Energy crop carbon emission reduction intelligent decision-making method and device based on digital twinning

Abstract

This invention discloses a digital twin-based intelligent decision-making method and device for carbon emission reduction of energy crops, belonging to the field of carbon emission technology. The method includes: acquiring multi-source big data on energy crops in a target area; constructing an agricultural virtual system based on the multi-source big data; performing topological structure analysis on a directed relational network model to determine a dynamic resilience index reflecting the disturbance resistance and recovery capability of the agricultural virtual system; determining one or more resilience control levers based on the evolution trend of the dynamic resilience index to generate multiple candidate control strategies for the resilience control levers; and determining the optimal control strategy based on the simulation results of inputting multiple candidate control strategies into a preset dynamic digital twin to monitor and warn of carbon emissions from energy crops. According to the embodiments of this application, the system resilience can be dynamically assessed, critical risks can be identified, and an intelligent management closed loop with self-sensing, self-analysis, and self-optimization capabilities can be formed to achieve early warning and decision-making.

Description

Technical Field

[0001] This invention belongs to the field of carbon emission technology, and specifically relates to a smart decision-making method and device for carbon emission reduction of energy crops based on digital twins. Background Technology

[0002] In recent years, with the improvement of agricultural observation and information systems, agricultural production processes have generated multi-source big data covering meteorological, soil, and management aspects, gradually building a foundation for agricultural big data applications at the regional scale. This type of big data includes at least daily meteorological element sequences, stratified soil parameters, and crop management parameters, providing key inputs for crop yield assessment and regional agricultural management. It can also serve as yield constraints or evaluation criteria in land use and planting structure decisions. Due to the complex and dynamic interactions between water, energy, and food, they collectively constitute a water-energy-food linkage system with trade-offs and synergistic characteristics. Currently, research on the impact of climate change on agriculture still has certain limitations. Existing studies often focus on single-dimensional yield prediction or static calculations due to insufficient systematic analysis, making it difficult to effectively reveal the nonlinear interactions between subsystems within the system. Furthermore, limited by the lack of dynamic feedback mechanisms, existing models using multi-dimensional big data struggle to capture the evolutionary trajectory of crop physiological development processes and environmental control targets over time under future scenarios. How to achieve synergistic evolution simulation and comprehensive early warning of the water-energy-food-carbon linkage in response to climate change is an urgent problem to be solved.

[0003] While some technologies have been developed to couple crop models with life cycle assessments for stress warnings and carbon emission detection, enabling multi-dimensional assessments, existing methods typically rely on quantified water and carbon levels and simulated scenarios based on historical data or big data to output risk levels. These methods cannot provide timely and proactive systematic intervention for warnings. By the time a warning is issued, the system may already be close to or on an irreversible deterioration trajectory, resulting in a lag and preventing true early warning from being achieved. Summary of the Invention

[0004] This application provides a digital twin-based intelligent decision-making method and device for carbon emission reduction in energy crops. It can dynamically assess system resilience, identify critical risks, and achieve real-time adaptive closed-loop control for agricultural water-energy-grain-carbon system optimization, forming an intelligent management closed loop with self-sensing, self-analysis, and self-optimization capabilities, enabling early warning and decision-making.

[0005] On the one hand, embodiments of this application provide a smart decision-making method for carbon emission reduction of energy crops based on digital twins, the method including: Acquire multi-source big data on energy crops in the target area, including historical data and real-time IoT monitoring data; Based on the aforementioned multi-source big data, an agricultural virtual system is constructed. The agricultural virtual system includes a water-energy-food-carbon process model of the physiological processes of energy crops and a directed correlation network model for characterizing the evolution of the internal coupling pattern of the system. A topology analysis was performed on the directed relational network model to determine a dynamic resilience index that reflects the disturbance resistance and recovery capability of the agricultural virtual system. Based on the evolution trend of the dynamic resilience index, one or more resilience control levers are determined to generate multiple candidate control strategies for the resilience control levers. Based on the simulation results of inputting the multiple candidate control strategies into a preset dynamic digital twin, the optimal control strategy is determined to monitor and provide early warning of carbon emissions from the energy crop.

[0006] Optionally, based on the multi-source big data, a directed relational network model is constructed, including: The time-series variables output by the water-energy-carbon quantification model are used as the initial network nodes; The convergent cross-mapping algorithm is used to analyze the causal relationship between any two node time series, and the normalized causal strength is used as the initial weight of the corresponding directed edge. Based on the initial weights and the initial network nodes, a directed relational network model is constructed.

[0007] Optionally, the historical data includes daily meteorological observation data and state variables of the energy crop, wherein the state variables are used to characterize the parameters affecting the carbon emissions of the energy crop; After constructing the directed association network model based on the initial weights and the initial network nodes, the method further includes: Obtain the global climate model corresponding to the target region; Extract simulation data from the global climate model and the historical data corresponding to the historical periods; Using a preset statistical algorithm, quantitatively evaluate the characterization indicators of the global climate model on the meteorological capabilities of the target region; Climate patterns that meet the similarity criteria are identified as the climate driving fields corresponding to the agricultural virtual system. The directed correlation network model is adjusted based on the climate driving field to obtain the driving model; The topology analysis of the directed network model to determine the dynamic resilience index reflecting the disturbance resistance and recovery capability of the agricultural virtual system includes: The real-time topology of the driving model is analyzed to determine the dynamic resilience index that reflects the disturbance resistance and recovery capability of the agricultural virtual system.

[0008] Optionally, the analysis of the real-time topology of the driving model to determine the dynamic resilience index reflecting the disturbance resistance and recovery capability of the agricultural virtual system includes: A topological analysis is performed on the driving model to obtain its topological features. Based on the aforementioned topological features and the output state of the process model, the state domain of the agricultural system is obtained through analysis. According to the state domain, preset sub-indicators are activated from preset indices and assigned different weight values. The preset sub-indicators include topological indices, dynamic indices and functional performance indices. Plot the change trajectory of the preset sub-indicators under different preset scenarios within a preset time window; Based on the change trajectory, an integral calculation is performed to obtain a dynamic resilience index, which includes a bearing capacity index and a recovery capacity index.

[0009] Optionally, determining one or more resilience control lever points based on the evolution trend of the dynamic resilience index to generate multiple candidate control strategies for the resilience control lever points includes: The key variable set of the agricultural virtual system under continuous time slices is mapped to a preset system dynamic state space to form a state evolution trajectory; In the system dynamics state space, the critical state domains of historical system performance are identified by clustering algorithm, and the distance vectors from the current system state point to the boundaries of each critical domain are calculated. When the minimum distance is less than the warning radius, based on the historical cause data of the critical domain, the source is traced back to the most relevant set of nodes in the directed relational network model that are in a preset state. The set of nodes is determined as the resilience control lever point; Based on the aforementioned resilience control lever points and corresponding critical patterns, strategy rules aimed at breaking or mitigating such specific critical evolutions are extracted from a preset strategy knowledge graph to generate multiple candidate control strategies.

[0010] Optionally, before determining one or more resilience control lever points based on the evolution trend of the dynamic resilience index to generate multiple candidate control strategies for the resilience control lever points, the method further includes: The radiation factor of the energy crop is obtained, which includes light energy utilization efficiency, total photosynthetically active radiation, and multiple correction functions. Based on the radiation factor and the preset driving simulation formula, radiation driving simulation results are generated; Based on the radiation-driven simulation results and the first preset transpiration formula, the evapotranspiration of the energy crop is determined. Based on the energy crop evapotranspiration and the second preset transpiration formula, water-driven simulation results are generated. Based on the radiation-driven simulation results and the moisture-driven simulation results, the evolution trend of the dynamic toughness index is plotted.

[0011] Optionally, the step of determining the optimal control strategy based on the simulation results of inputting the multiple candidate control strategies into a preset dynamic digital twin, in order to monitor and provide early warning of carbon emissions from the energy crop, includes: For each candidate control strategy, in the agricultural virtual system, a multi-dimensional efficiency vector is extracted from the simulation results corresponding to the candidate control strategy. The multi-dimensional efficiency vector includes at least: expected carbon emission reduction, expected carbon sink increment, resource consumption cost, system resilience improvement, and impact on grain yield. In a pre-defined multidimensional target space, the multidimensional effectiveness vectors of all candidate control strategies are non-dominated and sorted to select a subset of strategies that constitute the Pareto optimal frontier. No strategy in the subset of strategies can be better at a certain target without harming at least one other target. Receive user decision preferences for each dimension, including priority weight settings or setting constraint boundaries; Based on the decision preference, on the Pareto optimal frontier, an optimal compromise strategy that satisfies the decision preference is automatically calculated and determined as the final optimal control strategy through a preset decision method.

[0012] On the other hand, embodiments of this application provide a digital twin-based intelligent decision-making device for carbon emission reduction in energy crops, the device comprising: The acquisition module is used to acquire multi-source big data of energy crops in the target area, including historical data and real-time IoT monitoring data. A construction module is used to construct an agricultural virtual system based on the multi-source big data. The agricultural virtual system includes a water-energy-food-carbon process model of the physiological processes of energy crops and a directed correlation network model for characterizing the evolution of the internal coupling pattern of the system. The analysis module is used to perform topology analysis on the directed relational network model and determine dynamic resilience indicators that reflect the disturbance resistance and recovery capabilities of the agricultural virtual system. The identification module is used to determine one or more resilience control lever points based on the evolution trend of the dynamic resilience index, so as to generate multiple candidate control strategies for the resilience control lever points. The determination module is used to determine the optimal control strategy based on the simulation results of inputting the multiple candidate control strategies into the dynamic digital twin, so as to monitor and warn of the carbon emissions of the energy crop.

[0013] In another aspect, embodiments of this application provide an electronic device, the device comprising: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins as described in the first aspect.

[0014] In another aspect, embodiments of this application provide a computer storage medium storing computer program instructions, which, when executed by a processor, implement the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins as described in the first aspect.

[0015] In another aspect, embodiments of this application provide a computer program product in which the instructions are executed by the processor of an electronic device, causing the electronic device to perform the intelligent decision-making method for carbon reduction of energy crops based on digital twins as described in the first aspect.

[0016] The intelligent decision-making method and apparatus for carbon emission reduction of energy crops based on digital twins in this application can construct an agricultural virtual system by acquiring multi-source big data of energy crops in a target area. This agricultural virtual system can simulate the energy conversion process of energy crops under different environments. Then, through topological analysis in the agricultural virtual system, control lever points are identified, and multiple candidate control strategies are generated for the control lever points. The optimal control strategy is obtained by simulation and deduction in a preset dynamic digital twin to achieve intelligent decision-making for carbon emission reduction of farm crops. This realizes precise monitoring and proactive early warning of carbon emissions from energy crops, effectively overcoming the limitations of traditional methods in terms of insufficient systematic analysis, lack of dynamic feedback, and delayed early warning, and providing timely, proactive, and comprehensive decision support for carbon emission reduction of energy crops. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings. Figure 1 This is a flowchart illustrating an embodiment of the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins provided in this application. Figure 2 This is a schematic diagram of the structure of a digital twin-based intelligent decision-making device for carbon reduction of energy crops, provided in another embodiment of this application. Figure 3 This is a schematic diagram of the structure of an electronic device provided in another embodiment of this application. Detailed Implementation

[0018] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0019] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0020] To address the problems of existing technologies, this application provides a method and apparatus for intelligent decision-making on carbon reduction of energy crops based on digital twins. In this application, an agricultural virtual system can be constructed by acquiring multi-source big data on energy crops in a target area. This agricultural virtual system can simulate the energy conversion process of energy crops under different environments. Then, through topological analysis within this agricultural virtual system, control lever points are identified, and multiple candidate control strategies are generated for these lever points. These strategies are then simulated and deduced in a preset dynamic digital twin to obtain the optimal control strategy, thereby achieving intelligent decision-making for carbon reduction of farm crops. This enables precise monitoring and proactive early warning of carbon emissions from energy crops, effectively overcoming the limitations of traditional methods in terms of insufficient systematic analysis, lack of dynamic feedback, and delayed early warning. It provides timely, proactive, and comprehensive decision support for carbon reduction of energy crops.

[0021] The following section first introduces the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins, as provided in the embodiments of this application.

[0022] Figure 1 This illustration shows a flowchart of a digital twin-based intelligent decision-making method for carbon emission reduction in energy crops, provided in one embodiment of this application. Figure 1As shown, the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins may include S101-S105: S101, acquire multi-source big data on energy crops in the target area.

[0023] In this embodiment, the multi-source big data may include historical data and real-time IoT monitoring data, forming a dataset describing the status of energy crops and environmental conditions in the target area. As an optional implementation, energy crop-related data in the target area can be collected periodically by manual methods, such as farmers or researchers manually recording historical information like meteorological data, soil conditions, and crop growth stages, and then manually inputting this data into a database. Simultaneously, IoT devices can be manually inspected to read and record real-time monitoring data. Alternatively, daily meteorological observation data and energy crop status variables can be obtained by deploying fixed weather stations, soil sensors, and conducting regular manual sampling and analysis. After collection, this data is integrated into a data storage system through manual input or batch processing import.

[0024] S102, based on multi-source big data, constructs an agricultural virtual system.

[0025] In this embodiment, the agricultural virtual system may include a water-energy-food-carbon process model of the physiological processes of energy crops and a directed correlation network model for characterizing the evolution of the internal coupling pattern of the system. A simplified water-energy-food-carbon process model can be manually established using expert experience and historical statistical data. This model may employ linear or simple nonlinear equations to describe the relationships between the elements. Simultaneously, a static directed correlation network model is constructed by manually analyzing the correlations between variables, where the weights of nodes and edges are set based on empirical judgment. Alternatively, statistical regression methods can be used to fit the key parameters in the water-energy-food-carbon process model based on collected multi-source big data. Simultaneously, an initial directed correlation network model is constructed by manually drawing or based on simple correlation analysis, where each variable is considered a node, and known or assumed influence relationships between variables are represented as directed edges.

[0026] S103. Perform topology analysis on the directed relational network model to determine the dynamic resilience index used to reflect the disturbance resistance and recovery capability of the agricultural virtual system.

[0027] The topological characteristics of a directed network model can be calculated by manually examining the node connections, such as node degree and network density. Then, based on these static topological characteristics and expert experience, the system's resilience and recovery capabilities can be qualitatively evaluated and transformed into an empirical resilience index. Alternatively, basic graph theory algorithms, such as calculating network connectivity and shortest paths, can be used to analyze the topological structure of the directed network model. Subsequently, by manually setting thresholds or consulting pre-defined tables, these topological characteristics can be mapped to one or more static resilience indices, reflecting the system's stability under a specific structure.

[0028] S104 identifies one or more resilience control levers based on the evolution trend of dynamic resilience indicators, in order to generate multiple candidate control strategies for the resilience control levers.

[0029] One approach is to manually observe charts of dynamic resilience indicators over time to identify periods of significant decline or fluctuation. Then, combining this with an understanding of the agricultural virtual system's operating mechanism, the operator identifies key internal variables or processes that may have caused these changes and designates them as resilience control levers. Subsequently, based on empirical knowledge, the operator proposes control measures targeting these levers as candidate control strategies. Another approach is to set simple rules, such as triggering an alert when the dynamic resilience indicator falls below a preset threshold. The system then searches a predefined set of key variables related to the current alert state according to a pre-defined rule base and designates these variables as resilience control levers. Next, it selects strategies related to these levers from a pre-defined list of strategies encompassing general agricultural management practices as candidate control strategies.

[0030] S105, based on the simulation results of inputting multiple candidate control strategies into a preset dynamic digital twin, determines the optimal control strategy to monitor and warn of carbon emissions from energy crops.

[0031] One approach is to use each candidate control strategy as input, run multiple simulations in an agricultural virtual system, and record key outputs such as carbon emissions and yield of energy crops after each simulation. Then, by manually comparing the simulation results under different strategies, the strategy that performs best in carbon reduction is selected as the optimal control strategy. Another approach is to simulate each candidate control strategy in the agricultural virtual system and manually extract key performance indicators from the simulation results, such as carbon emissions and resource consumption. Subsequently, by manually assigning weights, these indicators are weighted and summed to obtain a comprehensive score. Finally, the strategy with the highest comprehensive score is selected as the optimal control strategy, and carbon emissions from energy crops are monitored and monitored based on this strategy.

[0032] In this embodiment, an agricultural virtual system can be constructed by acquiring multi-source big data on energy crops in a target area. This agricultural virtual system can simulate the energy conversion process of energy crops under different environments. Then, through topological analysis in this agricultural virtual system, control lever points are identified, and multiple candidate control strategies are generated for the control lever points. The optimal control strategy is obtained by simulation and deduction in a preset dynamic digital twin, so as to realize intelligent decision-making for carbon emission reduction of farm crops. This achieves precise monitoring and proactive early warning of carbon emissions from energy crops, effectively overcoming the limitations of traditional methods in terms of insufficient systematic analysis, lack of dynamic feedback, and delayed early warning, and providing timely, proactive, and comprehensive decision support for carbon emission reduction of energy crops.

[0033] In some other embodiments, S102 may include: The time-series variables output by the water-energy-carbon quantification model were used as the initial network nodes; The convergent cross-mapping algorithm is used to analyze the causal relationship between any two node time series, and the normalized causal strength is used as the initial weight of the corresponding directed edge. A directed relational network model is constructed based on the initial weights and initial network nodes.

[0034] In the embodiments of this application, when constructing an agricultural virtual system, it is necessary to accurately characterize the evolution of the coupling pattern within the system, especially how to effectively construct a directed relational network model based on multi-source big data to reflect the complex interactions in the physiological processes of energy crops. This is crucial for subsequent topology analysis and determination of resilience indicators.

[0035] Specifically, when constructing the directed relational network model, the time-series variables output by the water-energy-carbon quantification model are first used as initial network nodes. The water-energy-carbon quantification model is a crucial component of the agricultural virtual system, simulating the dynamic changes in water, energy, and carbon during the growth of energy crops. This model outputs a series of time-varying parameters, such as soil moisture content, crop photosynthetic rate, respiration rate, biomass, carbon storage, transpiration, and energy use efficiency. These time-series variables represent key states and processes within the agricultural virtual system. Using them as initial network nodes means treating these key parameters as fundamental elements of the network, which may have complex causal relationships. For example, by running the water-energy-carbon quantification model at preset time steps (such as daily or weekly), time-series data of its key physiological and ecological parameters can be obtained. This time-series data will serve as the basic data units for constructing the directed relational network, with each time series corresponding to a node in the network.

[0036] Secondly, a convergent cross-mapping algorithm is employed to analyze the causal relationship between any two node time series, and the normalized causal strength is used as the initial weight of the corresponding directed edge. Convergent Cross-Mapping (CCM) is a method for detecting causal relationships in complex systems, particularly suitable for nonlinear and non-stationary time series data. This algorithm determines causal relationships by observing the "mapping ability" of one system state to another, rather than simple statistical correlation. In practice, for any two initial network nodes (i.e., two time series variables), their corresponding historical time series data are extracted. For each pair of nodes A and B, cross-mapping from A to B and from B to A are performed respectively. That is, the state space of node A is constructed from its time series, and then it is observed whether this state space can accurately predict the time series of node B, and vice versa. If the state space of A can accurately predict B, it indicates that A has a causal influence on B. The CCM algorithm outputs a "mapping skill" or "prediction accuracy" metric, which can be used as a quantification of causal strength. Generally, as the embedding dimension increases, if the prediction accuracy converges, the causal relationship is established. The calculated causal strength values ​​are then normalized, for example, scaled to between 0 and 1, to eliminate differences in causal strength units and make them more suitable as weights for network edges. Normalization methods can include max-min normalization or Z-score normalization. If node A has a causal relationship with node B, a directed edge is added from A to B in the network, and the normalized causal strength value is assigned to this edge as its initial weight.

[0037] Finally, based on the initial weights and initial network nodes, a directed relational network model is constructed. This integrates the previously identified nodes and the directed causal relationships between them, along with their strengths, to form a complete network structure. This directed relational network model intuitively demonstrates the interaction patterns between various key physiological and ecological processes within the agricultural virtual system, including which factors influence which factors and the strength of these influences. In specific implementation, the time-series variables output from all water-energy-carbon quantification models are used as the set of network nodes. For any pair of nodes, if a causal relationship is found through convergent cross-mapping algorithm analysis, and the causal strength is determined, a directed edge from cause to effect is added to the network, and its weight is set to the corresponding causal strength value. This directed relational network model can be represented in a computer using data structures such as adjacency matrices and adjacency lists.

[0038] When constructing the directed correlation network model in the agricultural virtual system, this approach no longer relies solely on empirical judgment or simple statistical correlation. Instead, it innovatively uses key time-series variables output from the water-energy-carbon quantification model as network nodes and employs a convergent cross-mapping algorithm to accurately identify and quantify the complex nonlinear causal relationships between these nodes. This method overcomes the limitations of traditional correlation analysis, which cannot distinguish between causality and correlation, and struggles to handle nonlinear dynamic systems. By normalizing causal strength as the initial weight of directed edges, the constructed directed correlation network model can more realistically and precisely characterize the evolution of the water-energy-carbon coupling pattern in the physiological processes of energy crops. This provides a more solid and accurate foundation for subsequent topological analysis of the network, determination of dynamic resilience indicators, and identification of resilience regulation levers, thereby significantly improving the scientific rigor and effectiveness of the entire intelligent decision-making method for carbon emission reduction in energy crops, making the decision results more reliable and instructive.

[0039] In other embodiments, after constructing the directed relational network model based on the initial weights and initial network nodes, the method further includes: Obtain global climate models corresponding to the target region; Extract simulation data from global climate models and historical data corresponding to historical periods; Using pre-defined statistical algorithms, quantitatively assess the characterization indicators of global climate models on meteorological capabilities within the target region; Climate patterns that meet the similarity criteria are identified as the climate driving fields corresponding to the agricultural virtual system; The directed correlation network model is adjusted based on the climate driving field to obtain the driving model; Topological analysis of directed relational network models was performed to determine dynamic resilience indicators reflecting the disturbance resistance and recovery capabilities of agricultural virtual systems, including: The real-time topology of the driving model is analyzed to determine the dynamic resilience index that reflects the disturbance resistance and recovery capability of the agricultural virtual system.

[0040] Specifically, historical data includes not only daily meteorological observations such as temperature, precipitation, solar radiation, and humidity, but also state variables of energy crops. These state variables characterize key parameters affecting carbon emissions from energy crops, such as biomass, leaf area index, soil moisture content, carbon sequestration rate, and respiration rate. This detailed historical data provides a foundation for subsequent model construction and validation.

[0041] To incorporate forward-looking information on future climate change, this application obtains global climate models corresponding to the target region. Global climate models are complex numerical models based on physical laws and Earth system processes, capable of simulating the long-term evolution of the Earth's climate system. These models are typically published by international climate research institutions and contain climate prediction data under different future scenarios. After obtaining the global climate models, it is necessary to extract simulation data from the global climate models for the corresponding historical periods. This step aims to provide a benchmark for subsequent climate model evaluations; by comparing the model's simulation results for historical periods with actual observational data, the reliability of the model can be assessed.

[0042] Subsequently, pre-defined statistical algorithms are used to quantitatively evaluate the indicators that characterize the meteorological capabilities of the global climate model within the target region. These pre-defined statistical algorithms may include, but are not limited to, techniques such as bias correction, spatial downscaling, and temporal downscaling to make the global climate model output more applicable to regional scales. Characterization indicators may include root mean square error (RMSE), correlation coefficient, and bias, used to measure the degree of agreement between model simulation results and actual observational data on meteorological characteristics. For example, quantile mapping or linear scaling methods can be used to correct bias in the model output.

[0043] Climate models that meet the similarity criteria are identified as the climate driving fields corresponding to the agricultural virtual system. The similarity criteria refer to the climate model meeting a preset threshold requirement in its ability to characterize the meteorological conditions of the target area after evaluation using the aforementioned statistical algorithm. Models that meet these criteria are selected as climate driving fields, meaning that the model can reliably provide future climate scenarios as external input for the agricultural virtual system's simulation and prediction.

[0044] Building upon this, the directed correlation network model is adjusted based on the climate driving field to obtain the driving model. The climate driving field provides predictive information on future climate change, such as future trends in temperature, precipitation, and CO2 concentration. Based on this information, the nodes (such as crop physiological process variables) and edges (causal relationship strength) in the directed correlation network model can be dynamically adjusted. For example, when predicting future extreme drought scenarios, the weights and connection strengths of nodes related to water use efficiency and transpiration can be adjusted based on the predicted precipitation reduction from the climate driving field. This allows the model to more accurately reflect the dynamic response of energy crops to carbon emission reduction under future climate scenarios, forming a "driving model" that can be driven by climate change.

[0045] Finally, the real-time topology of the driving model is analyzed to determine dynamic resilience indices reflecting the disturbance resistance and recovery capabilities of the agricultural virtual system. The driving model is a dynamically changing system, and its topology adjusts in real time with changes in the climate driving field. By analyzing the real-time topology of the driving model, we can reveal how the coupling relationships between various elements (water, energy, food, and carbon) within the agricultural virtual system change under different climate scenarios. This analysis can more accurately determine the system's disturbance resistance (i.e., the system's ability to maintain a stable state under disturbance) and recovery capability (i.e., the speed and extent to which the system recovers to a normal state after disturbance) in the face of climate disturbances, thus obtaining more forward-looking and robust dynamic resilience indices.

[0046] By analyzing the real-time topology of the driving model, the determined dynamic resilience index can more accurately reflect the agricultural virtual system's ability to resist and recover from future climate disturbances, thereby significantly improving the long-term accuracy and robustness of intelligent carbon emission reduction decisions and providing a solid foundation for formulating more forward-looking and adaptive regulation strategies.

[0047] In other embodiments, the real-time topology of the driving model is analyzed to determine dynamic resilience indicators that reflect the disturbance resistance and recovery capabilities of the agricultural virtual system, including: Perform topological analysis on the driving model to obtain its topological characteristics; Based on the topological characteristics and the output state of the process model, the state domain of the agricultural system is analyzed and obtained. Based on the state domain, preset sub-indicators are activated from preset indices and assigned different weight values. The preset sub-indicators include topological indices, dynamic indices, and functional performance indices. Plot the change trajectory of preset sub-indicators under different preset scenarios within a preset time window; The dynamic resilience index is obtained by performing integral calculations based on the change trajectory. The dynamic resilience index includes the bearing capacity index and the recovery capacity index.

[0048] In this embodiment, firstly, a topological analysis is performed on the driving model to obtain its topological characteristics. Topological analysis aims to reveal the connection relationships and distribution characteristics of nodes and edges in the directed relational network model (i.e., the driving model), thereby understanding the network's inherent organizational rules, key nodes, information flow paths, and overall robustness. Specifically, various graph theory algorithms can be used for analysis, such as centrality analysis (calculating the degree centrality, betweenness centrality, proximity centrality, eigenvector centrality, etc. of nodes) to identify key nodes or information hubs in the network; community structure detection to identify tightly connected subgroups in the network; path analysis to evaluate the propagation efficiency of information or matter in the network; and connectivity analysis to evaluate the network's robustness when nodes or edges are removed. Through these analyses, a series of quantitative topological characteristics can be obtained, which are the foundation for understanding the system's behavior and resilience.

[0049] Secondly, based on the topological characteristics and the output states of the process model, the state domain of the agricultural system is analyzed. The state domain of an agricultural system refers to the region in a multidimensional space that describes the overall operational state of the virtual agricultural system at a specific point in time or under a specific context. It integrates the structural characteristics of the network (topological features) and the physiological and ecological processes within the system (the output states of the water-energy-food-carbon process model), thus providing a more comprehensive characterization of the system's current state. In practical implementation, key topological features obtained from topological analysis (such as centrality indices of key nodes, network density, clustering coefficients, etc.) can be integrated with key state variables output by the water-energy-food-carbon process model (such as crop biomass, soil moisture, carbon flux, energy use efficiency, etc.) to form a high-dimensional state vector. For this high-dimensional state vector, dimensionality reduction techniques such as Principal Component Analysis (PCA) or t-SNE can be used to map it to a low-dimensional space. Then, through clustering algorithms or expert knowledge, different state regions, such as healthy states, stress states, and critical states, can be identified in the dimensionality-reduced state space; each region constitutes a state domain.

[0050] Next, based on the state domain, preset sub-indicators are activated from a pre-defined index library and assigned different weight values. These preset sub-indicators include topological, dynamic, and functional performance indices. This step aims to dynamically select and weight different evaluation indices according to the specific state domain of the system, in order to more accurately and flexibly assess the system's resilience. The pre-defined index library can contain a variety of potential resilience evaluation indices, such as topological indices (e.g., network connectivity, modularity), dynamic indices (e.g., system response speed to disturbances, recovery time), and functional performance indices (e.g., carbon sequestration efficiency, food yield stability). The activation and weighting mechanism can be based on rules (e.g., activating water-related indices if the system is in a "water stress" state domain), machine learning models, or expert knowledge, thereby ensuring the relevance assessment's specificity and effectiveness.

[0051] Subsequently, the trajectory of preset sub-indicators under different preset scenarios within a preset time window is plotted. This step simulates a series of possible disturbance events (such as extreme weather, pest outbreaks, etc.) as preset scenarios, and records the numerical changes of each preset sub-indicator (topology indicator, dynamic indicator, functional performance indicator) in real time within the set time window. The recorded sub-indicator values ​​are plotted as curves or multi-dimensional trajectory graphs over time to intuitively display the system's response and recovery path under different scenarios, providing a data foundation for subsequent resilience quantification.

[0052] Finally, integration is performed based on the change trajectory to obtain dynamic resilience indices, which include withstand capability and resilience indices. Integration is an effective method for quantifying a system's performance under disturbances. By integrating the change trajectories of sub-indices, the system's performance throughout the entire disturbance-recovery process can be comprehensively evaluated. The withstand capability index is typically quantified by calculating the degree or duration of performance degradation in the initial stage of a disturbance, such as calculating the area integral of a functional performance index deviating from its normal level. The resilience index is quantified by calculating the time or recovery rate required for the system to recover from a degraded performance state to a normal or acceptable level, such as calculating the time required to recover from the lowest performance point to a preset threshold. Numerical integration methods such as Riemann integrals and trapezoidal integrals can be used for area calculation.

[0053] By performing topological analysis on the driving model and combining it with the output states of the process model, the state domain of the agricultural system can be comprehensively characterized, providing a solid foundation for resilience assessment. Based on this, according to the system's state domain, different weights are dynamically activated and assigned to topological, kinetic, and functional performance indicators, making resilience assessment more targeted and flexible, avoiding a "one-size-fits-all" approach. Furthermore, by plotting the changes of these sub-indices under different scenarios and performing integral calculations, the abstract concept of resilience can be quantified into specific endurance and resilience indicators, clearly revealing the system's resistance and recovery capabilities in the face of disturbances. This multi-dimensional and dynamic resilience assessment method enables intelligent decision-making on carbon reduction for energy crops to be based on a more accurate and comprehensive understanding of system resilience, thereby improving the effectiveness and robustness of decision-making.

[0054] In some other embodiments, S104 may include: The key variable set of the agricultural virtual system under continuous time slices is mapped to the preset system dynamic state space to form the state evolution trajectory; In the dynamic state space of this system, the critical state domains of historical system performance are identified by clustering algorithm, and the distance vectors from the current system state point to the boundaries of each critical domain are calculated. When the minimum distance is less than the warning radius, based on the historical causal data of the critical domain, the source is traced back to the most relevant set of nodes in the directed relational network model that are in a preset state. The node set is identified as a resilience control lever point; Based on the resilience regulation lever points and corresponding critical patterns, policy rules aimed at breaking or mitigating such specific critical evolutions are extracted from a pre-defined policy knowledge graph to generate multiple candidate regulation policies.

[0055] In this embodiment, a multi-dimensional state point is formed by combining the numerical values ​​of key variables in the system (e.g., carbon emissions, crop growth indicators, and water and fertilizer use efficiency output from the water-energy-food-carbon process model of energy crop physiological processes, as well as the activity or connection strength of key nodes in the directed relational network model) at continuous time points. These state points are connected in a pre-defined system dynamics state space to form the trajectory of the system's evolution over time, which helps to intuitively observe the system's dynamic behavior and potential critical transitions. In practice, dimensionality reduction techniques such as Principal Component Analysis (PCA) and t-SNE (t-distributed stochastic neighbor embedding) can be used to map the high-dimensional set of key variables to a low-dimensional (e.g., two-dimensional or three-dimensional) state space for visualization and analysis. Continuous time slices can be set as daily, weekly, or monthly data snapshots according to the actual monitoring frequency and analysis needs.

[0056] In the system's dynamic state space, clustering algorithms are used to identify the critical state domains of historical system performance, and the distance vectors from the current system state point to the boundaries of each critical domain are calculated. The purpose of this step is to analyze the state points in historical data where system performance declined or collapsed, and to categorize them using clustering algorithms to form regions representing "criticality," thus understanding when and where the system exhibited undesirable or high-risk states in the past. Calculating the distance from the current system state point to these critical domains quantifies the degree of risk currently faced by the system. Clustering algorithms such as K-means, DBSCAN, and hierarchical clustering can be used. The critical state domains of historical system performance can be defined based on preset thresholds; for example, when carbon emissions exceed a certain warning value, energy crop yields fall below a certain expected value, or system resilience indicators fall below a certain safety line, they can be considered critical states. The distance vectors can be calculated using Euclidean distance, Mahalanobis distance, etc., to reflect the proximity of the current state to different critical domains.

[0057] When the minimum distance is less than the warning radius, the process traces back to the most relevant set of nodes in the directed network model that are in a preset state, based on historical causal data of the critical domain. When the system approaches a critical state, it is necessary to quickly identify the key factors causing this state. This step, through reverse tracing, traces from the macroscopic critical state domain to specific nodes in the microscopic directed network model, thereby locating the root cause of the problem. The warning radius is an empirical value or a threshold obtained through training with historical data. Historical causal data can include external disturbances (such as extreme weather events), internal management measures (such as fertilizer application and irrigation), and the specific states of each node in the directed network model that led to the occurrence of the critical state. Reverse tracing can utilize causal reasoning algorithms (such as Granger causality and structural equation modeling) or rule bases based on expert knowledge to analyze which nodes in the directed network model (e.g., specific crop physiological processes, water resource utilization, energy input, etc.) exhibit abnormal or critical changes when the critical domain forms, and combine this with their importance in the network (such as centrality indicators) to determine the most relevant set of nodes. The preset state can refer to abnormal node activity, changes in connection strength, etc.

[0058] Identifying the set of nodes as leverage points for resilience regulation aims to clearly identify which specific system components or processes are key points that can be intervened to influence system resilience. This step is a direct confirmation of the results of the previous source tracing step. The identified set of nodes may include key physiological process nodes affecting carbon emissions, water and fertilizer management nodes, pest and disease management nodes, etc. These nodes are considered leverage points because of their critical position in the network and their significant impact on the system state.

[0059] Based on resilience regulation levers and corresponding critical patterns, strategy rules aimed at breaking or mitigating such specific critical evolutions are extracted from a pre-defined strategy knowledge graph to generate multiple candidate regulation strategies. Once a lever is identified, targeted intervention measures need to be developed. This step utilizes a pre-defined knowledge base to associate the lever with specific regulation strategies, thereby generating a series of feasible candidate solutions. The pre-defined strategy knowledge graph can be a structured database storing the relationships between various agricultural management practices, technical measures, and the system nodes and critical patterns they may affect. For example, if "soil moisture deficit" is identified as a lever, and the corresponding critical pattern is "crop wilting leading to decreased carbon absorption," the knowledge graph might recommend strategies such as "precision irrigation" and "mulch cultivation." Strategy rule extraction can be based on rule matching, semantic search, or machine learning recommendation algorithms. The generated candidate regulation strategies should be diverse to facilitate subsequent simulation, deduction, and optimization.

[0060] By identifying historical critical state domains and calculating the distance between the current state and the critical domain, early warning of system risks is achieved. More importantly, when the system approaches a critical state, it can accurately trace back to the most relevant specific nodes in the directed relational network model based on historical causal data, thereby identifying the key regulatory levers that truly affect the system's resilience. Based on this, combined with a pre-set strategy knowledge graph, it can efficiently and specifically extract strategy rules aimed at breaking or mitigating specific critical evolutions, generating diverse candidate regulatory strategies. This significantly improves the accuracy and effectiveness of intelligent decision-making for carbon emission reduction in energy crops, avoids blind or inefficient interventions, ensures the achievement of carbon emission reduction targets for energy crops, and enhances the agricultural system's resilience and recovery capabilities.

[0061] In some other embodiments, prior to S104, the method may further include: The radiation factor of energy crops is obtained, which includes light energy utilization efficiency, total photosynthetically active radiation, and multiple correction functions. Based on the radiation factor and the preset driving simulation formula, radiation-driven simulation results are generated. Based on radiation-driven simulation results and the first preset transpiration formula, the evapotranspiration of energy crops is determined. Based on the evapotranspiration of energy crops and a second preset transpiration formula, water-driven simulation results are generated. Based on radiation-driven and moisture-driven simulation results, the evolution trend of dynamic resilience index is plotted.

[0062] In this embodiment, the radiation factors of energy crops are first obtained. Radiation factors are key environmental parameters affecting photosynthesis, growth and development, and water use efficiency of energy crops, and their accurate acquisition is the foundation for subsequent simulations. Radiation factors typically include light use efficiency (LUE), photosynthetically active radiation (PAR), and several correction functions. LUE characterizes the efficiency with which energy crops convert absorbed light energy into biomass and can be estimated through field observations, remote sensing inversion, or empirical models. PAR refers to the solar radiation energy that plants can utilize for photosynthesis and can be obtained through meteorological station data, satellite remote sensing data, or radiative transfer models. The correction functions are used to correct for changes in LUE caused by environmental factors such as temperature, moisture, and carbon dioxide concentration. For example, these functions may include temperature correction functions and water stress correction functions, which are typically constructed based on crop physiological and ecological principles.

[0063] As an example, a geographic detector can be used to identify driving factors that significantly affect yield, and the q-index can be used to measure the explanatory power of each factor on the spatial differentiation of yield. The influencing factors identified in this step are mainly used to set the key parameters for crop yield prediction in step three: These are set using the following formula: Where q represents the explanatory power of environmental factors on the spatial distribution of output; h represents the number of spatial strata or zones of environmental impact factors; N h N and N represent the number of samples within layer h and the total number of samples in the study area, respectively; and These represent the variance of production within layer h and the total variance of production in the study area, respectively.

[0064] Subsequently, radiation-driven simulation results are generated based on radiation factors and preset driving simulation formulas. Preset driving simulation formulas are mathematical models or algorithms used to quantify the impact of radiation factors on the physiological processes of energy crops. For example, this formula can be a model based on the photosynthetic mechanism, such as a photosynthetic model for C3 or C4 plants. By inputting light energy utilization efficiency and total photosynthetically active radiation, combined with a correction function, the gross primary productivity (GPP) or net primary productivity (NPP) of the energy crop is calculated, thus obtaining the carbon fixation or biomass accumulation rate under radiation-driven conditions, which serves as the radiation-driven simulation result.

[0065] The following will use extreme weather events as an example to illustrate this point: For extreme weather events, it can be based on Tfut,y Calculate the number of days of high-temperature heatwave (HW).

[0066] Among them, T fut,y,d T represents the simulated daily temperature value on day d of year y; D is the total number of days in the crop growing season; T threshold The high temperature threshold is determined by taking historical baseline observation data T. obs,hist The 95th quantile of the same sequence; f(x) is an indicator function that takes the value 1 when the condition is met and the continuous duration is ≥5 days, otherwise it takes the value 0.

[0067] Based on the daily precipitation P of the output fut,y Extreme precipitation (R95p) is defined as the total amount of precipitation exceeding the historical precipitation threshold, and the calculation formula is as follows.

[0068] Among them, P fut,y,d This represents the simulated daily precipitation value for day d in year y; P95 is the historical baseline period observed daily precipitation sequence P obs,hist The 95th quantile of precipitation samples with ≥1 mm in China and Japan. This indicator is used to characterize the risk of heavy precipitation that may lead to increased nitrogen and phosphorus loss from farmland and an increase in indirect greenhouse gas emissions (Gind) under future scenarios.

[0069] Based on P fut,y The consecutive drought days (CDD) are defined as the maximum number of consecutive days in a year with daily precipitation less than 1 mm.

[0070] When CDD increases, it will affect crop yield through water stress on the one hand, and the reduction of effective precipitation Pe will lead to an increase in irrigation water demand BW, thereby increasing irrigation energy consumption emissions Girri.

[0071] The identified extreme events are input into subsequent crop physiological models to capture the nonlinear damage of extreme high temperatures to crop pollen viability and grain filling rate (for the DSSAT model) or the process of water stress causing crop stomatal closure or even permanent wilting (for the AquaCrop model).

[0072] Based on this, the evapotranspiration of energy crops is determined using radiation-driven simulation results and a first pre-defined evapotranspiration formula. The first pre-defined evapotranspiration formula is a model used to calculate the evapotranspiration of energy crops. It takes radiation-driven simulation results (such as net radiation and temperature) as input, combines other meteorological parameters (such as humidity and wind speed) and crop physiological parameters, and estimates the total amount of water lost by crops through transpiration and evaporation from the soil surface. For example, this formula can be the Penman-Monteith formula, the Priestley-Taylor formula, etc. These formulas can comprehensively consider energy balance and aerodynamic factors, thereby accurately estimating evapotranspiration.

[0073] As an example, a calibrated DSSAT model was used to simulate crop biomass accumulation under specific scenarios, driven by high-resolution future climate-driven field data based on specific shared socioeconomic pathways (SSP) and typical concentration pathways (RCP). The core of the DSSAT model lies in calculating photosynthetic product accumulation under different climate stresses.

[0074] In the formula, B represents the total dry matter biomass accumulated by the crop throughout its growth period; RUE represents the crop's light energy use efficiency; IPAR represents the total photosynthetically active radiation intercepted by the crop canopy; f(T), f(CO2), and f(W) represent the correction functions for the effects of temperature change, carbon dioxide concentration fertilizer effect, and water stress on crop growth, respectively. Simultaneously, the model outputs the daily actual evapotranspiration ET. c Used for subsequent water demand calculations.

[0075] Next, water-driven simulation results are generated based on energy crop evapotranspiration and a second pre-defined transpiration formula. The second pre-defined transpiration formula is a model used to further transform energy crop evapotranspiration into quantitative data reflecting the impact of water on crop physiological processes. For example, this formula can be a soil moisture balance model, simulating dynamic changes in soil moisture content by inputting parameters such as evapotranspiration, rainfall, and soil type, thereby assessing the degree of water stress on crops; or it can be a water use efficiency model, linking evapotranspiration to biomass accumulation. These simulation results can reflect the limiting effect of water conditions on crop growth and carbon cycling.

[0076] As an example, high-resolution future climate-driven field data based on SSP-RCP can be used to drive the AquaCrop model to simulate the biomass accumulation process of crops under specific scenarios. The AquaCrop model emphasizes the separation of transpiration and evaporation, which can better explain the water-saving mechanism achieved by water-scarce areas such as the North China Plain through "reducing bare ground cover" or "inhibiting stomatal conductance".

[0077] In the formula, The normalized crop water productivity parameter; Tr i ET represents the actual crop transpiration on day i. 0,i This represents the reference crop evapotranspiration for the corresponding date. This path focuses on characterizing the nonlinear limiting effect of water stress on yield.

[0078] In the AquaCrop model, ET is calculated from ET0. c The specific formula is as follows.

[0079] In the formula, ET c Indicates actual crop evapotranspiration; Tr represents actual crop transpiration, which in AquaCrop is driven by both canopy development and water stress; E represents actual soil evaporation; K s K represents the water stress coefficient, which is the coefficient of water stress when soil moisture is sufficient. s =1, when soil drought triggers stress, 0≤K s <1, used to reduce transpiration. K cb,x It represents the maximum standard crop transpiration coefficient, which represents the transpiration capacity of a crop under conditions of complete canopy coverage and no stress. K represents the corrected canopy cover, a core variable in AquaCrop, representing the photosynthetically active area, and changes dynamically with the reproductive period. e It represents the soil evaporation coefficient, which depends on the proportion of bare soil not covered by the canopy and the moisture content of the topsoil.

[0080] Finally, based on the radiation-driven and water-driven simulation results, the evolution trends of dynamic resilience indices were plotted. By using the aforementioned radiation-driven and water-driven simulation results as key inputs, and combining them with the water-energy-food-carbon process model and the directed correlation network model in the agricultural virtual system, the changes in dynamic resilience indices (such as endurance and resilience indices) over time can be calculated and visualized more comprehensively and accurately. This method, which comprehensively considers environmental driving factors, allows the evolution trends of dynamic resilience indices to more realistically reflect the disturbance resistance and recovery capabilities of the agricultural virtual system under different environmental disturbances.

[0081] This in-depth consideration of environmental drivers makes the evolution trend of the dynamic resilience index more realistic, significantly improving the accuracy of assessing the disturbance resistance and recovery capabilities of agricultural virtual systems. Based on this, subsequent identification of resilience regulation levers will be more precise, enabling the generation of more targeted and effective candidate regulation strategies, ultimately achieving accurate monitoring and early warning of carbon emissions from energy crops.

[0082] In some other embodiments, S105 may include: For each candidate control strategy, in the agricultural virtual system, a multi-dimensional efficiency vector is extracted from the simulation results corresponding to the candidate control strategy. The multi-dimensional efficiency vector includes at least: expected carbon emission reduction, expected carbon sink increment, resource consumption cost, system resilience improvement, and impact on grain yield. In a pre-defined multidimensional target space, the multidimensional effectiveness vectors of all candidate control strategies are non-dominated and sorted to select a subset of strategies that constitute the Pareto optimal frontier. No strategy in the subset of strategies can be better at one target without harming at least one other target. Receive user decision preferences for various dimensions, including priority weight settings or setting constraint boundaries; Based on decision preferences, at the Pareto optimal frontier, an optimal compromise strategy that satisfies the decision preferences is automatically calculated and determined as the final optimal control strategy through a preset decision method.

[0083] In this embodiment, after simulating each candidate control strategy, the system extracts a series of key performance indicators from the simulation results to form a multi-dimensional performance vector. For example, the expected carbon emission reduction quantifies the total carbon emission reduction projected after the strategy is implemented; the expected carbon sink increment assesses its improvement in carbon absorption capacity; resource consumption costs reflect the inputs required to implement the strategy, such as water, fertilizer, and energy; system resilience improvement measures the impact of the strategy on the agricultural virtual system's resistance to disturbances and its recovery capabilities; and the impact on grain yield assesses the potential impact of the strategy on energy crop yields. The extraction of these indicators ensures a comprehensive and quantitative evaluation of each strategy.

[0084] To effectively handle these multidimensional and potentially conflicting performance indicators, this application employs a non-dominated ranking method. This method compares and ranks the performance vectors of all candidate regulatory strategies within a predefined multidimensional objective space. Its core idea is to identify strategies that cannot be completely outperformed by other strategies across all objectives. A strategy is said to dominate the other strategy if it is superior to or at least not inferior to another strategy across all objectives, and strictly superior to it in at least one objective. Through non-dominated ranking, a subset of strategies constituting the Pareto optimal frontier can be selected. Any strategy in this subset possesses the property that if it wants to perform better on one objective, it must necessarily perform worse on at least one other objective. This provides decision-makers with a set of optimal trade-offs, avoiding suboptimal choices.

[0085] Given that practical decision-making often requires consideration of specific application scenarios and the decision-maker's subjective judgment, this application further designs a mechanism for receiving users' decision preferences across various dimensions. Users can express their preferences in multiple ways, such as by setting priority weights for different effectiveness dimensions (e.g., expected carbon emission reductions, resource consumption costs, and the impact on food production), clearly indicating which objectives are more important; or by setting specific constraints, such as requiring expected carbon emission reductions to reach a certain minimum threshold, or requiring resource consumption costs to not exceed a certain upper limit. The introduction of user preferences makes the decision-making process more flexible and targeted.

[0086] Based on this, the system will automatically calculate and determine an optimal trade-off strategy that satisfies user-inputted decision preferences on the pre-selected Pareto optimal frontier using a preset decision-making method. For example, a weighted sum method can be used, applying user-defined weights to the multidimensional efficacy vector of each strategy on the Pareto frontier, and then selecting the strategy with the largest weighted sum as the optimal solution; or an ideal point method can be used, calculating the distance between each strategy and the user-defined ideal target point, and selecting the strategy with the closest distance. In this way, the system can intelligently select the final control strategy that best meets the user's actual needs and priorities from a series of optimal trade-offs.

[0087] By extracting multi-dimensional performance vectors, each strategy is comprehensively and quantitatively evaluated, avoiding the potential bias of a single indicator. The construction of non-dominated ranking and the Pareto optimal front clearly demonstrates the trade-offs between different objectives, providing decision-makers with a truly optimal set of choices. More importantly, by incorporating user preferences for each dimension, this application enables the intelligent decision-making system to fully consider subjective needs and priorities in practical applications. This allows for the intelligent calculation and determination of an "optimal compromise strategy" that satisfies both multi-objective optimization principles and user-specific needs at the Pareto optimal front. This significantly enhances the scientific rigor, practicality, and operability of the decision-making process, ensuring that the selected strategy is not only theoretically optimal but also maximizes the balance between multiple objectives, including expected carbon emission reductions, expected carbon sink increases, resource consumption costs, system resilience improvement, and the impact on food production. This provides more accurate and intelligent decision support for carbon emission reduction in energy crops.

[0088] Based on the digital twin-based intelligent decision-making method for carbon emission reduction of energy crops provided in the above embodiments, this application also provides a specific implementation of a digital twin-based intelligent decision-making device for carbon emission reduction of energy crops. Please refer to the following embodiments.

[0089] First see Figure 2 The energy crop carbon reduction intelligent decision-making device 200 based on digital twin provided in this application embodiment may include: The acquisition module 201 is used to acquire multi-source big data of energy crops in the target area. The multi-source big data includes historical data and real-time IoT monitoring data. Module 202 is used to construct an agricultural virtual system based on multi-source big data. The agricultural virtual system includes a water-energy-food-carbon process model of the physiological processes of energy crops and a directed correlation network model for characterizing the evolution of the internal coupling pattern of the system. Analysis module 203 is used to perform topology analysis on the directed relational network model and determine dynamic resilience indicators that reflect the disturbance resistance and recovery capabilities of the agricultural virtual system. The identification module 204 is used to identify one or more resilience control levers based on the evolution trend of dynamic resilience indicators, so as to generate multiple candidate control strategies for the resilience control levers. The determination module 205 is used to determine the optimal control strategy based on the simulation results of inputting multiple candidate control strategies into a dynamic digital twin, so as to monitor and warn of carbon emissions from energy crops.

[0090] As an alternative implementation, the construction module 202 can also be used for: The time-series variables output by the water-energy-carbon quantification model were used as the initial network nodes; The convergent cross-mapping algorithm is used to analyze the causal relationship between any two node time series, and the normalized causal strength is used as the initial weight of the corresponding directed edge. A directed relational network model is constructed based on the initial weights and initial network nodes.

[0091] As an optional implementation, historical data includes daily meteorological observation data and state variables of energy crops, whereby the state variables are used to characterize parameters affecting carbon emissions from energy crops; the construction module 202 can also be used for: Obtain global climate models corresponding to the target region; Extract simulation data from global climate models and historical data corresponding to historical periods; Using pre-defined statistical algorithms, quantitatively assess the characterization indicators of global climate models on meteorological capabilities within the target region; Climate patterns that meet the similarity criteria are identified as the climate driving fields corresponding to the agricultural virtual system; The directed correlation network model is adjusted based on the climate driving field to obtain the driving model; Topological analysis of directed relational network models was performed to determine dynamic resilience indicators reflecting the disturbance resistance and recovery capabilities of agricultural virtual systems, including: The real-time topology of the driving model is analyzed to determine the dynamic resilience index that reflects the disturbance resistance and recovery capability of the agricultural virtual system.

[0092] As an alternative implementation, the construction module 202 can also be used for: Perform topological analysis on the driving model to obtain its topological characteristics; Based on the topological characteristics and the output state of the process model, the state domain of the agricultural system is analyzed and obtained. Based on the state domain, preset sub-indicators are activated from preset indices and assigned different weight values. The preset sub-indicators include topological indices, dynamic indices, and functional performance indices. Plot the change trajectory of preset sub-indicators under different preset scenarios within a preset time window; The dynamic resilience index is obtained by performing integral calculations based on the change trajectory. The dynamic resilience index includes the bearing capacity index and the recovery capacity index.

[0093] As an alternative implementation, the identification module 204 can also be used for: The key variable set of the agricultural virtual system under continuous time slices is mapped to the preset system dynamic state space to form the state evolution trajectory; In the dynamic state space of this system, the critical state domains of historical system performance are identified by clustering algorithm, and the distance vectors from the current system state point to the boundaries of each critical domain are calculated. When the minimum distance is less than the warning radius, based on the historical causal data of the critical domain, the source is traced back to the most relevant set of nodes in the directed relational network model that are in a preset state. The node set is identified as a resilience control lever point; Based on the resilience regulation lever points and corresponding critical patterns, policy rules aimed at breaking or mitigating such specific critical evolutions are extracted from a pre-defined policy knowledge graph to generate multiple candidate regulation policies.

[0094] As an alternative implementation, the identification module 204 can also be used for: The radiation factor of energy crops is obtained, which includes light energy utilization efficiency, total photosynthetically active radiation, and multiple correction functions. Based on the radiation factor and the preset driving simulation formula, radiation-driven simulation results are generated. Based on radiation-driven simulation results and the first preset transpiration formula, the evapotranspiration of energy crops is determined. Based on the evapotranspiration of energy crops and a second preset transpiration formula, water-driven simulation results are generated. Based on radiation-driven and moisture-driven simulation results, the evolution trend of dynamic resilience index is plotted.

[0095] As an alternative implementation, the determining module 205 can be used for: For each candidate control strategy, in the agricultural virtual system, a multi-dimensional efficiency vector is extracted from the simulation results corresponding to the candidate control strategy. The multi-dimensional efficiency vector includes at least: expected carbon emission reduction, expected carbon sink increment, resource consumption cost, system resilience improvement, and impact on grain yield. In a pre-defined multidimensional target space, the multidimensional effectiveness vectors of all candidate control strategies are non-dominated and sorted to select a subset of strategies that constitute the Pareto optimal frontier. No strategy in the subset of strategies can be better at one target without harming at least one other target. Receive user decision preferences for various dimensions, including priority weight settings or setting constraint boundaries; Based on decision preferences, at the Pareto optimal frontier, an optimal compromise strategy that satisfies the decision preferences is automatically calculated and determined as the final optimal control strategy through a preset decision method.

[0096] Figure 3 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0097] An electronic device may include a processor 301 and a memory 302 storing computer program instructions.

[0098] Specifically, the processor 301 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0099] Memory 302 may include mass storage for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. In one instance, memory 302 may include removable or non-removable (or fixed) media, or memory 302 may be non-volatile solid-state memory. Memory 302 may be internal or external to the integrated gateway disaster recovery device.

[0100] In one instance, memory 302 may be read-only memory (ROM). In one instance, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0101] Memory 302 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the intelligent decision-making method for carbon reduction of energy crops based on digital twins according to the first aspect of this disclosure.

[0102] The processor 301 reads and executes computer program instructions stored in the memory 302 to achieve... Figure 1 The embodiment shown presents an intelligent decision-making method for carbon emission reduction of energy crops based on digital twins.

[0103] In one example, the electronic device may also include a communication interface 303 and a bus 304. For example, Figure 3 As shown, the processor 301, memory 302, and communication interface 303 are connected through bus 304 and complete communication with each other.

[0104] The communication interface 303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0105] Bus 304 includes hardware, software, or both, that couples components of an electronic device together. For example, and not as a limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 304 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0106] This electronic device can execute the energy crop carbon emission reduction intelligent decision-making method based on digital twins as described in this application embodiment, thereby achieving a combination of... Figures 1-2 The description includes a digital twin-based intelligent decision-making and device for carbon reduction in energy crops.

[0107] Furthermore, in conjunction with the digital twin-based intelligent decision-making method for carbon reduction of energy crops in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the digital twin-based intelligent decision-making methods for carbon reduction of energy crops in the above embodiments.

[0108] In an optional embodiment, in conjunction with the digital twin-based intelligent decision-making method for carbon reduction of energy crops in the above embodiments, this application embodiment can provide a computer program product to implement it. The instructions in the computer program product are executed by the processor of an electronic device, enabling the electronic device to implement any of the digital twin-based intelligent decision-making methods for carbon reduction of energy crops in the above embodiments.

[0109] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0110] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0111] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0112] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0113] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A smart decision-making method for carbon emission reduction of energy crops based on digital twins, characterized in that, include: Acquire multi-source big data on energy crops in the target area, including historical data and real-time IoT monitoring data; Based on the multi-source big data, an agricultural virtual system is constructed. The agricultural virtual system includes a water-energy-food-carbon process model of the physiological processes of energy crops and a directed correlation network model for characterizing the evolution of the internal coupling pattern of the system. A topology analysis was performed on the directed relational network model to determine a dynamic resilience index that reflects the disturbance resistance and recovery capability of the agricultural virtual system. Based on the evolution trend of the dynamic resilience index, one or more resilience control levers are determined to generate multiple candidate control strategies for the resilience control levers. Based on the simulation results of inputting the multiple candidate control strategies into a preset dynamic digital twin, the optimal control strategy is determined to monitor and provide early warning of carbon emissions from the energy crop.

2. The method according to claim 1, characterized in that, Based on the aforementioned multi-source big data, a directed relational network model is constructed, including: The time-series variables output by the water-energy-carbon quantification model are used as the initial network nodes; The convergent cross-mapping algorithm is used to analyze the causal relationship between any two node time series, and the normalized causal strength is used as the initial weight of the corresponding directed edge. Based on the initial weights and the initial network nodes, a directed relational network model is constructed.

3. The method according to claim 2, characterized in that, The historical data includes daily meteorological observation data and state variables of the energy crop, the state variables being used to characterize the parameters affecting the carbon emissions of the energy crop; After constructing the directed association network model based on the initial weights and the initial network nodes, the method further includes: Obtain the global climate model corresponding to the target region; Extract simulation data from the global climate model and the historical data corresponding to the historical periods; Using a preset statistical algorithm, quantitatively evaluate the characterization indicators of the global climate model on the meteorological capabilities of the target region; Climate patterns that meet the similarity criteria are identified as the climate driving fields corresponding to the agricultural virtual system. The directed correlation network model is adjusted based on the climate driving field to obtain the driving model; The topology analysis of the directed network model to determine the dynamic resilience index reflecting the disturbance resistance and recovery capability of the agricultural virtual system includes: The real-time topology of the driving model is analyzed to determine the dynamic resilience index that reflects the disturbance resistance and recovery capability of the agricultural virtual system.

4. The method according to claim 3, characterized in that, The analysis of the real-time topology of the driving model to determine the dynamic resilience index reflecting the disturbance resistance and recovery capability of the agricultural virtual system includes: A topological analysis is performed on the driving model to obtain its topological features. Based on the aforementioned topological features and the output state of the process model, the state domain of the agricultural system is obtained through analysis. According to the state domain, preset sub-indicators are activated from preset indices and assigned different weight values. The preset sub-indicators include topological indices, dynamic indices and functional performance indices. Plot the change trajectory of the preset sub-indicators under different preset scenarios within a preset time window; Based on the change trajectory, an integral calculation is performed to obtain a dynamic resilience index, which includes a bearing capacity index and a recovery capacity index.

5. The method according to claim 1, characterized in that, The step of determining one or more resilience control lever points based on the evolution trend of the dynamic resilience index, and generating multiple candidate control strategies for the resilience control lever points, includes: The key variable set of the agricultural virtual system under continuous time slices is mapped to a preset system dynamic state space to form a state evolution trajectory; In the system dynamics state space, the critical state domains of historical system performance are identified by clustering algorithm, and the distance vectors from the current system state point to the boundaries of each critical domain are calculated. When the minimum distance is less than the warning radius, based on the historical cause data of the critical domain, the source is traced back to the most relevant set of nodes in the directed relational network model that are in a preset state. The set of nodes is determined as the resilience control lever point; Based on the aforementioned resilience control lever points and corresponding critical patterns, strategy rules aimed at breaking or mitigating such specific critical evolutions are extracted from a preset strategy knowledge graph to generate multiple candidate control strategies.

6. The method according to claim 1, characterized in that, Before determining one or more resilience control lever points based on the evolution trend of the dynamic resilience index to generate multiple candidate control strategies for the resilience control lever points, the method further includes: The radiation factor of the energy crop is obtained, which includes light energy utilization efficiency, total photosynthetically active radiation, and multiple correction functions. Based on the radiation factor and the preset driving simulation formula, radiation driving simulation results are generated; Based on the radiation-driven simulation results and the first preset transpiration formula, the evapotranspiration of the energy crop is determined. Based on the energy crop evapotranspiration and the second preset transpiration formula, water-driven simulation results are generated. Based on the radiation-driven simulation results and the moisture-driven simulation results, the evolution trend of the dynamic toughness index is plotted.

7. The method according to claim 1, characterized in that, The step of determining the optimal control strategy based on the simulation results of inputting the multiple candidate control strategies into a preset dynamic digital twin, in order to monitor and provide early warning of carbon emissions from the energy crop, includes: For each candidate control strategy, in the agricultural virtual system, a multi-dimensional efficiency vector is extracted from the simulation results corresponding to the candidate control strategy. The multi-dimensional efficiency vector includes at least: expected carbon emission reduction, expected carbon sink increment, resource consumption cost, system resilience improvement, and impact on grain yield. In a pre-defined multidimensional target space, the multidimensional effectiveness vectors of all candidate control strategies are non-dominated and sorted to select a subset of strategies that constitute the Pareto optimal frontier. No strategy in the subset of strategies can be better at a certain target without harming at least one other target. Receive user decision preferences for each dimension, including priority weight settings or setting constraint boundaries; Based on the decision preference, on the Pareto optimal frontier, an optimal compromise strategy that satisfies the decision preference is automatically calculated and determined as the final optimal control strategy through a preset decision method.

8. A digital twin-based intelligent decision-making device for carbon emission reduction in energy crops, characterized in that, The device includes: The acquisition module is used to acquire multi-source big data of energy crops in the target area, including historical data and real-time IoT monitoring data. A construction module is used to construct an agricultural virtual system based on the multi-source big data. The agricultural virtual system includes a water-energy-food-carbon process model of the physiological processes of energy crops and a directed correlation network model for characterizing the evolution of the internal coupling pattern of the system. The analysis module is used to perform topology analysis on the directed relational network model and determine dynamic resilience indicators that reflect the disturbance resistance and recovery capabilities of the agricultural virtual system. The identification module is used to determine one or more resilience control lever points based on the evolution trend of the dynamic resilience index, so as to generate multiple candidate control strategies for the resilience control lever points. The determination module is used to determine the optimal control strategy based on the simulation results of inputting the multiple candidate control strategies into the dynamic digital twin, so as to monitor and warn of the carbon emissions of the energy crop.

9. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the intelligent decision-making method for carbon emission reduction of energy crops based on digital twins as described in any one of claims 1-7.

Images