on the causal analysis results and the simulation results. It can be seen that the embodiments of the present application combine causal artificial intelligence technology and digital twin technology, which can deeply analyze the cause-effect relationship in agricultural production and reflect the changes in the agricultural production environment in real time, thereby providing more accurate and dynamic support for credit decisions, and further providing agricultural production units with adaptive credit strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on the disclosed drawings.

[0042] Figure 1 A flowchart of a credit strategy determination method disclosed in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of a credit strategy determination system disclosed in an embodiment of the present application;

[0044] Figure 3 A schematic diagram of each module in a credit strategy determination system disclosed in an embodiment of the present application;

[0045] Figure 4 A schematic diagram of a cause-effect diagram disclosed in an embodiment of the present application;

[0046] Figure 5 A flowchart of a method for identifying and estimating a structural causal model disclosed in an embodiment of the present application;

[0047] Figure 6 A flowchart of a causal analysis method based on a structural causal model disclosed in an embodiment of the present application;

[0048] Figure 7 A flowchart of a data collection and processing method disclosed in an embodiment of the present application;

[0049] Figure 8 A flowchart of a method for processing a virtual layer disclosed in an embodiment of the present application;

[0050] Fig. 9 This is a flowchart of another credit strategy determination method disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The embodiments based on the present application are all within the scope of protection of the present application.

[0052] In order to better understand the embodiments of the present application, some professional terms involved in the embodiments of the present application are explained below.

[0053] (1) Causal Artificial Intelligence (Causal AI) technology is a branch of artificial intelligence that focuses on discovering and modeling causal relationships between variables from data, rather than just correlations. Causal AI uses a series of statistical and machine learning methods to identify the effects of interventions, predict potential causal effects, and perform counterfactual analysis. In the embodiments of the present application, causal AI is used to analyze the causal relationship between external factors such as weather and insect pests and agricultural product output, providing a scientific basis for credit decisions.

[0054] (2) Digital Twin technology, which creates a virtual copy of a physical entity or system to simulate and analyze the entity. The digital twin technology can be updated in real time to reflect the state and behavior of its physical counterpart. In the application embodiment, the digital twin technology is used to build a virtual model of agricultural production, simulate the crop growth process and yield changes, so as to predict future yields and assess credit risks.

[0055] (3) Full-parameter digital twin technology refers to a digital twin model that includes three levels: physical layer, information processing layer, and virtual layer. This digital twin model not only replicates the structure and behavior of the physical entity, but also includes all data and information related to the entity, thereby achieving a more comprehensive and accurate simulation. In the embodiment of the present application, full-parameter digital twin technology is used to construct an agricultural production model that comprehensively considers multiple factors. The agricultural production model can also be expressed as a digital twin model, a virtual model, a virtual environment, etc.

[0056] (4) A credit decision support system is a collection of technologies and tools that assist financial institutions in making loan decisions. The credit decision support system helps financial institutions assess credit risks and formulate reasonable credit policies by analyzing various factors such as the borrower's financial status, credit history, and market conditions. The credit decision support system in the embodiment of this application is specifically aimed at the agricultural production field, combining causal AI and digital twin technology to provide more accurate credit assessment.

[0057] (5) Meteorological data refers to observations and records of atmospheric conditions and weather phenomena, including temperature, humidity, wind speed, precipitation, etc. In agricultural production, meteorological data has a direct impact on crop growth. In the embodiments of the present application, meteorological data is used as one of the input variables to construct and train causal models and digital twin models.

[0058] (6) Pest data, including information on the types, numbers, occurrence time, and impact range of insects that are harmful to crops. Pests are one of the common disasters in agricultural production and have a significant impact on crop yield and quality. In the implementation of this application, pest data is also used as a key input to simulate and predict the potential impact of pest data on agricultural product yield.

[0059] (7) Yield data refers to the production volume of agricultural products within a certain period of time. Yield data is an important basis for evaluating agricultural production efficiency and formulating credit policies. In the application examples, yield data is not only used to train causal models and digital twin models, but also used to verify the prediction accuracy of the model and the effectiveness of credit decisions.

[0060] In the current environment, agricultural production is facing challenges from multiple uncertainties such as climate change, pests and diseases, and soil degradation, which directly affect the output and quality of agricultural products, and thus affect the risk assessment and decision-making of agricultural credit. Traditional methods for determining credit strategies rely on historical data and empirical judgments, and lack an in-depth understanding of the complexity and dynamic changes of agricultural production, resulting in the fact that the credit strategies provided to agricultural production units may not accurately reflect the real risks and potential of agricultural production.

[0061] To this end, the embodiments of the present application disclose a credit strategy determination method, device, program product and storage medium, which combine causal artificial intelligence technology and digital twin technology, can deeply analyze the cause-effect relationship in agricultural production, and reflect the changes in the agricultural production environment in real time, thereby providing more accurate and dynamic support for credit decisions, and further providing agricultural production units with adaptive credit strategies.

[0062] See also Figure 1 , a flowchart of a credit strategy determination method disclosed in an embodiment of the present application, the credit strategy determination method may include the following steps:

[0063] S101. Use causal artificial intelligence technology to conduct causal analysis on agricultural production data to obtain causal analysis results.

[0064] Among them, agricultural production data can include: meteorological data, pest data, historical yield data, sensor data, market supply and demand conditions, etc. The results of causal analysis can reflect the causal relationship between external factors and yield.

[0065] S102. Use digital twin technology to simulate the agricultural production process corresponding to the agricultural production data to obtain simulation results.

[0066] In the embodiment of the present application, the digital twin technology is used to simulate the corresponding agricultural production process based on agricultural production data to obtain simulation results. The digital twin technology can be a full-parameter digital twin, and the simulation results include the simulated output of agricultural products.

[0067] S103. Determine the target credit strategy based on the causal analysis results and simulation results.

[0068] The credit policy in the embodiment of the present application may include: credit limit, interest rate, loan term, etc.

[0069] It can be seen that the embodiment of the present application combines causal artificial intelligence technology and digital twin technology, which can deeply analyze the cause-effect relationship in agricultural production and reflect the changes in the agricultural production environment in real time, thereby providing more accurate and dynamic support for credit decisions of agricultural products, and further providing agricultural production units with adaptive credit strategies.

[0070] In a possible implementation, S101 in the credit strategy determination method provided in the implementation of the present application includes: using a structural causal model (Structura l Causa l Model l, SCM) to perform causal analysis on agricultural production data to obtain causal analysis results.

[0071] Among them, SCM is a statistical method that combines graph theory and causal inference, which can handle complex causal relationships and perform accurate causal inference when there are hidden variables and unobserved confounding factors. The SCM in the embodiment of the present application can be a graph-based SCM.

[0072] The agricultural production data in the embodiment of the present application includes: meteorological data, pest data, historical yield data and agricultural policy data, and the causal analysis results include the causal relationship between at least two of the following: meteorological conditions, pest occurrence, yield, agricultural policy, and unobserved factors.

[0073] In the embodiment of the present application, using SCM to perform causal analysis on agricultural production data to obtain causal analysis results can specifically include: testing whether the causal effect in the structural causal model can be identified based on the agricultural production data; if the causal effect can be identified, determining the parameters corresponding to the causal effect; and determining the causal analysis results based on the parameters.

[0074] It can be seen that the causal relationship between multiple influencing factors can be analyzed and determined in the embodiment of the present application, so that more accurate support can be provided for the credit decision of agricultural products. Moreover, the structural causal model is used for causal analysis in the embodiment of the present application, so that more complex causal relationships can be handled, and accurate causal inference can be made for hidden variables and unobserved confounding factors, so that more accurate support can be provided for the credit decision of agricultural products.

[0075] In a possible implementation, S102 in the credit strategy determination method provided in the implementation of this application includes:

[0076] a1. Use digital twin technology to create a virtual environment corresponding to agricultural production data.

[0077] The virtual environment in the embodiment of the present application can be a virtual farmland that can accurately reflect actual agricultural production conditions, and can include crop model construction, soil model construction, climate model integration, etc.

[0078] a2. Use the virtual environment to simulate the agricultural production process and obtain simulation results.

[0079] In the embodiments of the present application, data in a virtual environment can be used to simulate various processes of agricultural production, and the agricultural production process may include: crop growth process, soil changes, climate change, etc.

[0080] In the implementation of this application, agricultural management adjustment strategies can also be determined based on the simulation results. Among them, agricultural management adjustment strategies can include irrigation plans, fertilization strategies, and pest and disease control measures, etc. In this way, corresponding management strategies can be provided for agricultural production units to ensure the normal growth of crops and promote the sustainable development of agricultural production.

[0081] It can be seen that in the embodiments of the present application, various changes in the agricultural production process can be simulated in the corresponding virtual environment, so that more accurate simulation results can be obtained, and more accurate support can be provided for credit decisions on agricultural products.

[0082] In a possible implementation, the agricultural production data in the credit strategy determination method provided in the implementation of the present application includes: first sensor data and first monitoring data, and the method further includes:

[0083] b1. Acquire second sensor data and second monitoring data.

[0084] The second sensor data is collected after the first sensor data, and the second monitoring data is collected after the first monitoring data.

[0085] In the embodiments of the present application, environmental sensors such as temperature, humidity, and light intensity, as well as soil sensors for monitoring soil moisture, fertility, and pH value can be installed in the farmland. The sensors collect sensor data such as crop growth status, soil conditions, and climate change in real time. Monitoring equipment can be used to track the use of agricultural machinery such as irrigation systems, fertilization equipment, and harvesters to obtain monitoring data.

[0086] b2. Adjust the virtual environment based on the second sensor data and the second monitoring data.

[0087] It can be seen that in the embodiments of the present application, sensor data and monitoring data collected in real time or near real time can be used to dynamically adjust the virtual environment, so as to accurately reflect the current agricultural production status. More accurate simulation results can provide more accurate support for credit decisions of agricultural products.

[0088] In a possible implementation, S103 in the credit strategy determination method provided in the implementation of this application includes:

[0089] c1. Based on the causal analysis results and simulation results, determine the credit risks and credit returns corresponding to multiple candidate credit strategies.

[0090] c2. Determine the target credit strategy from multiple candidate credit strategies based on credit risk and credit return.

[0091] It can be seen that in the embodiment of the present application, based on the causal analysis results and simulation results, accurate credit risk assessment and credit return assessment can be performed on multiple candidate credit strategies, and a more suitable credit strategy can be determined.

[0092] See also Figure 2 and Figure 3 , Figure 2 is a schematic diagram of a credit strategy determination system disclosed in an embodiment of the present application, Figure 3A schematic diagram of each module in a credit strategy determination system disclosed in an embodiment of the present application. The credit strategy determination system in the embodiment of the present application mainly includes a causal AI module, a digital twin module and a decision support module. Among them, the causal AI module is used to provide the causal relationship of the impact of external factors on output. The causal AI module will perform data preprocessing, causal inference, and model verification. The causal AI module will input the causal relationship data into the decision support module, and the causal relationship data is the result of causal analysis; the digital twin module is used to provide real-time simulation data of the agricultural production process. The digital twin module will perform data collection, environmental simulation, and trend prediction. The data twin module will input real-time simulation data into the decision support module, and the real-time simulation data is the simulation result; the decision support module is used to comprehensively analyze and evaluate credit risks. The decision support module will perform risk assessment and strategy formulation. The decision support module may provide further data demand feedback to the causal AI module, and may provide simulation parameter adjustment feedback to the digital twin module.

[0093] The causal AI module and the digital twin module in the embodiment of the present application can collaborate with each other through data interfaces and APIs. The causal AI module passes the identified causal relationship to the decision support module in the form of parameters or models, and the digital twin module inputs the real-time simulation data and prediction results to the decision support module in the form of data streams. The outputs of the causal AI module and the digital twin module together constitute the input of the decision support module, and the decision support module performs risk assessment and credit strategy formulation based on these inputs. During the collaboration process, the decision support module may, based on the preliminary analysis results, propose further data requirements or simulation parameter adjustments to the causal AI module or the digital twin module in order to conduct a deeper analysis or a more accurate simulation. This two-way interaction ensures that the entire credit strategy determination system can flexibly adapt to changes in the agricultural production environment and adjust credit strategies in a timely manner to meet new challenges.

[0094] The credit strategy determination method in the embodiment of the present application may further include: a data collection module and a user interaction interface.

[0095] It should be noted that the credit strategy determination method provided in the above implementation can be implemented by the credit strategy determination system. The credit strategy determination method is described in more detail below in conjunction with the credit strategy determination system.

[0096] (a) Causal AI module

[0097] The causal AI module in the embodiment of the present application is responsible for identifying the causal relationship between variables from historical and real-time data, extracting and modeling the causal relationship between key factors affecting agricultural product output. This module includes data preprocessing algorithms, causal inference algorithms, and model verification algorithms. It uses statistical and machine learning methods to analyze data, identify potential causal effects, quantify these causal effects into indicators that can be used for decision support, and conduct counterfactual analysis. Among them, the boundaries of the causal AI module are clear, focusing on the causal analysis of data, without involving the collection of real-time data or the simulation of agricultural production processes.

[0098] The causal AI module is mainly used for data preprocessing, causal inference, model verification, etc. The following describes the processing flow of the causal AI module.

[0099] The data preprocessing step of the causal AI module is a key step in ensuring data quality. At this stage, data cleaning is first performed to remove irrelevant information and noise, fill in missing values, and ensure the integrity and availability of the data set. Next, variable selection is performed to screen out the key variables that have the greatest impact on production through feature importance evaluation and dimensionality reduction techniques. Outlier detection can also be performed to identify and process data points that may distort the analysis results. Through these steps, the causal AI module is able to use high-quality data for subsequent causal inference.

[0100] Causal inference is the core of the causal AI module, which uses a series of algorithms such as causal tests and propensity score matching to identify and quantify causal effects between variables. Among them, causal tests are used to analyze whether one time series can predict the future values ​​of another time series, thereby revealing the causal relationship between variables. Propensity score matching is a statistical method used to simulate the conditions of randomized controlled trials in observed data, and to reduce the bias in the estimation of treatment effects by matching the observation units of the treatment group and the control group.

[0101] The model validation step of the causal AI module evaluates the generalization ability and stability of the causal model by using techniques such as cross-validation and bootstrapping. These validation techniques can detect whether the causal model is overfitting the training data and how the causal model performs on new, unseen data. Model validation is an important step in improving the accuracy and credibility of causal analysis results.

[0102] The causal AI technology used in the causal AI module in the embodiment of the present application can be Causal Graph Neural Networks (CGNN). Among them, CGNN is an advanced causal inference method that combines the representation learning and causal inference capabilities of graph neural networks, and can capture complex relationships between variables, including nonlinear relationships and high-order interactions. In the presence of hidden variables and unobserved confounding factors, CGNN can estimate the causal effects between variables by learning the graph structure representation of the data.

[0103] The input data of CGNN in the embodiment of the present application may include: meteorological data, pest data, yield data, agricultural policy data and other related factors. Among them, meteorological data includes climate factors closely related to agricultural production such as temperature, precipitation, humidity, wind speed, etc.; pest data covers the frequency, severity, time and location of occurrence of different types of pests; yield data includes historical yield records, which include crop types, yields, growth cycles, etc.; agricultural policy data involves credit policies, subsidy policies, market access policies and other factors that may affect farmers' planting decisions.

[0104] In the embodiment of the present application, CGNN can be combined with a graph-based structural causal model (Structura l Causa lMode l, SCM), and a graph-based SCM is used as the core algorithm for causal inference. Among them, SCM is a statistical method that combines graph theory and causal inference, which can handle complex causal relationships and perform accurate causal inference when there are hidden variables and unobserved confounding factors. The following is a detailed description of the principles of SCM and its application process in the embodiments of the present application.

[0105] ①The principle of SCM.

[0106] The basic principle of SCM is to represent the causal relationship as a directed acyclic graph (DAG), where nodes represent random variables and edges represent the causal relationship between variables. The key features of SCM include: causal graph, structural equation, causal effect identification and parameter estimation.

[0107] The SCM in the embodiment of the present application constructs a causal graph to represent the potential causal relationship between variables. Each node represents a variable, such as weather conditions, pest occurrence, etc., and the edge represents the direct impact between variables. Figure 4 , which is a schematic diagram of a cause-effect diagram provided in an embodiment of the present application. Figure 4The intermediate nodes include: meteorological conditions, pest occurrence, credit policy, market supply and demand, output and unobserved factors; the edges include: the direct impact between meteorological conditions and output, unobserved factors, the direct impact between pest occurrence and output, unobserved factors, the direct impact between credit policy and output, unobserved factors, and the direct impact between market supply and demand and output, unobserved factors.

[0108] The SCM in the embodiment of the present application defines a structural equation for each variable, which describes how the variable depends on other variables and potential unobserved factors. Please refer to Table 1 below, which is an example table of structural equations, and Table 1 describes the dependencies between multiple variables.

[0109] Table 1 Structural equation example table

[0110]

[0111] In the example in Table 1, there are four main variables: yield (Y), meteorological conditions (X1), pest occurrence (X2), and credit policy (X3). Each variable has a structural equation that describes how the variable depends on the other variables and potential unobserved factors (U).

[0112] In the yield equation in Table 1, yield (Y) is a function of meteorological conditions (X1), pest occurrence (X2) and credit policy (X3). Among them, β1, β2 and β3 are corresponding coefficients, which represent the direct impact of meteorological conditions, pest occurrence and credit policy on yield; ε is the error term, which represents other variables that affect yield besides meteorological conditions, pest occurrence and credit policy.

[0113] In the meteorological condition equation in Table 1, meteorological conditions (X1) depend on time (T) and unobserved factors (U). Among them, α1 is the coefficient, which represents the influence of time on meteorological conditions; δ1 is the influence of unobserved factors on meteorological conditions; η1 is the error term.

[0114] In the pest occurrence equation in Table 1, pest occurrence (X2) depends on crop type (C) and unobserved factors (U). Among them, α2 is the coefficient, which represents the effect of crop type on pest occurrence; δ2 is the effect of unobserved factors on pest occurrence; η2 is the error term.

[0115] In the credit policy equation in Table 1, credit policy (X3) depends on economic indicators (E) and unobserved factors (U). Among them, α3 is the coefficient, which represents the impact of economic indicators on credit policy; δ3 is the impact of unobserved factors on credit policy; η3 is the error term.

[0116] In the example of Table 1, the unobserved factor (U) is a latent variable, which may include factors such as soil quality, farmers' planting techniques, market changes, etc. These factors may affect all variables in SCM, but are not directly observed in the current analysis. Through the structural equation in Table 1, the causal effects between variables can be estimated, and appropriate methods such as maximum likelihood estimation (MLE) can be used in statistical software to estimate the coefficients in the structural equation, which helps to understand how different factors affect the output of agricultural products and provide a scientific basis for credit decisions. It is understandable that Table 1 is only an illustrative description and should not be construed as a limitation on the embodiments of the present application.

[0117] In the embodiment of the present application, the SCM uses an identification algorithm to determine which causal effects can be identified from the data, and uses an estimation algorithm to calculate the parameters of these causal effects. Among them, a recursive identification algorithm can be used to identify and estimate the SCM to ensure that the causal effects in the SCM can be accurately identified and estimated from the collected agricultural production data.

[0118] See also Figure 5 , is a flowchart of a method for identifying and estimating a structural causal model provided in an embodiment of the present application. Figure 5 The identification and estimation of SCM using a recursive identification algorithm is described.

[0119] The recognition phase in the embodiments of the present application may include the following steps: model specification, application of a recursive recognition algorithm, evaluation of recognition conditions, and model modification.

[0120] In the model specification step, all variables in the SCM and their potential causal relationships are explicitly specified based on theories and previous studies in the field of agricultural production. These causal relationships are represented by a causal graph, in which nodes represent variables and edges represent causal effects.

[0121] In the recursive identification algorithm application step, the recursive identification algorithm determines the identification status of the SCM by analyzing each structural equation in the SCM. For each endogenous variable, that is, a variable affected by other variables, the recursive identification algorithm traces all direct and indirect causal paths and evaluates whether there is enough independent information to estimate all parameters related to the variable.

[0122] In the identification condition evaluation step, the recursive identification algorithm evaluates each causal effect in the SCM to determine whether each causal effect can be uniquely identified from the agricultural production data. If all causal effects in the SCM can be identified, then the SCM is considered to be recursively identified.

[0123] In the model modification step, if the recursive identification algorithm finds that some causal effects cannot be identified from agricultural production data, the SCM may need to be modified. The modification may include redefining the relationship between variables, adding or deleting paths, or introducing new measurement variables to improve the identification of the SCM.

[0124] The estimation stage in the embodiment of the present application may include the following steps: parameter estimation, model fitting evaluation, parameter significance test and result interpretation and application.

[0125] In the parameter estimation step, once the SCM is confirmed to be recursively identified, an estimation algorithm can be used to calculate the parameters of each causal effect in the SCM. This involves MLE or other appropriate estimation methods, aiming to find the best parameter values ​​so that the covariance matrix predicted by the model is as close as possible to the covariance matrix of the actual observed data.

[0126] In the model fit assessment step, after parameter estimation, fit indices such as chi-square test, comparative fit index (CFI), root mean square error of approximation (RMSEA) are used to assess the goodness of fit of the SCM. These indices provide an overall assessment of the consistency of the SCM with the data.

[0127] In the parameter significance test step, a significance test is performed on the estimated parameters to determine whether the estimated parameters are statistically significant. This step can be completed by calculating the standard error of the parameter estimate and the corresponding p-value.

[0128] In the step of interpretation and application of results, the estimation results are interpreted to determine which causal effects are significant, and the SCM is interpreted and discussed based on these findings to obtain the causal analysis results. The causal analysis results can be used to guide agricultural production credit decisions, such as evaluating the potential impact of different credit policies on agricultural product output, etc.

[0129] It can be seen that by using a recursive recognition algorithm in the embodiments of the present application, it is possible to ensure that the causal effect of SCM is not only theoretically reasonable but also statistically verifiable, thereby providing solid data support for agricultural production credit decisions.

[0130] ② Description of the application of SCM in the embodiments of this application.

[0131] See also Figure 6 , which is a flow chart of a causal analysis method based on a structural causal model disclosed in an embodiment of the present application. Figure 6 The causal analysis method based on the structural causal model in includes the following steps:

[0132] 1. Data preparation: Collect and organize agricultural production data, including meteorological data, pest data, yield data, etc., and convert these data into a format suitable for SCM analysis.

[0133] 2. Causal graph construction: Based on domain knowledge and prior information, a preliminary causal graph is constructed to determine the possible causal relationships between variables. This step may require the participation of domain experts and multiple rounds of iterations.

[0134] 3. Model identification: Use the SCM identification algorithm to test whether the constructed causal diagram can be identified from the collected data, that is, to test the identifiability of the causal diagram and determine whether there is enough information to estimate all causal effects.

[0135] 4. Parameter estimation: Parameter estimation is performed on the identified causal effects, which usually involves using statistical software to perform complex calculations such as maximum likelihood estimation or Bayesian estimation.

[0136] 5. Model validation: Use cross-validation, sensitivity analysis and other methods to verify the robustness of SCM and ensure the reliability of model results.

[0137] 6. Interpretation and application of results: Interpret the results of SCM as actual cause-effect relationships in agricultural production, obtain causal analysis results, and apply the causal analysis results to credit decision support.

[0138] It can be seen that in the embodiment of the present application, SCM provides a data-driven causal inference method for agricultural production credit decisions, enabling decision makers to more accurately understand and predict how various factors in agricultural production interact with each other, and how these interactions affect the final yield and credit risk.

[0139] (2) Digital twin module.

[0140] The digital twin module is responsible for creating and maintaining a virtual model of agricultural production, including the physical layer, information processing layer, and virtual layer. The module has real-time data processing capabilities and uses high-precision sensor data, real-time monitoring data, and advanced simulation technology to build a dynamically updated digital copy of the agricultural production process to ensure that real-time simulation data can accurately reflect the actual situation of agricultural production. Among them, the digital twin module not only simulates crop growth and environmental changes, but also predicts future trends and potential problems, providing real-time data support for decision-making.

[0141] The following is a detailed description of the physical layer, information processing layer and virtual layer in the digital twin module, as well as the application of the digital twin module.

[0142] ①Physical layer.

[0143] The physical layer is the basis of the digital twin module, which involves all physical entities in the actual agricultural production environment, such as crops, soil, agricultural machinery, etc. Among them, real-time data can be collected by deploying various sensors and monitoring equipment.

[0144] Specifically, environmental sensors such as temperature, humidity, and light intensity can be installed in farmland, as well as soil sensors for monitoring soil moisture, fertility, and pH. These sensors collect data on crop growth conditions, soil conditions, climate change, and other real-time data. Monitoring equipment can also be used to track the use of agricultural machinery such as irrigation systems, fertilization equipment, and harvesters.

[0145] In addition to the above agricultural production data collection methods, the following data collection methods can also be considered: public data set utilization, online data collection, mobile device application, community participation, and existing infrastructure collection. Among them, the public data set utilization method can integrate public agricultural data sets from government departments and research institutions to provide macro-level data support for the digital twin model; the online data collection method can collect farmers' feedback on agricultural production practices, challenges and needs through social media analysis, online questionnaires, etc.; the mobile device application method can develop mobile applications to enable farmers to use smartphones to record and upload key information such as crop growth and meteorological data; the community participation method can encourage farmers and the public to participate in data collection through community surveys and citizen science projects to enrich the micro data of agricultural production; the existing infrastructure collection method can use agricultural service points and rural information stations as nodes for data collection and upload to improve the convenience and efficiency of data collection.

[0146] Please refer to Table 2 below for an example of data collection methods. Table 2 provides a variety of data collection methods.

[0147] Table 2. Example of data collection method

[0148]

[0149]

[0150] Table 2 provided in the examples of this application summarizes the various low-cost data collection methods used in the examples of this application. These methods together constitute a diversified data collection system, which helps the examples of this application to effectively expand the data sources without adding too much economic burden, improve the diversity and richness of data, and provide comprehensive data support for agricultural production credit decisions. At the same time, these methods also help to increase farmers' participation and trust in the project, and promote the dissemination and application of agricultural technology.

[0151] ②Information processing layer.

[0152] In the application embodiment, the information processing layer is a crucial component of the digital twin module. The information processing layer ensures that a large amount of raw data collected from the physical layer can be effectively converted and utilized to support the construction and operation of the digital twin model.

[0153] See also Figure 7 , is a flow chart of a data collection and processing method disclosed in an embodiment of the present application. The data collection and processing method can be implemented based on the physical layer and the information processing layer, wherein the data collection part includes: collecting data from sensors and monitoring equipment, and the data processing part includes: data preprocessing, data fusion and feature extraction. The following is combined Figure 7 The data processing part is explained in detail.

[0154] Data preprocessing in the information processing layer includes: data cleaning, data standardization, and data conversion.

[0155] Data cleaning includes identifying and correcting errors, outliers, and inconsistencies in the data, such as removing or filling missing data points, smoothing noisy data, and identifying and processing outliers. Data standardization is to ensure the consistency and comparability of data. The data needs to be standardized, including scaling the data to a uniform range or converting data in different formats to a uniform format. Data conversion is the conversion of raw data into a format suitable for analysis, such as converting time series data into a form that can be used for statistical analysis, or converting non-numeric data into numerical data for quantitative analysis.

[0156] Data fusion in the information processing layer combines data from multiple sources to provide more comprehensive and accurate information, including: multi-source data integration, data alignment, and data association.

[0157] Among them, multi-source data integration integrates data from different sensors and monitoring equipment into a unified data set, such as combining data from soil moisture sensors with data from climate monitoring stations to obtain more comprehensive agricultural environmental information. Data alignment ensures that the timestamp and spatial location information between different data sources are consistent for effective data comparison and analysis. Data association is to identify and establish relationships between data, such as associating crop growth data with corresponding climate condition data to analyze the impact of climate change on crop growth.

[0158] Feature extraction in the information processing layer is to identify and extract key information from the processed data that is helpful for model analysis and prediction, including: key indicator identification, feature selection, and feature construction.

[0159] Among them, key indicator identification is to determine which data features are most important for agricultural production and credit decisions, such as crop growth rate, probability of pests and diseases, soil fertility, etc. Feature selection is to select the most informative features from a large number of data features to reduce the complexity and computational requirements of the model, which may involve the use of statistical tests, machine learning algorithms or other feature selection techniques. Feature construction is to build new features from existing data to better capture the complexity of agricultural production, such as combining data from multiple sensors to create a comprehensive crop health indicator.

[0160] In the embodiment of the present application, through the three steps of data preprocessing, data fusion and feature extraction, the information processing layer can convert the original and scattered data into a clear and structured data set, providing accurate and reliable input for the digital twin model, which not only helps to improve the predictive ability of the digital twin model, but also provides strong data support for agricultural production and credit decisions.

[0161] ③Virtual layer.

[0162] In the application embodiments, the design and implementation of the virtual layer is the key to ensuring the effectiveness of the digital twin model.

[0163] See also Figure 8 , is a flow chart of a method for processing a virtual layer according to an embodiment of the present application, the method comprising: virtual environment creation, simulation, and real-time update. Figure 7 The virtual layer is described in detail.

[0164] 1. Create a virtual environment.

[0165] Virtual environment creation is the cornerstone of the virtual layer, which requires the construction of a virtual farmland that can accurately reflect actual agricultural production conditions. Virtual environment creation can include the following steps: crop model construction, soil model development, and climate model integration.

[0166] In the step of building a crop model, existing crop growth models can be developed or adopted, which can simulate the growth cycle, physiological response and yield formation of specific crop species. Among them, the crop model needs to include the response mechanism to factors such as light, temperature, water and nutrition.

[0167] In the soil model development step, a soil model is created that describes the physical, chemical, and biological properties of the soil and how these properties affect crop water and nutrient absorption. The soil model should include parameters such as soil structure, organic matter content, pH, and nutrient levels.

[0168] In the climate model integration step, climate models are integrated to simulate meteorological conditions, such as temperature, precipitation, wind speed, and humidity, and how these conditions change over time. The climate model should be able to provide historical climate data and predictions of future climate trends.

[0169] 2. Simulation.

[0170] Simulation is the core activity of the virtual layer, which uses data in the virtual environment to simulate various processes and phenomena of agricultural production. Simulation includes: crop growth process simulation, soil change dynamics simulation, climate change impact assessment, etc.

[0171] In the simulation of crop growth process, crop models can be used to simulate key growth stages such as sowing, germination, growth, flowering, fruiting and harvesting. The simulation results should be able to predict the growth and yield potential of crops.

[0172] In soil dynamics simulation, soil models are used to simulate changes in soil conditions, such as water movement, nutrient cycling, and soil erosion. These simulations help understand strategies for maintaining and improving soil fertility.

[0173] In climate change impact assessment, climate models are used to evaluate the potential impacts of long-term and short-term climate change on agricultural production, including the simulation of extreme climate events such as droughts, floods and heat waves.

[0174] 3. Update in real time.

[0175] To ensure the accuracy and practicality of the virtual layer, the virtual environment needs to be continuously updated based on the real-time data provided by the physical layer. This step involves: data synchronization mechanism, model parameter adjustment, feedback and decision support.

[0176] In the data synchronization mechanism, a data synchronization mechanism is established to transmit the sensor data collected by the physical layer to the virtual environment in real time or near real time, including crop growth status, soil moisture and temperature, and real-time meteorological data.

[0177] In model parameter adjustment, model parameters in the virtual environment are adjusted based on real-time data to ensure that the model can accurately reflect the current agricultural production situation. This may involve adjusting parameters such as crop growth rate, soil moisture content, and climate conditions.

[0178] In feedback and decision support, the updated virtual environment is used for predictive analysis and the results are fed back to agricultural production managers and decision makers. This helps them to adjust agricultural management strategies in a timely manner, such as irrigation plans, fertilization strategies, and pest and disease control measures.

[0179] It can be seen that in the embodiments of the present application, through the steps of virtual environment creation, simulation, and real-time update, the virtual layer can not only provide a highly accurate and real-time updated simulation platform for agricultural production, but also provide strong technical support for agricultural production management and credit decision-making. The establishment and maintenance of this virtual layer will greatly improve the efficiency and sustainability of agricultural production, and also bring new opportunities and challenges to the field of agricultural credit.

[0180] ④Application of digital twin module.

[0181] In the embodiment of the present application, the framework can be implemented through full-parameter digital twin technology to build a comprehensive, highly accurate and real-time updated digital twin model, which not only includes meteorological, insect pest and yield data, but also can integrate other relevant factors such as soil quality, planting technology, etc.

[0182] The digital twin model in the embodiment of the present application can realize multi-factor integration, yield prediction, credit decision support, etc. Specifically, multiple key factors of agricultural production such as soil quality and planting technology are integrated into the digital twin model; the digital twin model is used to predict yield and provide decision support for agricultural production; the simulation results of the model are applied to credit decision-making, the potential risks and benefits of agricultural production are evaluated, and accurate credit evaluation is provided for farmers and financial institutions.

[0183] (III) Decision support module.

[0184] The decision support module integrates the outputs of the causal AI module and the digital twin module to provide credit decision recommendations. This module uses optimization algorithms and risk assessment models to evaluate the potential risks and benefits under different credit strategies based on the causal analysis results and simulation results, and formulates credit strategies to provide a basis for the final credit decision. Among them, the boundary of this module lies in the decision-making process, which requires processing and analyzing the data from the causal AI module and the digital twin module, and converting them into feasible credit strategies.

[0185] See also Fig. 9 , is a flow chart of another credit strategy determination method provided in an embodiment of the present application, the method comprising: receiving causal analysis results and simulation results, combining optimization algorithms, risk assessment, providing credit decision suggestions, interactive decision support, continuous learning and adaptation. Fig. 9 The decision support module is described in detail.

[0186] In the step of receiving causal analysis results and simulation results, the decision support module receives the causal analysis results from the causal AI module, which reveals the key factors and their interactions that affect the yield and quality of agricultural products. At the same time, it also receives the real-time simulation results provided by the digital twin module, which reflects the current and predicted agricultural production conditions.

[0187] In the step of combining optimization algorithms, the decision support module uses optimization algorithms such as linear programming, genetic algorithms, ant colony optimization, etc. to evaluate the potential risks and benefits under different credit strategies and obtain optimization results. Among them, these algorithms can find optimal or near-optimal solutions in complex decision spaces, taking into account the uncertainty and dynamic changes of agricultural production.

[0188] In the risk assessment step, the decision support module uses the risk assessment model to assess the risks that various credit strategies may bring, including credit default risk, market volatility risk, and natural disaster risk, and obtains risk assessment results. Among them, the risk assessment model can predict and quantify risks based on historical data, market analysis, and agricultural production simulation results.

[0189] In the step of providing credit decision recommendations, the decision support module provides credit decision recommendations based on risk assessment results and optimization results, that is, provides preliminary credit strategies. These recommendations include the determination of credit limits, the setting of interest rates, the selection of loan terms, and possible risk mitigation measures. Among them, credit decision recommendations are aimed at balancing risks and benefits, ensuring the effective use of credit resources, and protecting the interests of financial institutions.

[0190] In the interactive decision support step, the decision support module can provide interactive decision support, allowing users to adjust the parameters of the credit strategy, such as loan conditions and agricultural production practices, to observe the impact of different decisions on risk and return. This interactive analysis helps decision makers better understand the consequences of decisions and choose the most appropriate credit strategy.

[0191] In the step of continuous learning and adaptation, the decision support module can continuously learn and adapt through machine learning methods as time goes by and new data is collected, updating the risk assessment model and optimization algorithm in the decision support module to reflect the latest trends and changes in agricultural production and markets.

[0192] It can be seen that in the embodiment of the present application, by receiving causal analysis results and simulation results, combining optimization algorithms, risk assessment, providing credit decision recommendations, interactive decision support, continuous learning and adaptation, the decision support module provides a comprehensive, dynamic and intelligent framework for agricultural production credit decisions, which can not only help financial institutions make more scientific and reasonable credit decisions, but also promote the sustainable development and risk management of agricultural production.

[0193] Suppose an agricultural producer wants to obtain a loan to expand production. First, the historical yield data, weather data, and market conditions of the corresponding farmland can be collected. Then the causal AI module is used to analyze this data to identify the key factors affecting yield, and the digital twin module is used to simulate the current state of the farmland and possible future production conditions. Then the decision support module is used to synthesize this information, evaluate the risks and potential benefits of the loan, and provide reasonable credit limits and interest rate recommendations for agricultural producers.

[0194] The following is an exemplary description of an application scenario provided in an embodiment of the present application.

[0195] (1)Scene setting.

[0196] Xiao Zhang is a rice farmer who wants to expand his planting area and increase production, so he needs a loan to buy seeds, fertilizers and farm machinery.

[0197] (2) Data collection.

[0198] In addition to using soil moisture sensors and climate monitoring stations to collect data, Xiao Zhang also recorded data from manual observations, including crop growth conditions and the occurrence of pests and diseases. Xiao Zhang also used a smartphone app to record crop planting dates, fertilization and irrigation schedules, and photos and videos of crop growth.

[0199] In addition, the bank has worked with local agricultural cooperatives to share data collected by the cooperatives on market prices and demand; the bank has also participated in online agricultural communities to collect information on discussions among other farmers on planting techniques and market trends.

[0200] (3) Data processing

[0201] In the information processing layer, the credit strategy determination system integrates and preprocesses data from various sources, including data cleaning, data standardization, and feature extraction, to build a comprehensive agricultural production data set.

[0202] In the virtual layer, based on the processed comprehensive data, a virtual rice planting model was created using digital twin technology, simulating the impact of crop growth cycle and environmental conditions.

[0203] (4) Decision support.

[0204] The Causal AI module analyzes the cause-and-effect relationships between crop growth data, environmental conditions, and market trends, identifying key factors that influence yield and profitability.

[0205] The decision support module combines the simulation results of the virtual layer and the causal analysis results of the causal AI module to evaluate the risks and benefits of the loan. The decision support module also uses optimization algorithms such as genetic algorithms to consider the optimal impact of different loan plans on banks and farmers.

[0206] (5) Loan decision.

[0207] Based on the above analysis, the bank's credit strategy determination system recommends a loan of a certain amount and gives a reasonable interest rate and repayment plan. At the same time, the credit strategy determination system also provides planting suggestions, such as the best sowing time, fertilizer application amount and irrigation frequency.

[0208] (6) Feedback on results.

[0209] Xiao Zhang adjusted his planting plan based on the bank's advice and obtained a loan. As the planting season progressed, Xiao Zhang continued to update crop growth data through the mobile app, which was fed back into the digital twin module in real time. The bank monitored the simulation results of the rice planting model and adjusted loan conditions and provided further agricultural consulting services in a timely manner.

[0210] Through this scenario example, we can see how the detailed process in the embodiment of this application integrates multiple data collection methods to provide more comprehensive and accurate agricultural production and credit decision support. This process involves not only sensor data, but also manual observation, social media information, mobile application data and data from other partners, thus realizing a diversified data collection and analysis system.

[0211] The following is an exemplary description of another application scenario provided by an embodiment of the present application.

[0212] In this scenario example, the data collection module is responsible for collecting data from various sources, including weather stations, farmland monitoring systems, market databases, etc. The collected data is preprocessed to ensure data quality. Then, the data is sent to the causal AI module and the digital twin module for analysis and simulation. The causal analysis results and simulation results are used in the decision support module to formulate credit strategies. Among them, the data processing process includes: data collection, data preprocessing, data analysis and simulation, decision support, feedback and iteration. The data processing process of this scenario example is described in detail below.

[0213] (1) Data collection.

[0214] Collect climate data from meteorological stations in the area where the farmland is located, including temperature, precipitation, wind speed, and sunshine duration, which are crucial for predicting crop growing conditions; use various sensors installed in the farmland to collect data such as soil moisture, pH value, nutrient content, as well as images and video materials of crop growth conditions; obtain historical price data, demand trends, and competitor information from market databases, which help evaluate the market prospects and potential profits of agricultural products.

[0215] (2) Data preprocessing.

[0216] Check the collected data and remove erroneous, duplicate or incomplete records to ensure data accuracy and consistency; convert unstructured data (such as images and videos) into structured data to facilitate subsequent analysis and processing, such as using image recognition technology to extract growth status information from crop photos; standardize the data to make it conform to a unified format and scope to facilitate comparison and integration of data from different sources and types.

[0217] (3) Data analysis and simulation.

[0218] Use causal inference algorithms to analyze the causal relationship between meteorological, soil and market data, and identify key factors affecting crop yield and returns; input the preprocessed data into the digital twin model to simulate the growth and yield changes of crops under different environmental conditions.

[0219] (4) Decision support.

[0220] Combine the causal analysis results of the causal AI module and the simulation results of the digital twin module to evaluate the potential risks and benefits under different credit strategies; formulate credit strategies based on the risk assessment results, including loan amount, interest rate, repayment period and possible risk mitigation measures.

[0221] (5) Feedback and iteration.

[0222] Feed the output of the decision support module back to agricultural producers and financial institutions to help them understand the basis for their decisions and potential impacts. As new data is collected and market conditions change, the data processing process is continuously updated and optimized to improve the accuracy and adaptability of decisions.

[0223] Through this data processing process, the embodiment of the present application can provide scientific and accurate support for credit decision-making in agricultural production. This not only helps agricultural producers obtain more appropriate credit resources, but also helps financial institutions manage credit risks more effectively and promote the sustainable development of agricultural production.

[0224] It can be seen that the embodiment of the present application provides a comprehensive view of agricultural production by integrating multi-source data from weather stations, farmland monitoring systems and market databases. This comprehensive data utilization can more accurately predict crop yields and market demand, thereby providing a more reliable basis for credit decisions.

[0225] Among them, the multi-source data fusion technology adopted in the embodiment of the present application can effectively integrate and process large-scale multi-source heterogeneous data, including meteorological data, soil data, market data and social media information, etc., ensuring the comprehensiveness and diversity of the data, providing a solid foundation for subsequent analysis and decision-making, and helping to more accurately simulate and predict various possibilities of agricultural production, thereby reducing credit risks and improving the reliability of decision-making.

[0226] The causal AI module used in the embodiment of this application for causal relationship analysis can more accurately identify and quantify the key factors affecting yield and income, deeply explore the causal relationship behind the data, reveal the causal chain that is not easy to observe in agricultural production, and provide deep insights for credit decisions, rather than just superficial correlations. Compared with traditional correlation analysis, it can better reveal the internal mechanism of agricultural production, distinguish the direct and indirect effects between factors, provide a more scientific basis for credit decisions, help formulate targeted agricultural practices and credit strategies, and more effectively improve agricultural production efficiency and credit investment returns.

[0227] In the embodiment of the present application, a digital twin module is used to create a digital twin model of agricultural production, realize the synchronization of real-time data and the dynamic update of the model, and can simulate crop growth and market changes in a virtual environment, providing a highly flexible testing and prediction tool for agricultural production and credit decision-making. Different agricultural strategies can be tested without interfering with actual production, and the simulation results can help agricultural producers and financial institutions evaluate the effectiveness and risks of the strategies. The simulation optimization strategy can improve the success rate of actual operations and reduce credit risks.

[0228] The decision support module in the embodiment of the present application combines the optimization algorithm and the risk assessment model to provide a scientific and accurate method for formulating credit strategies. Among them, the optimization algorithm can find the optimal solution among multiple options and balance risks and benefits; the risk assessment model can predict and quantify potential market and production risks. In this way, accurate decision support can help financial institutions formulate more reasonable loan conditions, improve loan recovery rates, and improve the effectiveness and efficiency of credit strategies.

[0229] In the embodiment of this application, causal AI and full-parameter digital twin technology are combined to achieve an in-depth understanding and accurate simulation of agricultural production. Compared with traditional methods, it can reflect changes in the agricultural production environment in real time, effectively deal with hidden variables and unobserved confounding factors, and improve the accuracy of credit risk assessment and the efficiency of credit resource allocation. Among them, by accurately simulating the output of agricultural products, the credit strategy is more in line with the actual needs and market changes of agricultural production, thereby improving the accuracy of the credit strategy; by collecting and analyzing data related to agricultural production in real time, it can timely reflect changes in the agricultural production environment and improve the timeliness of credit decisions; by real-time monitoring and predicting agricultural production risks, it can timely adjust credit strategies, reduce the uncertainty and risk of agricultural production, and thus enhance risk management capabilities; by optimizing the allocation of credit resources and supporting the sustainable practice of agricultural production, it can improve the overall efficiency and sustainability of agricultural production, thereby promoting sustainable agricultural development. Through digital twin technology, agricultural producers can better understand and manage the agricultural production process, improve output and quality, and thus improve agricultural production efficiency.

[0230] The embodiments of the present application can effectively solve the problems and limitations of traditional credit strategy determination methods. Through the integration of real-time data and comprehensive consideration of external factors, more accurate and dynamic credit support can be provided for agricultural production. In addition, the combined use of causal AI and digital twin technology not only improves the accuracy of risk assessment, but also provides strong technical support for the sustainable development of agricultural production, marking that agricultural production credit decision-making has entered a new stage, with important practical application value and broad market prospects.

[0231] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0232] Furthermore, the embodiment of the present application also provides a credit device, including:

[0233] At least one processor, the processor is used to call the computer instructions in the memory so that the credit device executes any one of the implementation methods of the above-mentioned credit policy determination method.

[0234] Furthermore, an embodiment of the present application also provides a computer program product, including instructions, which, when executed on a computer, enable the computer to execute any one of the implementation methods of the above-mentioned credit policy determination method.

[0235] Furthermore, an embodiment of the present application also provides a computer-readable storage medium for storing a computer program. When the computer program is executed on a computer, the computer executes any one of the implementation methods of the above-mentioned credit policy determination method.

[0236] It can be known from the description of the above implementation mode that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., including several instructions for enabling a computer device (which can be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0237] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0238] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0239] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0240] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining a credit strategy, characterized in that: The method comprises: Use causal artificial intelligence technology to conduct causal analysis on agricultural production data and obtain causal analysis results; Using digital twin technology to simulate the agricultural production process corresponding to the agricultural production data to obtain simulation results; Based on the causal analysis results and the simulation results, a target credit strategy is determined.

2. The method according to claim 1, characterized in that The agricultural production data include: meteorological data, pest data, historical yield data and agricultural policy data; The causal analysis of agricultural production data using causal artificial intelligence technology to obtain causal analysis results includes: A structural causal model is used to perform causal analysis on agricultural production data to obtain causal analysis results; the causal analysis results include the causal relationship between at least two of the following: meteorological conditions, pest occurrence, yield, agricultural policy, and unobserved factors.

3. The method according to claim 2, characterized in that The agricultural production data is analyzed using the structural causal model to obtain causal analysis results, including: Based on the agricultural production data, test whether the causal effect in the structural causal model can be identified; If the causal effect can be identified, determining a parameter corresponding to the causal effect; The causal analysis result is determined based on the parameter.

4. The method according to claim 1, characterized in that The digital twin technology is used to simulate the agricultural production process corresponding to the agricultural production data to obtain simulation results, including: Using digital twin technology to create a virtual environment corresponding to the agricultural production data; The agricultural production process is simulated by using the virtual environment to obtain simulation results. The agricultural production process includes: crop growth process, soil changes, and climate changes.

5. The method according to claim 4, characterized in that The agricultural production data includes: first sensor data and first monitoring data, and the method further includes: Acquire second sensor data and second monitoring data, where the second sensor data is collected after the first sensor data, and the second monitoring data is collected after the first monitoring data; Based on the second sensor data and the second monitoring data, the virtual environment is adjusted.

6. The method according to claim 1, characterized in that Determining a target credit strategy based on the causal analysis result and the simulation result includes: Determining the credit risks and credit returns corresponding to a plurality of candidate credit strategies based on the causal analysis results and the simulation results; The target credit strategy is determined from the plurality of candidate credit strategies based on the credit risk and the credit return.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Based on the simulation results, agricultural management adjustment strategies are determined.

8. A credit strategy determination device, characterized in that: The system comprises at least one processor, wherein the processor is used to call computer instructions in a memory so that the credit strategy determination device executes the credit strategy determination method according to any one of claims 1 to 7.

9. A computer program product, characterized in that The method comprises instructions which, when executed on a computer, cause the computer to execute the credit strategy determination method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that: Used to store a computer program, which, when executed on a computer, enables the computer to execute the credit strategy determination method according to any one of claims 1 to 7.

Images