Carbon emission monitoring method and system based on agricultural carbon neutralization

Abstract

The invention relates to the technical field of carbon neutralization, and provides a carbon emission monitoring method and system based on agricultural carbon neutralization. The carbon emission monitoring method based on agricultural carbon neutralization comprises the following steps: constructing a multi-dimensional database based on carbon emission based on agricultural planting data of a current area, and determining dynamic characteristics of carbon emission; according to the carbon emission dynamic characteristics, constructing a dynamic analysis model, and determining a carbon neutralization efficiency index set; according to the carbon neutralization efficiency index set, a carbon emission thermodynamic diagram of the current area is constructed, the carbon emission thermodynamic diagram generates a key sensitive factor sorting sequence which is transformed in real time based on the carbon neutralization time-space evolution trend, and the color depth of the thermodynamic diagram is adjusted based on the key sensitive factor sorting sequence. According to the application, through multi-dimensional data fusion, dynamic coupling of respiration source analysis and metabolic entropy change, based on sensitive factor sorting and thermodynamic diagram rendering rules, meteorological sudden change response and edge calculation are combined, a monitoring result is displayed in real time through a thermodynamic diagram, and the carbon emission condition is determined.

Description

Technical Field

[0001] This application relates to the field of carbon neutralization technology, and particularly to a carbon emission monitoring method and system based on agricultural carbon neutralization. Background Art

[0002] Agricultural carbon neutralization is an important path to achieve sustainable development, and accurate monitoring of carbon emission dynamics is the key to optimizing agricultural management. Currently, agricultural carbon emission monitoring mostly relies on single-dimensional static data, such as soil respiration flux or farming activity records, lacking the integration of multi-source heterogeneous data in the crop growth and metabolism process. Traditional methods are difficult to capture the dynamic coupling relationship between root respiration, microbial activities and meteorological factors in real time, resulting in spatio-temporal deviations in the assessment of carbon emission intensity. In addition, existing technologies mostly use fixed-threshold models to generate heat maps, and cannot adaptively adjust visualization parameters according to the evolution of key sensitive factors. Especially in the correlation analysis of rhizosphere metabolic characteristics and farming behaviors, a dynamic monitoring system with feedback correction ability has not been established, restricting the precise regulation of carbon neutralization efficiency.

[0003] There are significant bottlenecks in the technical implementation of existing monitoring systems: Sensor networks are mostly limited to the collection of single physical indicators, such as carbon dioxide concentration or soil temperature and humidity, failing to integrate multi-modal data such as root electrophysiological signals, isotope tracing and three-dimensional impedance imaging, resulting in insufficient accuracy in the analysis of respiration sources. At the same time, traditional models rely on offline data training and are difficult to dynamically correct weights through transfer learning. There is a lag between the carbon emission prediction results and the actual metabolic entropy change. During the heat map generation process, disturbance factors such as meteorological mutations and farmer operations often cause the factor ranking and color mapping rules to be inaccurate. Lack of a distributed verification mechanism based on blockchain and federated learning makes it difficult to ensure the credibility and real-time nature of monitoring data. These defects restrict the dynamic optimization of the agricultural carbon neutralization path, and there is an urgent need to construct an integrated monitoring method that combines multi-dimensional perception, adaptive modeling and intelligent visualization. Summary of the Invention

[0004] This application proposes a carbon emission monitoring method and system based on agricultural carbon neutralization. In this application, multi-source data such as root electrophysiology, carbon dioxide isotope and three-dimensional impedance imaging are integrated. This application uses multi-dimensional data fusion, through the dynamic coupling of respiration source analysis and metabolic entropy change, optimizes the heat map rendering rules based on sensitive factor ranking and federated learning, combines meteorological mutation response and edge computing, and can display the monitoring results in real time through a heat map to determine the carbon emission status.

[0005] In a first aspect, an embodiment of this application provides a carbon emission monitoring method based on agricultural carbon neutralization, including: Construct a multi-dimensional database based on carbon emissions based on the agricultural planting data of the current region, and determine the dynamic characteristics of carbon emissions; Construct a dynamic analysis model based on the dynamic characteristics of carbon emissions to determine the carbon neutralization efficiency index set; Construct a carbon emission heat map for the current region based on the carbon neutralization efficiency index set. Among them, the carbon emission heat map generates a sorted sequence of key sensitive factors that change in real time based on the spatio-temporal evolution trend of carbon neutralization, and adjusts the color depth of the heat map based on the sorted sequence of key sensitive factors.

[0006] During the implementation of this application, when monitoring the carbon neutralization of crops in a certain region, first, through agricultural planting data, multi-dimensional data integration and dynamic analysis of a multi-dimensional database are carried out to judge the real-time tracking of the carbon emission data of crops and evaluate the real-time carbon emission efficiency, determine the dynamic characteristics of carbon emissions. Based on the dynamic carbon emission characteristics and the pre-set carbon neutralization efficiency evaluation index set, the spatio-temporal evolution of the current region will be realized in combination with the heat map, and the key sensitive factors affecting the carbon neutralization effect will be sorted, so as to intuitively observe the priority areas of carbon neutralization. Furthermore, by adjusting the color depth of different regions of the heat map, the carbon neutralization effect of the current region is reflected, improving the monitoring efficiency and response speed.

[0007] Combined with the first aspect, a three-dimensional sensing component is arranged in the rhizosphere layer of the agricultural crops in the current region; among them, the three-dimensional sensing component includes a bioelectric sensor, a carbon dioxide isotope induction sensor, and an electrical impedance tomography device; The bioelectric sensor is used to determine the active respiration area corresponding to the root electrophysiological signal in the rhizosphere layer; The carbon dioxide isotope induction sensor is used to determine the respiration source according to the spectral measurement value of the active respiration area; The electrical impedance tomography device is used to generate a three-dimensional impedance map of the root system according to the respiration unit.

[0008] During the implementation of this application, multi-modal in-situ monitoring of crop respiration metabolism is carried out. First, the combination of the bioelectric sensor and the carbon dioxide isotope sensor solves the problem that traditional single-point monitoring cannot distinguish the root active area from the microbial respiration source. Through the cross-modal association of electrophysiological signals and isotope spectra, the respiration source localization accuracy is improved, and it is actually improved to the centimeter level during implementation. Secondly, the electrical impedance tomography device can determine the respiration flux distribution of the three-dimensional structure of the crop roots in the current region through dynamic impedance gradients, solving the problem of insufficient characterization of root spatial heterogeneity by two-dimensional sensing technology. By jointly constructing a spatio-temporal holographic map of respiration metabolism, the analysis of the carbon exchange mechanism of the soil-crop system is upgraded from empirical speculation to quantitative visualization, realizing the visual display of crop carbon neutralization.

[0009] Combined with the first aspect, the three-dimensional impedance map further includes a respiration flux value, and the steps for obtaining the respiration flux value are as follows: When the sampling point of the root electrophysiological signal reaches the preset power density under the preset monitoring frequency band, a first instruction is triggered; In response to the first instruction, a first characteristic frequency band scan is performed on the active respiration area to determine effective respiration hotspots; According to the spatio-temporal correlation database and the effective respiration hotspots, spectral measurement is carried out. When the respiration contribution rate of the effective respiration hotspots in the spectral measurement is higher than the preset contribution value, a second instruction is triggered; According to the second instruction, the offset value of the respiration contribution rate and the impedance gradient data are spatio-temporally matched, and when the matching is successful, a respiration flux value is generated.

[0010] During the implementation of this application, through a multi-level trigger sensing mechanism, the problems of high false detection rate of plant respiration hotspots and lag in flux calculation in traditional technologies are solved. First, the preset frequency band power density threshold triggers the first instruction, realizing the active filtering of invalid noise signals. For example, mechanical vibration interference focuses the sensor computing power on the effective respiration area, improving the hotspot positioning efficiency. Secondly, the second instruction is triggered by the respiration contribution rate threshold. By combining the spatio-temporal matching of the impedance gradient and the isotope offset value, the instantaneous interference error caused by meteorological mutations is eliminated. For example, the impedance artifacts caused by rainwashing reduce the calculation error rate of the respiration flux value. Through signal triggering combined with feature scanning, then combined with contribution verification, and finally spatio-temporal fusion, a four-order optimization chain is formed, reducing the amount of redundant data processing while ensuring the monitoring accuracy, and can collect data on the respiration metabolism of concealed crops.

[0011] Combined with the first aspect, the construction of a multi-dimensional database based on carbon emissions to determine the dynamic characteristics of carbon emissions includes: Establish a spatio-temporal correlation database integrating planting type, intensity of farming activities, soil respiration flux and meteorological factors; Real-time calculate the crop root respiration entropy change data through the effective respiration hotspots, and combine the blockchain distributed storage of farmers' farming behavior data. Use the dynamic time warping algorithm to couple the sensor data with the farmers' operation logs to generate a spatio-temporal evolution map of carbon emission intensity with feedback on respiration metabolism characteristics.

[0012] During the implementation of this application, first, through a multi-source heterogeneous data spatio-temporal fusion mechanism, that is, through a spatio-temporal correlation database, discrete data such as planting types and farming activities are four-dimensionally grid-coded (longitude / latitude / time / depth) with soil respiration flux, and meteorological mutations are coupled with farming behaviors and weather, thereby improving the resolution of carbon emission traceability. For example, it is improved from the hectare level to the square meter level. Secondly, through blockchain distributed storage, combined with the technology of dynamic time warping, technical collaboration is achieved, so that the time baselines of sensor data and manual logs do not drift. Then, by aligning the respiratory entropy change data with farmer operation events, it can reach the millisecond level, and the time accuracy error of the carbon emission intensity map will also be directly reduced.

[0013] Combined with the first aspect, The construction of a dynamic analysis model according to the dynamic characteristics of carbon emissions includes: Adopt a soil respiration flux sensor synchronous construction feedback loop based on the dynamic characteristics of carbon emissions to generate a time series supervision; According to the transfer learning algorithm configured on the feedback loop, map agricultural planting data to a dynamic Bayesian network; According to the dynamic Bayesian network, determine the respiratory metabolic entropy weight model, and real-time correct the carbon emission intensity weight, generate a spatio-temporal carbon footprint gradient prediction matrix with a self-correction function of respiratory efficiency, and generate a dynamic analysis model according to the spatio-temporal carbon footprint gradient prediction matrix.

[0014] During the implementation of this application, first, a dynamic self-correction analysis model will be constructed to solve the problems of environmental response lag and data heterogeneity conflict in traditional carbon footprint prediction. The time series supervision feedback loop will perform on-line calibration of the soil respiration flux sensor, and then reduce the synchronization error between the sensor data and the root metabolic entropy change. The dynamic characteristics can be extracted in real time. Finally, the transfer learning-driven dynamic Bayesian network, for example: quickly adapt the rice paddy training model to the wheat field scenario to solve the problem of lack of cold start data in new planting areas. Finally, the respiratory entropy weight model corrects the weight matrix in real time through the thermodynamic efficiency gradient, and the prediction accuracy of the prediction matrix under mutation conditions such as heavy rain and drought will be significantly improved compared with the static model.

[0015] Combined with the first aspect, the determination of the carbon neutralization efficiency index set includes: According to the carbon footprint gradient prediction matrix, fuse the crop photosynthetic efficiency feature vector corresponding to the agricultural planting data and the soil respiration efficiency spectrum; Adopt a graph embedding pruning algorithm to construct a multi-dimensional index dimension reduction channel, generate a three-dimensional carbon neutralization level map with respiratory metabolic efficiency constraints through a spatio-temporal dynamic evolution equation, and adaptively adjust according to the index weight with the root respiration entropy change rate.

[0016] In the implementation process of this application, by constructing a dynamic index system under the constraint of crop respiration metabolism, the problems of only being able to use static weights in traditional carbon neutrality assessment and the inability to integrate multi-dimensional indicators are overcome. First, the fusion of the photosynthesis-respiration double efficiency spectrum can thermodynamically couple the carbon absorption of crops with soil emissions, improving the response sensitivity of the index set to scenarios such as sudden changes in light intensity and root damage. Second, the graph embedding pruning technology can combine manifold learning to extract the main control factors of the spatio-temporal evolution of carbon emissions, increasing the dimensionality compression rate of multi-dimensional indicators and retaining a high amount of metabolic characteristic information. Finally, the three-dimensional hierarchical map dynamically adjusts the weights based on the respiratory entropy change rate, shortening the update frequency of the carbon neutrality rating under drought stress.

[0017] Combined with the first aspect, constructing a carbon emission heat map of the current region according to the carbon neutrality efficiency index set, including Based on the three-dimensional hierarchical map of the carbon neutrality efficiency index set, a coupling feature evaluation model of the sensitive factor importance with a time sliding window mechanism is constructed by extracting the coupling characteristics of the farming behavior sequence of farmers and the soil respiration metabolism waveform through a spatio-temporal convolutional neural network; among them, the soil respiration metabolism waveform is collected by a soil respiration sensor array composed of bioelectric sensors in the three-dimensional perception component. Generate a factor ranking sequence that changes in real time with meteorological data based on the graph attention mechanism, and establish a dynamic mapping function of the color depth of the heat layer based on the non-linear relationship between the ranking position and the root respiratory entropy change rate.

[0018] In the implementation process of this application, a multi-modal dynamic visualization solution will be combined to solve the disconnection between the static factor weights and the real-time meteorological response in the generation of traditional heat maps. First, the spatio-temporal convolutional neural network combines the time sliding window mechanism to extract cross-scale coupling characteristics from the farming behavior sequence of crops and the soil respiration waveform in the current region, improving the recognition accuracy of sensitive factors. Second, based on the graph attention mechanism, real-time reconstruction of factor ranking for meteorological data is carried out to solve the problem that fixed ranking models cannot adapt to some special environments, such as heavy rain. Finally, based on the non-linear color mapping of the root respiratory entropy change rate, the rendering delay of the heat layer is compressed, for example: compressed from the minute level to the sub-second level, and the dynamic correspondence between the color depth and the metabolic intensity is also achieved.

[0019] Combined with the first aspect, constructing a carbon emission heat map of the current region according to the carbon neutrality efficiency index set further includes: During the generation of the sensitive factor ranking sequence, synchronously collect the vibration spectrum data of wearable farm tools and the metabolic characteristics of the soil respiration sensor array. Construct a distributed factor weight correction model based on federated learning, dynamically optimize the association rules between the ranking sequence and the color depth of the heat map based on the reinforcement learning algorithm, and implement a closed-loop verification of the farmer operation feedback data and the heat layer rendering parameters based on the blockchain smart contract.

[0020] During the implementation of this application, this application solves the dual problems of human operation interference and model drift in traditional heat map generation through a multi-source heterogeneous data closed-loop verification mechanism. First, during the process of determining key sensitive factors, the synchronous collection of soil metabolic characteristics is carried out, and the cross-media correlation analysis of the factors affecting crops, soil conditions, and the change of root respiration entropy is carried out, so as to improve the quantification accuracy of the impact of farming behavior on carbon emissions. Secondly, the distributed weight correction model driven by federated learning reduces the factor sorting error rate through the gradient encrypted exchange between multiple nodes while ensuring data privacy. Finally, the rendering parameter closed-loop verification chain constructed by the blockchain smart contract verifies the color depth adjustment of the carbon emission heat map through the operation feedback of multiple farmer nodes, which is more accurate and solves the problem that the traditional centralized model is vulnerable to single-point data pollution.

[0021] Combined with the first aspect, constructing the carbon emission heat map of the current region according to the carbon neutrality efficiency index set further includes: Input the key sensitive factor sorting sequence into the spatio-temporal adversarial generation network, and configure the correlation matrix of multi-spectral remote sensing data and farmer operation logs to generate a set of rendering parameters for the heat layer with respiratory metabolism efficiency constraints; According to the set of rendering parameters for the heat layer, establish a color depth gradient threshold dynamically adjusted by the meteorological factor mutation response function, and integrate the synchronous visualization of the metabolic characteristics of the carbon emission heat map; among them, the multi-spectral remote sensing data is collected by the drone device.

[0022] During the implementation of this application, multi-source data and edge intelligent rendering technology are integrated to solve the problems of remote sensing data lag and mobile terminal visualization distortion in traditional heat map generation. In the traditional technical solution, carbon neutrality mainly uses satellite determination, while this application combines drone technology. The spatio-temporal adversarial generation network optimizes the metabolic efficiency constraints of the rendering parameters of the carbon neutrality heat layer through the correlation matrix of farmer logs and multi-spectral data, and the utilization rate of drone remote sensing data will also be improved. The meteorological mutation response function will dynamically adjust the color depth threshold through the air pressure gradient change rate, so that the rendering delay of the carbon neutrality heat map will be compressed under extreme weather, for example: compressed from the minute level to the sub-second level.

[0023] In the second aspect, this application proposes a carbon emission monitoring system based on agricultural carbon neutrality, including: Carbon emission characteristic acquisition module: Based on the agricultural planting data of the current region, construct a multi-dimensional database based on carbon emissions to determine the dynamic characteristics of carbon emissions; Carbon neutrality index set determination module: Used to construct a dynamic analysis model according to the dynamic characteristics of carbon emissions to determine the carbon neutrality efficiency index set; Carbon Emission Visualization Module: It is used to construct a carbon emission heat map of the current region according to the carbon neutralization efficiency index set. Among them, the carbon emission heat map generates a sorted sequence of key sensitive factors that change in real time based on the spatio-temporal evolution trend of carbon neutralization, and adjusts the color depth of the heat map based on the sorted sequence of key sensitive factors.

[0024] During the implementation of this application, when monitoring the carbon neutralization of crops in a certain area, first, through agricultural planting data, multi-dimensional data integration and dynamic analysis of a multi-dimensional database are carried out to judge the real-time tracking of the carbon emission data of crops and evaluate the real-time carbon emission efficiency, determine the dynamic characteristics of carbon emissions. Based on the dynamic carbon emission characteristics and the pre-set carbon neutralization efficiency evaluation index set, the spatio-temporal evolution of the current region will be combined with the heat map to sort the key sensitive factors affecting the carbon neutralization effect, so as to intuitively observe the priority areas of carbon neutralization. Furthermore, by adjusting the color depth of different regions of the heat map, the carbon neutralization effect of the current region is reflected, improving the monitoring efficiency and response speed.

[0025] Other features and advantages of this application will be described in the following specification, and part of them will become obvious from the specification, or be understood by implementing this application. The objectives and other advantages of this application can be achieved and obtained through the structures specifically pointed out in the written specification, claims, and drawings.

[0026] The technical solutions of this application will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0027] The drawings are used to provide a further understanding of this application, and constitute a part of the specification. Together with the embodiments of this application, they are used to explain this application and do not constitute a limitation to this application. In the drawings: Figure 1 It is a flowchart of a carbon emission monitoring method based on agricultural carbon neutralization in an embodiment of this application; Figure 2 It is a flowchart of monitoring crops through a three-dimensional perception component in an embodiment of this application; Figure 3 It is a flowchart of calculating the respiratory flux value of a three-dimensional impedance spectrum in an embodiment of this application; Figure 4 It is a flowchart of the dynamic characteristics of carbon emissions in an embodiment of this application; Figure 5 It is a flowchart of constructing a dynamic analysis model in an embodiment of this application; Figure 6 It is a flowchart of the carbon neutralization efficiency index set in an embodiment of this application; Figure 7A logic diagram for constructing a carbon emission heat map of the current area in an embodiment of the present application; Figure 8 It is a logic diagram of closed-loop verification in an embodiment of the present application; Figure 9 It is a logic diagram for synchronous visualization of metabolic characteristics in the embodiments of this application; Figure 10 This is a system composition diagram of a carbon emission monitoring system based on agricultural carbon neutrality in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The preferred embodiments of the present application are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application and are not used to limit the present application.

[0029] Traditional carbon emission monitoring is mainly based on online monitoring methods, gas monitoring using non-dispersive infrared monitoring technology, high-sensitivity spectral monitoring using cavity ring-down spectroscopy technology, and gas monitoring using tunable diode laser absorption spectroscopy technology. They are mainly based on greenhouse gas monitoring, often combined with remote sensing and satellite technology to achieve three-dimensional carbon emission monitoring. However, for crops, which are densely planted and may also involve greenhouse technology, traditional technologies are often unable to achieve accurate carbon neutrality monitoring for crop monitoring. This application proposes a carbon emission monitoring method based on agricultural carbon neutrality combined with a thermal map to achieve the spatiotemporal evolution of the current area and intuitively observe the carbon neutral area. The specific real-time process of this application is as follows: Example 1

[0030] See also Figure 1 , Figure 1 This is a schematic flow chart of a carbon emission monitoring method based on agricultural carbon neutrality provided in an embodiment of the present application.

[0031] In its specific implementation, it first builds a multidimensional database based on carbon emissions based on the agricultural planting data of the current area to determine the dynamic characteristics of carbon emissions. This application focuses on multidimensional data fusion and dynamic feature analysis. In the crop planting area, multidimensional data such as soil respiration flux sensors, drone multispectral data and farmers' fertilization logs are integrated through a multidimensional database.

[0032] The carbon emission characteristics are judged through a multidimensional database. For example, when the sensor detects an abnormal increase in the root respiration entropy change rate in a certain area, the multidimensional database associates the meteorological data to find that the high temperature during the same period has caused a surge in soil microbial activity, and based on the farmer's log, the cause is locked in, such as excessive application of nitrogen fertilizer.

[0033] Furthermore, the carbon emission weight of this area is dynamically corrected. The carbon neutrality abnormal factors that should originally be classified as natural breathing sources are re-labeled as conventional factors or corrected factors in the current area because of the dominant emissions caused by agricultural activities. Then, through contact control measures, after farmers receive instructions and adjust the irrigation frequency, the carbon emission intensity of the corresponding area decreases. Therefore, the multi-dimensional data fusion of this application has a high-precision traceability ability for complex emission sources.

[0034] In its specific implementation, a dynamic analysis model will be constructed according to the dynamic characteristics of carbon emissions to determine the carbon neutrality efficiency index set; in this process, the carbon emission pattern of a certain historical period can be migrated to the newly built model, that is, the dynamic analysis model, through transfer learning. By setting different emergency conditions, the carbon neutrality efficiency index set is determined. For example, when a sudden heavy rainfall causes a sudden change in the soil impedance gradient, the dynamic Bayesian network of the dynamic analysis model autonomously adjusts the photosynthesis-respiration efficiency weight based on real-time sensor data, and then the carbon neutrality efficiency index is switched within a period of time. For example, it switches from a moderately inefficient state to a severely imbalanced state. Compared with the traditional fixed-threshold model that warns of emission risks in advance, farmers can install drainage channels accordingly, thereby weakening the carbon emission peak and enhancing the advantages of rapid response and self-correction to environmental mutations.

[0035] In its specific implementation, according to the carbon neutrality efficiency index set, a carbon emission heat map of the current area is constructed. The existing technology mainly reflects the static carbon emission situation in the form of a carbon emission pie chart, etc. Among them, the carbon emission heat map generates a sorting sequence of key sensitive factors that changes in real time based on the spatio-temporal evolution trend of carbon neutrality, and adjusts the color depth of the heat map based on the sorting sequence of key sensitive factors. In this process, according to the sorting sequence of sensitive factors, such as sensitive factors like temperature > root activity > irrigation frequency, etc., a heat map is generated, and the temperature-sensitive area is marked with a red gradient. If there is a cold snap that causes a sudden drop in the surface temperature of the crops, this application will synchronously update the sorting based on federated learning as: root activity > microbial biomass > temperature, and then the color depth of the heat map will be dynamically adjusted accordingly, highlighting the changes in the carbon emission hot spots in the root frost area. The planting management end of the crops applies a biochar insulation layer to the frost-damaged area of the crop roots according to the visualization result, so that the carbon sink efficiency of this area is restored to the pre-disaster level, thereby maintaining the carbon neutrality level. The dynamic mapping of the heat map parameters of this application can quickly provide rectification suggestions for the actual situation of complex scenarios.

[0036] Example 2: This application provides a method for monitoring crops through a three-dimensional sensing component. Refer to Figure 2 Specifically: In specific implementation, the bioelectric sensor is used to determine the active respiration area corresponding to the root electrophysiological signal in the rhizosphere layer; for example, the bioelectric sensor will detect a strong electric signal with an expected characteristic frequency in a certain area of the rhizosphere layer, and then automatically mark this area as an active respiration hotspot.

[0037] The carbon dioxide isotope induction sensor is used to identify the type of crop and determine the corresponding respiration source. For example, the isotope sensor synchronously measures the real-time value of carbon dioxide isotope in the soil, which is highly consistent with the respiration characteristic value of the crop. For the area misjudged as microbial respiration by the traditional single sensor, the corresponding type of crop can be determined through the root-dominated emission area of this application, and then a precise fertilization instruction for farmers can be generated. After precisely applying the bacterial fertilizer, the carbon emission in the current area can be reduced, and the precise distinction of the respiration source type can be achieved through multi-modal sensing.

[0038] In specific implementation, the electrical impedance tomography device of this application is used to generate a three-dimensional impedance map of the root system according to the respiration element. It will detect an abnormal impedance gradient on the surface soil in the current area, and the generated three-dimensional map shows a circular fracture zone existing in the root system at a preset depth. Then, combined with the isotope data, it is determined that the current area is an anaerobic or aerobic respiration hotspot, and a respiration abnormality determination based on three-dimensional impedance imaging is generated. Through the abnormality determination, the impedance gradient can return to the normal value range after the farmer digs a trench for drainage, so that the methane emission flux decreases. Therefore, three-dimensional impedance imaging has technical advantages in capturing hidden respiration abnormalities.

[0039] The bioelectric sensor of this application will judge whether there is an offset in the root active area, and the isotope sensor detects a mutation of oxygen isotope in the soil, indicating abnormal irrigation water penetration. The impedance imaging of this application will show that new capillary root clusters are formed in the root system in the offset area. When there is an offset, an alarm will be triggered and an irrigation optimization plan will be generated to improve the carbon neutralization effect. In actual implementation, the uniformity of root distribution is improved, verifying the early warning value of multi-sensor data fusion for carbon metabolism imbalance.

[0040] Example 3: This application proposes a method for calculating the respiration flux value of a three-dimensional impedance map. Refer to Figure 3 , specifically: In specific implementation, when the root electrophysiological signal reaches the sampling point of the preset power density under the preset monitoring frequency band, the first instruction is triggered. If the power density of the root electrophysiological signal is monitored to increase suddenly in a specific frequency band, the first instruction is triggered to perform a high-frequency scan of the target area to lock the elliptical respiratory hotspot of a preset diameter. Because traditional full-band scanning requires data processing, and this application only processes key data through threshold triggering, it will reduce computing power consumption. Then, through isotope verification, it is determined that the respiratory contribution rate of the hotspot area has increased, the error has decreased, and the effectiveness of the trigger mechanism for resource optimization, that is: in response to the first instruction, the first characteristic frequency band scan is performed on the active respiratory area to determine the effective respiratory hotspot; This application encounters extreme weather, such as sandstorms. The impedance gradient data of this application shows abnormal fluctuations in the surface layer if the isotope respiration contribution rate is lower than the preset value. The system determines that it is an impedance artifact caused by wind erosion and refuses to generate flux values. Compared with the traditional method of mistakenly increasing carbon emission values, this application controls the misjudgment rate within a stable threshold through time and space matching to ensure the reliability of monitoring in extreme weather.

[0041] If the power density of the root system is detected to exceed the standard for a certain duration in a specific frequency band, a secondary scan is triggered to find the underground respiratory hotspot. The time-space matching shows that the impedance gradient change rate in the area is highly correlated with the offset of the isotopic carbon dioxide. If it is determined to be a methane hotspot where anaerobic bacteria burst and emit, a second instruction will be generated, allowing farmers to dig aeration ditches accordingly. If the carbon emission flux value decreases after the aeration ditch is dug, the ability to detect deep metabolic events based on a multi-level trigger mechanism is realized.

[0042] Embodiment 4: This application proposes a method for determining the dynamic characteristics of carbon emissions, see Figure 4 , specifically: This application will first build a spatiotemporal correlation database integrating planting types, agricultural activity intensity, soil respiration flux and meteorological factors to achieve four-dimensional data fusion traceability: In this process, the spatiotemporal database of this application associates the possible events of the current crop area quota, such as rainstorm events, combine harvester operation logs and root respiration entropy change data, and dynamically corrects the carbon emission intensity of the crop area. If it is identified that the respiration flux in the soil compaction area caused by agricultural machinery rolling has increased sharply, it will generate the weekly average of carbon emissions in the crop area after farmers adopt strip farming, and ensure the four-dimensional grid coding's ability to accurately locate emissions caused by mechanical disturbances.

[0043] This application is for blockchain-DTW data alignment: In the specific implementation process, the rice transplanting time of the current area stored in the blockchain is aligned with the root activity jump time recorded by the sensor through the DTW algorithm, and it is determined that the time delay is due to the biological bacteria agent of the farmer's temporary treatment measures. The system of the present application will automatically mark this operation as a carbon sink optimization event, and generate a bacteria agent application-carbon absorption efficiency curve to promote the update of the regional carbon emission prediction model and improve the model goodness of fit.

[0044] The present application will also generate a warning for the spatio-temporal evolution map of the current crop area: In the current crop, the spatio-temporal map shows that there are abnormal pulse respiration intensities at some monitoring points during a certain period. If it is traced that there may be illegal midnight irrigation by farmers during this period, and then the irrigation blockchain deposit record is obtained, resulting in anaerobic respiration caused by a sudden drop in soil oxygen content. Accordingly, an irrigation time optimization plan is generated, and the evolution judgment of similar events in the current area is carried out through the equivalent plan, so as to reduce the incidence rate and make the map have the warning value of concealed human emissions.

[0045] Example 5: The present application proposes a way to construct a dynamic analysis model, see Figure 5 , specifically: During the construction process of the dynamic feature analysis model in the present application, real-time calibration is carried out for the feedback loop: In the specific implementation, if the feedback loop detects continuous emergencies of a certain sensor, for example, baseline drift occurs after rainfall, a calibration instruction will be automatically triggered. According to the transfer learning algorithm configured on the feedback loop, the agricultural planting data is mapped to the dynamic Bayesian network; the data of this node is reconstructed through the respiration entropy change gradient distribution of adjacent sensors to correct the output error. Compared with the traditional quarterly manual calibration, the present application improves the stability of the sensor network and ensures the continuity of carbon footprint prediction.

[0046] The present application will also perform cross-scenario adaptation for transfer learning: In the specific implementation, any training data will be migrated from the Bayesian network. Through the common features extracted by transfer learning, such as the root oxygen content-respiration efficiency curve, the model adaptation time is extremely shortened from the conventional. The newly formed model has a low prediction error in the first week. For example, the error of the traditional method is high, more than 37%. The present application can ensure the effectiveness of cross-crop knowledge transfer.

[0047] The present application will also perform anti-mutation optimization for the entropy weight model: In specific implementation, according to the dynamic Bayesian network, a respiratory metabolic entropy weight model is determined, and the carbon emission intensity weight is corrected in real time to generate a spatio-temporal carbon footprint gradient prediction matrix with a self-correcting function for respiratory efficiency. Then, based on the spatio-temporal carbon footprint gradient prediction matrix, a dynamic analysis model is generated. For example, when the entropy weight model monitors an abnormal increase rate of root respiration entropy, the cause of the abnormality is determined, and then the weight of the rainstorm impact factor is increased. The carbon emission peak can be predicted in advance through the prediction matrix, and then a control instruction is issued. The control instruction is the analysis result of the dynamic analysis model. If farmers enable the drainage system accordingly, the actual peak intensity is lower than the predicted value. For the detection of the respiratory state of plants in extreme environments, it has the ability to dynamically adjust the detection accuracy.

[0048] Example 6: This application provides a method for determining the carbon neutralization efficiency index set. Refer to Figure 6 , specifically: During the process of determining the carbon neutralization efficiency index set in this application, dual-efficiency spectrum fusion optimization will be carried out: First, when this application is implemented, according to the carbon footprint gradient prediction matrix, the photosynthetic efficiency eigenvector of agricultural planting data corresponding to the soil respiration efficiency spectrum will be fused; for example: the photosynthetic efficiency eigenvector, photosynthetic photon flux density and soil respiration efficiency spectrum are fused to determine the photoinhibition area caused by strong light at noon.

[0049] Then, a multi-dimensional index dimensionality reduction channel is constructed using the graph embedding pruning algorithm, and a three-dimensional carbon neutralization level map with respiratory metabolic efficiency constraints is generated through the spatio-temporal dynamic evolution equation. According to the dynamic adaptive adjustment of the index weight with the root respiration entropy change rate, the graph embedding pruning algorithm compresses the 12-dimensional index of the current area into 3-dimensional main factors. For example: light intensity / root temperature / impedance gradient, and the dynamic map will output the generation of the carbon neutralization level of the current area. For example: from "Carbon Neutralization Grade B" to "Carbon Neutralization Grade C". After farmers set up sunshade nets accordingly, the photosynthesis-respiration efficiency ratio is increased, and the utilization rate of light energy for dual-spectrum fusion is also improved.

[0050] This application uses the graph embedding pruning algorithm to construct a multi-dimensional index dimensionality reduction channel, and generates a three-dimensional carbon neutralization level map with respiratory metabolic efficiency constraints through the spatio-temporal dynamic evolution equation, that is, for manifold learning dimensionality reduction: The graph embedding pruning algorithm of this application extracts some main control factors from multiple original indexes: For example: soil oxygen content at a depth of 5 cm (weight 0.41), coefficient of variation of root diameter (0.33), and average daily photosynthetically active radiation (0.26).

[0051] Compared with the principal component analysis method, the three-dimensional channels generated by this application shorten the model training time, and the graph embedding pruning algorithm is adopted. Therefore, the prediction error is stable within the preset value under abnormal conditions such as rainy weather of crops, realizing the intelligent noise reduction ability for complex indicators.

[0052] This application dynamically adapts and adjusts according to the index weight with the change rate of root respiration entropy, that is, it adjusts in real time for the dynamic weight: When it is monitored that the change rate of root respiration entropy drops suddenly, the three-dimensional map of this application will automatically adjust the weight. For example, the weight of soil moisture content is increased from 0.18 to 0.63. The rating model after dynamic adjustment of this application will give an early warning of the degradation of the soil layer carbon emission function, and then adjust the carbon neutrality level. For example, it is restored from level D to level B within 6 hours. At this time, the carbon trading valuation increases.

[0053] Example 7: This application provides a method for constructing a carbon emission heat map of the current area. Refer to Figure 7 , specifically: This application will, based on the three-dimensional grade map of the carbon neutral efficiency index set, extract the coupling characteristics of the farmer's farming behavior sequence and the soil respiration metabolism waveform through a spatio-temporal convolutional neural network, and construct a sensitivity factor importance evaluation model with a time sliding window mechanism; among them, the soil respiration metabolism waveform is collected by a soil respiration sensor array composed of bioelectric sensors in the three-dimensional perception component. In this process, cross-scale feature extraction is realized. For example, ST-CNN analyzes the coupling relationship between the farmer's fertilization interval (once every 2 days) and the peak interval of the soil respiration waveform, and then identifies the factors affecting carbon emissions. For example, the pulsed nitrous oxide emission caused by excessive urea. Then, the importance of the sensitivity factor is determined through the time sliding window model. For example, the time period from 0 to 6 hours after fertilization is set as the sensitive period, and the color depth of this period in the heat map is set to 3.2 times the reference value. Then, the corresponding fertilization adjustment value is generated. After the farmer adjusts the fertilization strategy accordingly, it is judged whether the pulse emission intensity decreases. If it decreases, it indicates that this application can achieve the early warning effect of cross-scale feature extraction.

[0054] This application will generate a factor sorting sequence that changes in real time with meteorological data based on the graph attention mechanism, and establish a dynamic mapping function for the color depth of the heat layer based on the non-linear relationship between the sorting position and the change rate of root respiration entropy, that is, for meteorological dynamic sorting: For example, the GAT model adjusts the factor sorting from the conventional temperature > root activity > depth dynamics to soil oxygen content > wind speed > leaf transpiration based on real-time wind speed and air pressure data.

[0055] Feedback to the carbon emission heat map, the color depth of the corresponding area will be increased within the preset time, so that the meteorological-driven dynamic sorting makes the environmental meteorology also an important link in carbon emission monitoring, realizing the dynamic sorting of the influence of meteorological factors.

[0056] This application combines and establishes a dynamic mapping function for the color depth of the thermal layer based on the non-linear relationship between the sorting rank and the root respiration entropy change rate, realizing non-linear color mapping: By establishing a color depth - respiration entropy change rate function: For example: , When the entropy change rate ΔH in a certain area suddenly increases from 2.7 kJ / mol·h to 5.1 kJ / mol·h, the color depth of the carbon emission thermal map changes in the RGB gradient in the corresponding area. By capturing the color changes in the corresponding area in real time through a mobile terminal, the real-time response efficiency of non-linear mapping is judged, and the accuracy of carbon emission measurement is improved.

[0057] Example 8: This application provides a method for closed-loop verification. Refer to Figure 8 , specifically: During the generation process of the sensitive factor sorting sequence, this application synchronously collects the metabolic characteristics of the soil respiration sensor array; realizes the fusion of metabolic characteristics: When there is a correlation between the synchronous collection of sensitive factors and the underground soil respiration waveform, the sensitive factors and metabolic characteristics are positively correlated. For example, changes in the sensitive factors of the root system lead to an increase or decrease in the respiration entropy increase rate, and the carbon emission thermal map automatically adjusts the color depth of the operation area, realizing the quantification of medium data fusion and evaluating the changes in carbon emissions.

[0058] This application constructs a distributed factor weight correction model based on federated learning, and dynamically optimizes the correlation between the sorting sequence and the color depth of the thermal map through a reinforcement learning algorithm, that is: realizes federated learning collaborative optimization: This application shares the factor weight gradient through federated learning and collaboratively trains without exchanging the original data. When a certain area in the crop planting area, for example, node B, encounters pests and causes abnormal metabolism, the model quickly corrects the sorting rule based on the historical gradients of nodes A / C, so that the carbon emission thermal map adjusts the pest impact factor weight within 2 hours, improving the accuracy of carbon emission monitoring and also combining the group optimization advantages of distributed learning.

[0059] This application realizes the closed-loop verification of the farmer operation feedback data and the thermal layer rendering parameters based on the blockchain smart contract, that is, realizes the blockchain closed-loop verification: The smart contract of this application requires that the adjustment of the color depth of the heat map needs to verify the operation feedback of nodes in multiple node areas. For example, if the deviation of the blockchain timestamp is lower than the setting, the malicious tampering rendering parameter of the nodes in this area will be set to light green for the high-risk area. Then, anomalies are identified through hash value comparison and the node is frozen, ensuring the global consistency of the carbon emission heat map layer and reducing the incidence of mis-rendering events.

[0060] Example 9: This application provides a way to synchronize and visualize metabolic characteristics. Refer to Figure 9 , specifically: This application will input the sorted sequence of key sensitive factors into the spatio-temporal adversarial generation network, and configure the correlation matrix of multi-spectral remote sensing data and farmers' operation logs to generate a set of heat map layer rendering parameters with respiratory metabolic efficiency constraints, that is, to achieve ST-GAN data fusion. For example, when ST-GAN fuses the multi-spectral data of drones, the chlorophyll content in the 710nm band and the average sowing depth of farmers' sowing depth logs of 8.2 cm, a set of heat map rendering parameters will be generated. The spatio-temporal adversarial generation network identifies the decrease value of the photosynthesis-respiration efficiency ratio in the area where the sowing is too deep, and then marks this area with dark red on the carbon emission heat map to achieve the measurement of metabolic efficiency through air-ground data fusion and output the result of constraint optimization.

[0061] This application establishes a color depth gradient threshold dynamically adjusted by a meteorological factor mutation response function according to the heat map layer rendering parameter set, and integrates the synchronization and visualization of metabolic characteristics of the carbon emission heat map. Among them, the multi-spectral remote sensing data is collected by drone equipment, that is, to achieve the synchronous influence of meteorological mutation and edge computing: For example, when the air pressure drops or rises suddenly, the meteorological response function will increase or decrease the color depth gradient threshold. The carbon emission heat map will increase the color depth of the meteorological influence area within a preset time, and based on the fast adaptation ability of the dynamic threshold to meteorological mutations, improve the accuracy of carbon emission measurement.

[0062] Then, through the deployed edge computing nodes, such as drone equipment, the carbon emission heat map generation model is compressed. When the time stamp deviation between the heat map layer received by the mobile APP and the cloud metabolic data is very low, the real-time visualization efficiency of edge-cloud collaboration is achieved. The traditional solution can only achieve the aggregation of carbon emission remote sensing data in a region, cannot achieve the synchronization of meteorological and edge computing effects, and can only achieve gas monitoring.

[0063] Example 10: This application provides a system composition diagram of a carbon emission monitoring system based on agricultural carbon neutrality. Refer to Figure 10 , specifically: In its specific implementation, it first builds a multidimensional database based on carbon emissions based on the agricultural planting data of the current area to determine the dynamic characteristics of carbon emissions. This application focuses on multidimensional data fusion and dynamic feature analysis. In the crop planting area, multidimensional data such as soil respiration flux sensors, drone multispectral data and farmers' fertilization logs are integrated through a multidimensional database.

[0064] The carbon emission characteristics are judged through a multidimensional database. For example, when the sensor detects an abnormal increase in the root respiration entropy change rate in a certain area, the multidimensional database associates the meteorological data to find that the high temperature during the same period has caused a surge in soil microbial activity, and based on the farmer's log, the cause is locked in, such as excessive application of nitrogen fertilizer.

[0065] Furthermore, the carbon emission weight of the area is dynamically corrected, and the carbon neutrality anomaly factors that should have been classified as natural respiration sources are relabeled as regular factors or corrected factors in the current area. This is because emissions are dominated by agricultural activities, and then through contact control measures, after farmers receive instructions and adjust the irrigation frequency, the carbon emission intensity of the corresponding area decreases. Therefore, the multi-dimensional data fusion of this application has high-precision traceability capabilities for complex emission sources.

[0066] In its specific implementation, a dynamic analysis model will be constructed according to the dynamic characteristics of carbon emissions to determine the carbon neutrality efficiency indicator set; in this process, the carbon emission pattern of a certain period in history can be migrated to the newly created model, namely the dynamic analysis model, through transfer learning. By setting different emergency conditions, the efficiency indicator set of carbon neutrality is determined. For example, when sudden heavy rainfall causes a sudden change in soil impedance gradient, the dynamic Bayesian network of the dynamic analysis model autonomously adjusts the photosynthesis-respiration efficiency weight based on real-time sensor data, and then the carbon neutrality efficiency indicator is switched within a period of time, for example, from a moderately inefficient state to a severely unbalanced state. Compared with the traditional fixed threshold model that warns of emission risks in advance, farmers can install drainage channels based on this, thereby weakening the peak of carbon emissions and enhancing the advantages of rapid response and self-correction to environmental mutations.

[0067] In its specific implementation, according to the carbon neutralization efficiency index set, a carbon emission heat map of the current region is constructed. The prior art mainly reflects the static carbon emission situation in the form of a carbon emission pie chart, etc. Among them, the carbon emission heat map generates a sorting sequence of key sensitive factors that changes in real time based on the spatio-temporal evolution trend of carbon neutralization, and adjusts the color depth of the heat map based on the sorting sequence of key sensitive factors. In this process, according to the sorting sequence of sensitive factors, for example: temperature > root activity > irrigation frequency and other sensitive factors, a heat map is generated, and the temperature-sensitive area is marked with a red gradient. If there is a cold snap that causes the surface temperature of the crops to drop suddenly, this application will synchronously update the sorting based on federated learning to: root activity > microbial biomass > temperature, and then the color depth of the heat map will be dynamically adjusted accordingly, highlighting the changes in the carbon emission hotspots in the root frostbite area. The planting management end of the crops applies a biochar insulation layer to the frostbite area of the crop roots according to the visualization results, so that the carbon sink efficiency of this area is restored to the pre-disaster level, thereby maintaining the carbon neutralization level. The dynamic mapping of the heat map parameters of this application to the actual situation of complex scenarios can quickly provide rectification suggestions.

[0068] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalent technologies, this application is also intended to include these changes and modifications.

Claims

1. A carbon emission monitoring method based on agricultural carbon neutrality, characterized in that, Including: Construct a multi-dimensional database based on carbon emissions based on the agricultural planting data of the current region, and determine the dynamic characteristics of carbon emissions; Construct a dynamic analysis model according to the dynamic characteristics of carbon emissions, and determine the carbon neutralization efficiency index set; Construct a carbon emission heat map of the current region according to the carbon neutralization efficiency index set, wherein the carbon emission heat map generates a sorted sequence of key sensitive factors that change in real time based on the spatio-temporal evolution trend of carbon neutralization, and adjusts the color depth of the heat map based on the sorted sequence of key sensitive factors.

2. The carbon emission monitoring method based on agricultural carbon neutralization according to claim 1, characterized in that, A three-dimensional sensing component is configured in the rhizosphere layer of the agricultural crops in the current region; wherein, the three-dimensional sensing component includes a bioelectric sensor, a carbon dioxide isotope induction sensor, and an electrical impedance tomography device; The bioelectric sensor is used to determine the active respiration region corresponding to the root electrophysiological signal in the rhizosphere layer; The carbon dioxide isotope induction sensor is used to determine the respiration source according to the spectral measurement value of the active respiration region; The electrical impedance tomography device is used to generate a three-dimensional impedance map of the root system according to the respiration unit.

3. The carbon emission monitoring method based on agricultural carbon neutrality according to claim 2, characterized in that, The three-dimensional impedance map includes a respiration flux value, and the obtaining step of the respiration flux value is as follows: When the root electrophysiological signal reaches a sampling point with a preset power density in a preset monitoring frequency band, a first instruction is triggered; In response to the first instruction, perform a first characteristic frequency band scan on the active respiration region to determine effective respiration hotspots; According to the effective respiration hotspots, perform spectral measurement. When the respiration contribution rate of the effective respiration hotspots in the spectral measurement is higher than a preset contribution value, a second instruction is triggered; According to the second instruction, perform spatio-temporal matching on the offset value of the respiration contribution rate and the impedance gradient data, and when the matching is successful, generate a respiration flux value.

4. The carbon emission monitoring method based on agricultural carbon neutrality according to claim 3, characterized in that, The construction of the multi-dimensional database based on carbon emissions and the determination of the dynamic characteristics of carbon emissions include: Establish a spatio-temporal association database integrating planting types, intensity of farming activities, soil respiration flux, and meteorological factors; According to the spatio-temporal association database and the effective respiration hotspots, calculate the respiratory entropy change data of crop roots in real time, and combine the blockchain distributed storage of farmers' farming behavior data. Use the dynamic time warping algorithm to couple sensor data with farmers' operation logs to generate a spatio-temporal evolution map of carbon emission intensity with feedback on respiratory metabolism characteristics.

5. The carbon emission monitoring method based on agricultural carbon neutrality according to claim 1, characterized in that, The construction of the dynamic analysis model according to the dynamic characteristics of carbon emissions includes: According to the transfer learning algorithm configured on the feedback loop, map the agricultural planting data to a dynamic Bayesian network; According to the dynamic Bayesian network, determine the respiratory metabolism entropy weight model, and real-time correct the carbon emission intensity weight to generate a spatio-temporal carbon footprint gradient prediction matrix with a self-correction function of respiratory efficiency. And according to the spatio-temporal carbon footprint gradient prediction matrix, generate a dynamic analysis model.

6. The carbon emission monitoring method based on agricultural carbon neutrality according to claim 5, wherein, The determination of the carbon neutralization efficiency index set includes: According to the carbon footprint gradient prediction matrix, fuse the crop photosynthetic efficiency eigenvector corresponding to the agricultural planting data and the soil respiration efficiency spectrum; Adopt a graph embedding pruning algorithm to construct a multi-dimensional index dimensionality reduction channel, generate a three-dimensional carbon neutralization grade map with constraints on respiratory metabolism efficiency through a spatio-temporal dynamic evolution equation, and adaptively adjust according to the index weight with the change rate of root respiratory entropy.

7. The carbon emission monitoring method based on agricultural carbon neutralization according to claim 1, wherein, The construction of the carbon emission heat map of the current region according to the carbon neutralization efficiency index set includes: A three-dimensional hierarchical atlas based on a carbon neutrality efficiency index set extracts the coupling characteristics of farmers' farming behavior sequences and soil respiration metabolism waveforms through a spatio-temporal convolutional neural network, and constructs a sensitivity factor importance evaluation model with a time sliding window mechanism; among them, the soil respiration metabolism waveform is collected by a soil respiration sensor array composed of bioelectric sensors in a three-dimensional perception component. Generate a factor ranking sequence that changes in real time with meteorological data based on a graph attention mechanism, and establish a dynamic mapping function for the color depth of a heat map layer based on the non-linear relationship between the ranking position and the root respiration entropy change rate.

8. The carbon emission monitoring method based on agricultural carbon neutrality according to claim 7, characterized in that, According to the carbon neutrality efficiency index set, constructing a carbon emission heat map for the current region further includes: During the generation of the sensitivity factor ranking sequence, synchronously collect the metabolic characteristics of the soil respiration sensor array. Construct a distributed factor weight correction model based on federated learning, dynamically optimize the association rules between the ranking sequence and the heat map color depth through a reinforcement learning algorithm, and implement a closed-loop verification of the farmer operation feedback data and the heat map layer rendering parameters based on a blockchain smart contract.

9. The carbon emission monitoring method based on agricultural carbon neutralization according to claim 1, wherein, According to the carbon neutrality efficiency index set, constructing a carbon emission heat map for the current region further includes: Input the key sensitivity factor ranking sequence into a spatio-temporal adversarial generation network, and configure the association matrix of multi-spectral remote sensing data and farmer operation logs to generate a heat map layer rendering parameter set with respiratory metabolism efficiency constraints. According to the heat map layer rendering parameter set, establish a color depth gradient threshold dynamically adjusted by a meteorological factor mutation response function, and integrate the synchronous visualization of the metabolic characteristics of the carbon emission heat map; among them, the multi-spectral remote sensing data is collected by an unmanned aerial vehicle device.

10. A carbon emission monitoring system based on agricultural carbon neutrality, characterized in that, Include: Carbon emission feature collection module: Based on the agricultural planting data of the current region, construct a multi-dimensional database based on carbon emissions to determine the dynamic characteristics of carbon emissions. Carbon neutrality index set determination module: Used to construct a dynamic analysis model according to the dynamic characteristics of carbon emissions to determine the carbon neutrality efficiency index set. Carbon emission visualization module: Used to construct a carbon emission heat map for the current region according to the carbon neutrality efficiency index set, where the carbon emission heat map generates a real-time changing key sensitivity factor ranking sequence based on the carbon neutrality spatio-temporal evolution trend, and adjusts the heat map color depth based on the key sensitivity factor ranking sequence.

Images