Method and device for predicting carbon reserve of simulation area based on large language model

Abstract

The invention discloses a method and device for predicting carbon reserves of a simulation area based on a large language model, and the method comprises the steps: obtaining historical land cover data of the simulation area through employing the large language model, and determining the initial carbon reserves and an initial land utilization transfer matrix of the simulation area based on the historical land cover data; for each scene to be predicted, acquiring a scene factor corresponding to the scene to be predicted, and performing scene correction on the initial land utilization transfer matrix by using the scene factor to obtain a scene land transfer matrix; based on the scene land transfer matrix, performing land utilization transfer planning prediction on each scene to be predicted by using a Markov chain to obtain scene land cover data; according to the method, the predicted carbon reserves corresponding to the scene land cover data are determined based on the scene land cover data, accurate scene land cover data and predicted carbon reserves can be determined based on different scenes to be predicted, carbon reserve prediction of the scenes to be predicted is realized, and the reliability of carbon reserve prediction is effectively improved.

Description

Technical Field

[0001] The present disclosure generally relates to the field of data prediction technology, and in particular to a method and device for predicting and simulating regional carbon storage based on a large language model. Background Art

[0002] Land use change is one of the most important factors affecting the carbon cycle of terrestrial ecosystems, and its contribution to the increase in atmospheric CO2 concentration is second only to the burning of fossil fuels. Studies have shown that from 1850 to 1998, 136±5×10 9 Carbon emissions from terrestrial ecosystems to the atmosphere through land-use change account for approximately 33% of the total CO2 increase currently in the atmosphere.

[0003] In related technologies, calculations are usually made based on actual land use change results, and then urban design is guided by studying the carbon emissions corresponding to different land use types. Therefore, it is difficult to explore the relationship between land use changes and urban development needs, and there is a lack of targeted policy recommendations for land use management. Summary of the invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desired to provide a method and apparatus for predicting and simulating regional carbon reserves based on a large language model, which realizes the carbon reserve prediction of the prediction scenario and effectively improves the reliability of the carbon reserve prediction.

[0005] In a first aspect, an embodiment of the present application provides a method for predicting and simulating regional carbon storage based on a large language model, comprising:

[0006] Acquiring historical land cover data of the simulation area using a large language model, and determining an initial carbon stock and an initial land use transfer matrix of the simulation area based on the historical land cover data;

[0007] For each scenario to be predicted, a scenario factor corresponding to the scenario to be predicted is obtained, and the scenario factor is used to perform scenario correction on the initial land use transfer matrix to obtain a scenario land transfer matrix;

[0008] Based on the scenario land transfer matrix, using Markov chain to predict the land use transfer planning for each scenario to be predicted, to obtain scenario land cover data;

[0009] Based on the scenario land cover data, a predicted carbon storage corresponding to the scenario land cover data is determined.

[0010] In some embodiments, the obtaining of historical land cover data of the simulation area and determining the initial carbon storage and initial land use transfer matrix of the simulation area based on the historical land cover data comprises:

[0011] Acquire initial input information, and use a large language model to perform semantic extraction on the initial input information to obtain initial input features;

[0012] Matching the initial input feature with preset regional land information, determining regional information corresponding to the initial input information, and using the area corresponding to the regional information as the simulation area;

[0013] Acquiring corresponding historical land cover data based on the regional information;

[0014] The initial carbon storage of the simulation area is determined based on the land cover data of the current year in the historical land cover data, and the initial land use transfer matrix is ​​determined based on the historical land cover data.

[0015] In some embodiments, for each scenario to be predicted, scenario correction is performed on the initial land use transfer matrix to obtain a scenario land transfer matrix, including:

[0016] Acquire input scenario restriction information, and use a large language model to perform semantic analysis on the scenario restriction information to obtain a scenario development type corresponding to the scenario restriction information; the scenario development type is a future development type corresponding to the simulation area;

[0017] Determining a scenario factor for scenario correction of the initial land use transfer matrix based on the scenario development type;

[0018] Based on the scenario factors, the initial land use transfer matrix is ​​scenario modified to obtain the scenario land transfer matrix.

[0019] In some embodiments, for each scenario to be predicted, scenario correction is performed on the initial land use transfer matrix to obtain a scenario land transfer matrix, including:

[0020] Acquire input scenario restriction information, and use a large language model to perform semantic analysis on the scenario restriction information to obtain a scenario type corresponding to the scenario restriction information;

[0021] Based on the scenario type, using the large language model to determine simulation parameters and driving factors corresponding to the scenario type, and based on the simulation parameters and driving factors to determine the scenario factor corresponding to the scenario type;

[0022] Based on the scenario factors, the initial land use transfer matrix is ​​scenario modified to obtain the scenario land transfer matrix.

[0023] In some embodiments, it also includes:

[0024] Determining the land use transfer data of the simulation area based on the historical land cover data and the corresponding initial carbon storage and the scenario land cover data and the corresponding predicted carbon storage; the land use transfer data at least includes the total transfer amount, transfer ratio and transfer direction of each type of land;

[0025] Based on the initial carbon stock, the predicted carbon stock and the land use transfer data, a land use transfer planning suggestion corresponding to the simulation area is generated.

[0026] In some embodiments, generating a land use transfer planning suggestion corresponding to the simulation area based on the initial carbon storage, the predicted carbon storage and the land use transfer data includes:

[0027] When the predicted carbon stock corresponding to any of the scenarios to be predicted is greater than the initial carbon stock, recommending the scenario land cover data corresponding to the scenario to be predicted as the land use transfer planning suggestion; or

[0028] When the predicted carbon stocks corresponding to all the scenarios to be predicted are less than or equal to the initial carbon stocks, feedback requests to adjust the scenario factor for at least one scenario to be predicted to obtain an adjusted scenario factor;

[0029] The corresponding predicted carbon storage is reacquired based on the adjusted scenario factors until the obtained predicted carbon storage is greater than the initial carbon storage.

[0030] In a second aspect, an embodiment of the present application provides a device for predicting and simulating regional carbon storage based on a large language model, comprising:

[0031] An acquisition module, used for acquiring historical land cover data of a simulation area using a large language model, and determining an initial carbon storage and an initial land use transfer matrix of the simulation area based on the historical land cover data;

[0032] A correction module, for performing scenario correction on the initial land use transfer matrix for each scenario to be predicted, to obtain a scenario land transfer matrix;

[0033] A prediction module, for performing land use transfer planning prediction for each scenario to be predicted based on the scenario land transfer matrix using a Markov chain to obtain scenario land cover data;

[0034] A determination module is used to determine the predicted carbon storage corresponding to the scenario land cover data based on the scenario land cover data.

[0035] In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the embodiment of the present application when executing the program.

[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the embodiment of the present application.

[0037] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, characterized in that when the computer program is executed by a processor, it implements the method described in the embodiment of the present application.

[0038] The method and device for predicting and simulating regional carbon storage based on a large language model provided in the embodiments of the present application can determine the land transfer matrix consistent with the scenario to be predicted based on different scenarios to be predicted, and then perform planning and prediction in combination with a Markov chain with good convergence for random prediction, so that more accurate scenario land cover data can be obtained under the scenario to be predicted, and carbon storage calculation is performed based on the scenario land cover data to obtain predicted carbon storage, thereby realizing the carbon storage prediction for the scenario to be predicted and effectively improving the reliability of the carbon storage prediction.

[0039] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0041] Figure 1 A schematic diagram of a process for predicting and simulating regional carbon storage based on a large language model provided in an embodiment of the present application is shown;

[0042] Figure 2 A schematic diagram of a process for predicting and simulating regional carbon storage based on a large language model provided in another embodiment of the present application is shown;

[0043] Figure 3 A schematic flow chart of a method for predicting and simulating regional carbon storage based on a large language model provided in another embodiment of the present application is shown;

[0044] Figure 4 A schematic flow chart of a method for predicting and simulating regional carbon storage based on a large language model provided in yet another embodiment of the present application is shown;

[0045] Figure 5A schematic diagram showing the principle of a method for predicting and simulating regional carbon storage based on a large language model provided in an embodiment of the present application is shown;

[0046] Figure 6 A block diagram of a device for predicting and simulating regional carbon storage based on a large language model provided in one embodiment of the present application is shown;

[0047] Figure 7 A schematic diagram of the structure of a computer system of an electronic device or server suitable for implementing an embodiment of the present application is shown. DETAILED DESCRIPTION

[0048] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0049] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0050] In order to further illustrate the technical solution provided by the embodiment of the present application, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiment of the present application provides the method operation instruction steps shown in the following embodiments or drawings, more or less operation instruction steps may be included in the method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided by the embodiment of the present application. The method may be executed in the order of the method shown in the embodiment or drawings or in parallel during the actual processing process or when the device is executed.

[0051] Please refer to Figure 1 , Figure 1 The flowchart of the method for predicting and simulating regional carbon storage based on a large language model provided by an embodiment of the present application is shown. Figure 1 As shown, the method includes:

[0052] Step 101: Use a large language model to obtain historical land cover data of a simulation area, and determine an initial carbon storage and an initial land use transfer matrix of the simulation area based on the historical land cover data.

[0053] It should be noted that the simulation area is an area that needs to carry out land use transfer planning. The initial carbon storage determined based on the historical land cover data of the simulation area can provide reliable parameter comparison data for carbon storage prediction after the subsequent land use transfer planning. The land use transfer matrix is ​​used to provide land use type transfer probability information. The initial land use transfer matrix can fully express the land use transfer situation of the continued development of the historical current land use planning. Based on the initial land use transfer matrix, the correction that conforms to the development scenario of the simulation area can effectively predict the development of land use transfer under the development scenario, and then realize the carbon storage prediction based on the development scenario.

[0054] In a feasible embodiment, the present application introduces large language model technology to assist in analyzing the user's initial input information, thereby effectively reducing the user's input standardization requirements when making planning predictions, that is, the user does not need to have a high level of land cover information to make relevant predictions, effectively reducing the difficulty of user planning predictions and improving user experience.

[0055] Specifically, if Figure 2 As shown, step 101, obtaining historical land cover data of the simulation area, and determining the initial carbon storage and initial land use transfer matrix of the simulation area based on the historical land cover data, including:

[0056] Step 1011, obtaining initial input information, using a large language model to perform semantic extraction on the initial input information, and obtaining initial input features.

[0057] It should be noted that the initial input information is text information entered by the user in the interactive interface, which is used to represent the area that needs to be planned and predicted. For example, the user enters "My research area is Beijing", that is, "My research area is Beijing" is the initial input information entered by the user, and the initial input information is semantically extracted using the large language model to obtain the core semantics - "Beijing", and then the feature vector corresponding to "Beijing" is obtained by feature extraction of the core semantics, which is used for feature matching in the large language model.

[0058] Step 1012, matching the initial input features with preset regional land information, determining the regional information corresponding to the initial input information, and using the area corresponding to the regional information as the simulation area.

[0059] It should be noted that the large language model will extract the initial input features and perform information matching according to the pre-set rules of analysis and reminder information, wherein the analysis and reminder rules include preset regional land information, such as the source of the preset regional land information.

[0060] Exemplarily, the analysis reminder message that can be preset in advance can be "Please parse the regional information entered by the user and determine whether it belongs to one of the provincial, municipal, and county levels in the National Land Cover Database." Therefore, the large language model matches the initial input features with the features corresponding to the provinces, municipalities, and counties in the National Land Cover Database. If the match is successful, it is determined that the initial input information belongs to one of the provinces, municipalities, and counties in the National Land Cover Database. If the match is unsuccessful, the user is fed back to modify the initial input information. For example, the large language model uses an interactive interface to show the user "It has been confirmed that the research area you specified is Beijing, which belongs to the provincial region in the National Land Cover Database." Or "I am very sorry, I cannot confirm that 'XXX' is a valid region in the National Land Cover Database. Please re-enter or correct the information of the research area."

[0061] Among them, the large language model can extract features of provinces, cities, and counties in the "National Land Cover Database" in advance to obtain a feature library. After the initial input features are extracted according to the initial input information, the initial input features are compared with the feature library to determine the feature distance (such as cosine distance or Euclidean distance) between the initial input features and each feature in the feature library. If the feature distance is greater than the preset distance threshold, it is determined that the initial input feature matches the feature, and the area and level corresponding to the feature are the area information corresponding to the initial input information entered by the user, and the area is determined to be a simulation area.

[0062] Step 1013: Acquire corresponding historical land cover data based on the regional information.

[0063] After the regional information corresponding to the user's initial input information has been determined, the historical land cover data corresponding to the regional information can be obtained according to the historical land cover data corresponding to each region stored in the database. The historical land cover data may be the land cover data of the latest four periods. The data of the latest four periods are the data reported (or uploaded) after statistics in the latest four land cover data statistical cycles.

[0064] The land cover data include but are not limited to the land use types corresponding to each location in the region and the area corresponding to each land use type.

[0065] Step 1014: determine the initial carbon storage of the simulation area based on the land cover data of the current year in the historical land cover data, and determine the initial land use transfer matrix based on the historical land cover data.

[0066] It should be noted that before determining the initial carbon storage of the simulation area, it is also necessary to obtain the carbon sink factor corresponding to each land use type based on the historical land cover data of the simulation area, and then determine the carbon storage of the current year based on the land cover data and carbon sink factor of the current year, and use it as the initial carbon storage.

[0067] Furthermore, in a feasible embodiment, the initial land use transfer matrix is ​​determined using land cover data of two consecutive periods (the current year and the previous year), and the following formula can be used specifically:

[0068]

[0069] Among them, P ij is the relative proportion of the i-th land use type in the previous year transferred to the j-th land use type in the current year, A T1 (i) is the land cover data matrix corresponding to the previous year, A T2 (j) is the land cover data matrix corresponding to the current year.

[0070] It should be noted that the land cover data matrix may correspond to the regional map, that is, the land use type of each position point in the matrix is ​​determined according to the pixel position corresponding to the regional map.

[0071] In a feasible embodiment, the database may store land cover data in the form of regional images, that is, the land cover data corresponding to each position in the region may be determined by image recognition.

[0072] Specifically, the images of the previous year and the current year corresponding to the simulation area are obtained respectively, and the corresponding image gray value matrix is ​​obtained as the land cover data matrix of the corresponding year, that is, A T1 (i) and A T2 (j). Then the land use transfer matrix is ​​calculated using the above formula.

[0073] In another feasible embodiment, after respectively acquiring the images of the previous year and the current year corresponding to the simulation area and obtaining the corresponding image gray value matrix, the image gray value matrix is ​​converted into a one-dimensional image array. Finally, the confusion matrix is ​​used to analyze the two image arrays to obtain a land use transfer matrix.

[0074] It should be understood that in the embodiment of the present application, the probability analysis of the confusion matrix is ​​used to construct the predicted values ​​and actual values ​​of the land use transfer in the previous year and the current year as the land use transfer matrix.

[0075] It should be noted that in the embodiment of the present application, the current year and the previous year can be natural years or statistical periods of land cover data, that is, the current annual cycle and the previous annual cycle. The specific settings can be made according to user needs and data sources, and this application does not make specific limitations.

[0076] Step 102, for each scenario to be predicted, a scenario factor corresponding to the scenario to be predicted is obtained, and the scenario factor is used to perform scenario correction on the initial land use transfer matrix to obtain a scenario land transfer matrix.

[0077] It should be noted that the scenario factor is a parameter factor used to correct the land use transfer data in the initial land use transfer matrix.

[0078] In a feasible embodiment, scenario restriction information input by the user can be received based on an interactive interface, and a large language model can be used to perform semantic analysis on the scenario restriction information to obtain a scenario development type corresponding to the scenario restriction information; the scenario development type is the future development type corresponding to the simulation area; based on the scenario development type, an adjustment ratio for scenario correction of the initial land use transfer matrix is ​​determined; based on the adjustment ratio, scenario correction is performed on the initial land use transfer matrix to obtain a scenario land transfer matrix.

[0079] For example, assuming that scenario factors are used to modify the land use transfer matrix in a weighted manner, the scenario restriction information input by the user may be economic development restriction information for the simulation area, such as business as usual development, economic development and ecological protection, etc. The large language model determines the scenario factors corresponding to the scenario restriction information input by the user.

[0080] Among them, the business-as-usual development type fully considers the historical inertia of regional development, that is, it is based on the development law of simulating regional land use transfer. Since it is closer to the current land use planning, there is no need for additional corrections on the basis of the initial land use transfer matrix. The corresponding scenario factor is 1, that is, no adjustment or modification is made to the initial land use transfer matrix.

[0081] Economic development type aims to meet material needs, usually giving priority to economic efficiency, even at the expense of natural land use types. For example, in order to meet the needs of a growing population, more land use types can be transferred to urbanization, including but not limited to transferring more existing grassland, cultivated land, forest land, water bodies and unused land to urban types. Therefore, the scenario factors distinguish the types of land use transfers. For example, the scenario factors of transfers to towns are all increased, that is, greater than 1, and the scenario factors of transfers to natural environments (such as grassland, forest land and water bodies) are all reduced, that is, greater than 0 and less than 1.

[0082] The ecological protection type is a type that takes ecological protection as its core goal. It is the opposite of the economic development type in terms of macro-level, which means strictly protecting ecological land (i.e., forest land, grassland, water area, and cultivated land) and protecting ecological diversity. Therefore, the scenario factors are opposite to the economic development type. For example, the scenario factors transferred to towns are all reduced, that is, reduced to greater than 0 and less than 1, and the scenario factors transferred to the natural environment (such as grassland, forest land, and water body) are all increased, that is, greater than 1.

[0083] In a feasible embodiment, the scenario restriction information input by the user may also include further restriction content, for example, the economic development type may be further restricted to high-speed economic development, economic development and low-speed economic development. At this time, the large language model may determine the corresponding adjustment ratio based on the further information on "economic development". For example, the range of the scenario factors preset in advance is 0-2. The scenario factors transferred to towns in the high-speed economic development scenario are all improved, that is, greater than 1 and closer to 2, and the scenario factors transferred to natural environments (such as grasslands, woodlands and water bodies) are all reduced, that is, greater than 0 and less than 1 and closer to 0. In the low-speed economic development scenario, the scenario factors transferred to towns are all improved, that is, greater than 1 but closer to 1, and the scenario factors transferred to natural environments (such as grasslands, woodlands and water bodies) are all reduced, that is, greater than 0 and less than 1 but also closer to 1.

[0084] Furthermore, after obtaining the scenario factors, the transfer data at the corresponding positions in the initial land transfer matrix are scenario-corrected to obtain the scenario land transfer matrix. Exemplarily, the rows of the initial land transfer matrix are multiple original land use types, and the columns are multiple transferred land types. The scenario factors are used to correct the values ​​of the rows or columns affected by the scenario, such as multiplying them with the scenario factors to obtain the scenario land transfer matrix. For example, in the high-speed economic development scenario, the column with the land use type of "town" is multiplied by the corresponding scenario factor greater than 1 and close to 2 to achieve the purpose of scenario correction of the initial land transfer matrix.

[0085] Therefore, the present application can utilize a large language model to extract the scenario development type and scenario development speed corresponding to the simulation area based on the scenario restriction information of the simulation area input by the user, thereby determining the scenario factors that meet the scenario restriction information input by the user and are reasonably proportioned, so that the scenario land transfer matrix obtained by scenario correction based on the scenario restriction factors can not only meet the user's prediction needs, but also be more in line with the law of land transfer changes, thereby improving the reliability of the scenario land transfer matrix.

[0086] In another feasible embodiment, the scenario type corresponding to the scenario restriction information can also be a type corresponding to other user needs, such as the current scenario, similar scenarios and derived scenarios. Among them, the current scenario is the current information data of the simulation area, including but not limited to conventions and regulations, government policies, environmental resources, development models, etc. The development model includes population, economy, society, technology, etc., which are not specifically limited in this application. Similar scenarios are scenario types that are fine-tuned based on the scenario restriction information input by the user and the government policies, environmental resources, etc. under the current scenario of the simulation area. Derived scenarios are scenario types that may be formed in the predicted area, which are predicted by the large language model based on the information of the current scenario and the scenario restriction information.

[0087] Specifically, the scenario restriction information input by the user is obtained, and the large language model is used to determine the simulation parameters and driving factors corresponding to the scenario type, and the scenario factors corresponding to the scenario type are determined by giving the simulation parameters and driving factors; based on the scenario factors, the initial land use transfer matrix is ​​scenario-corrected to obtain the scenario land transfer matrix.

[0088] It should be noted that the driving factor is an indicator factor that can affect land use change based on the analysis of land use change characteristics and laws. In the embodiment of the present application, the driving factor can be selected as follows:

[0089]

[0090] The simulation parameters are weight values ​​determined based on historical data of driving factors. Specifically, the weight values ​​corresponding to the evaluation indicators can be determined using the following formula:

[0091]

[0092] g j =1-e j

[0093]

[0094]

[0095] Among them, P ij is the information entropy of the j-th driving factor in the i-th year, e j is the intermediate parameter, g j is the utility value of the jth indicator, w j is the weight value of the j-th driving factor, x' ij Represents the standard data of the j-th driving factor in the i-th year.

[0096] It should be understood that since the driving factors have few dimensions, units and quantities, in order to further eliminate the differences in dimensions, units and orders of magnitude between different indicators, this application uses the range normalization method to standardize the driving factors to 0-1.

[0097] In a feasible embodiment, the following formula is used for range normalization:

[0098]

[0099] Where i = 1, 2, ..., m, j = 1, 2, ..., n; m represents the number of years of data, n represents the number of driving factors, x represents the number of driving factors, ij represents the original data of the j-th driving factor in the i-th year, x' ij Represents the standard data of the j-th driving factor in the original i-th year.

[0100] Furthermore, after obtaining the simulation parameters and driving factors, the large language model further reviews and determines the simulation parameters and driving factors, such as whether the driving factors corrected by the simulation parameter weights are within a reasonable range, such as whether the land use of each ecological element exceeds the ecological red line corresponding to the simulation area, whether the increase in transportation facilities caused by urbanization exceeds a reasonable proportion, etc. If the simulation parameters and driving factors are unsuccessful in the review and determination, the user will be fed back with unsuccessful information through the interactive interface to remind the user to re-enter the scenario restriction information, so that the development forecast of the simulation area meets the basic rules in the stage of setting the forecast conditions, avoiding the generation of a large number of unnecessary forecast results, improving the forecast efficiency, and ensuring the reliability of the forecast results.

[0101] Optionally, under different scenario types, data from different periods of different driving factors can be selected. For example, the current development scenario selects all driving factors and their corresponding simulation parameters, and similar scenarios and derivative scenarios can refer to the corresponding development trends to select the corresponding driving factors. For example, when the development trend tends to economic development, the multiple driving factors corresponding to the ecological factor B are selected as the minimum value, and the multiple driving factors corresponding to the architectural factor C and the social factor D are selected as the maximum value. When the development trend tends to ecological protection, the multiple driving factors corresponding to the ecological factor B are selected as the maximum value, and the multiple driving factors corresponding to the architectural factor C and the social factor D are selected as the minimum value.

[0102] Furthermore, after the simulation parameters and driving factors are determined, the scenario factors can be determined according to the weights and sums of the driving factors and their corresponding simulation parameters under different scenario types. Then, the scenario factors are used to perform scenario correction on the entire initial land transfer matrix, that is, the scenario factors are multiplied by (each item in) the initial land transfer matrix to obtain the corrected scenario land transfer matrix.

[0103] Therefore, the embodiment of the present application fully considers the characteristics and laws of land use changes and the scenario restrictions of user needs to modify the initial land transfer matrix, so that the revised scenario land transfer matrix can not only meet the scenario types required by users, but also fully consider the influence of historical driving factors in the simulated area, so that the land use transfer planning prediction according to the scenario land transfer matrix can be more reliable.

[0104] Step 103, based on the scenario land transfer matrix, using the Markov chain to perform land use transfer planning prediction for each scenario to be predicted, and obtain scenario land cover data.

[0105] It should be noted that the Markov chain has the property of converging to its stationary distribution in a random process. Therefore, after determining the scenario land transfer matrix, the Markov chain can be used to well predict the final land use planning data according to the development of the scenario land transfer matrix, that is, it can determine the final scenario land cover data according to the development of the land transfer matrix.

[0106] Step 104: Determine the predicted carbon storage corresponding to the scenario land cover data based on the scenario land cover data.

[0107] It should be understood that after determining the scenario land cover data under the scenario type development, the scenario land cover data and the carbon sink factor can be used to calculate the predicted carbon storage corresponding to the scenario land cover data.

[0108] Therefore, the method for predicting and simulating regional carbon storage based on a large language model provided in the embodiment of the present application can determine the land transfer matrix that is consistent with the scenario to be predicted based on different scenarios to be predicted, and then combine it with the Markov chain with good convergence for random prediction to perform planning and prediction, so that more accurate scenario land cover data can be obtained under the scenario to be predicted, and carbon storage calculation is performed based on the scenario land cover data to obtain predicted carbon storage, which realizes the carbon storage prediction of the scenario to be predicted and effectively improves the reliability of carbon storage prediction.

[0109] In a possible embodiment, Figure 3 As shown, the method for predicting and simulating regional carbon storage based on a large language model proposed in the embodiment of the present application also includes:

[0110] Step 301 : determining land use transfer data of a simulation area based on historical land cover data and its corresponding initial carbon storage and scenario land cover data and its corresponding predicted carbon storage.

[0111] Among them, land use transfer data at least include the total transfer amount, transfer ratio and transfer direction of each type of land.

[0112] Step 302: Generate land use transfer planning suggestions corresponding to the simulation area based on the initial carbon storage, predicted carbon storage and land use transfer data.

[0113] In a possible embodiment, if Figure 4 As shown, step 302 generates land use transfer planning suggestions corresponding to the simulation area based on the initial carbon storage, predicted carbon storage and land use transfer data, including:

[0114] Step 3021: When the predicted carbon storage corresponding to any scenario to be predicted is greater than the initial carbon storage, the scenario land cover data corresponding to the scenario to be predicted is recommended as a land use transfer planning suggestion.

[0115] or

[0116] Step 3022: When the predicted carbon storage corresponding to all scenarios to be predicted is less than or equal to the initial carbon storage, a feedback request is made to adjust the scenario factor for at least one scenario to be predicted to obtain an adjusted scenario factor.

[0117] Step 3023: reacquire the corresponding predicted carbon storage based on the adjusted scenario factors until the predicted carbon storage is greater than the initial carbon storage.

[0118] It should be noted that in the embodiment of the present application, the purpose of carbon reserve prediction is to seek future land use transfer planning that can increase the carbon reserves in the simulated area. Therefore, the scenario in which the predicted carbon reserves increase relative to the initial carbon reserves and its corresponding scenario land cover data are selected as land use transfer planning suggestions for recommendation.

[0119] In a specific embodiment, Figure 5 As shown, predictions are made according to three scenario types to be predicted (current scenario, similar scenario and derived scenario) to obtain three predicted carbon stocks corresponding to the three scenario types, and then it is determined whether at least one of the three predicted carbon stocks is greater than the initial carbon stock. If so, the land cover data corresponding to the predicted carbon stock greater than the initial carbon stock is recommended as a land use transfer planning suggestion. If not, the user feedback module is used to send a "prediction unsuccessful" feedback message to the user, receive the scenario restriction information re-entered by the user, and use the large language model to re-determine the scenario land transfer matrix based on the re-entered scenario restriction information and the review conditions, and then re-obtain the predicted carbon stock, until there is a prediction result in which the predicted carbon stock is greater than the initial carbon stock, and the land use transfer planning suggestion is output.

[0120] For example, in an embodiment where the predicted carbon storage is greater than the initial carbon storage, the land cover data is:

[0121]

[0122] Based on the land cover data and its corresponding predicted carbon storage, the land use transition data of the simulation area is determined:

[0123] 1. Total amount transferred:

[0124] Total grassland transfer: approximately 3,139,183,192 square units;

[0125] Total town transfer volume: approximately 2,245,202,052 square units;

[0126] Total amount of cultivated land transferred: approximately 1,048,318,262 square units;

[0127] Total forest land transfer: approximately 1,347,725,474 square units;

[0128] Total water volume transferred: approximately 66,486,174 square units;

[0129] Total transfer of unused land: approximately 774,579 square units.

[0130] 2. Transfer ratio:

[0131] The percentage of grassland transferred from 2011 to 2019: about 98.6%;

[0132] The proportion of transfer from urban areas in 2011 to urban areas in 2019: about 82.8%;

[0133] The proportion of cultivated land transferred from 2011 to 2019: about 89.6%;

[0134] The percentage of forest land transferred from 2011 to 2019: about 95.4%;

[0135] The percentage of water bodies transferred from 2011 to 2019: about 91.5%;

[0136] The proportion of unused land transferred from 2011 to 2019: about 97.5%.

[0137] 3. Main transfer directions of various types of land:

[0138] Grassland: Mainly remains stable, with small amounts shifting to towns, cultivated land, woodlands and water bodies.

[0139] Towns: Most urban land use remains stable, with a small amount transferred from cultivated land, grassland and unused land.

[0140] Cultivated land: The land was mainly transferred from cultivated land to urban and forest land, indicating the impact of urban expansion and forest protection policies.

[0141] Forest land: Mainly transferred from forest land to towns and cultivated land, which may be related to urbanization and agricultural structure adjustment.

[0142] Water body: Little change, mainly restricted by natural factors such as topography and landform.

[0143] Unused land: Mainly remain stable, with a small amount transferred to other types, which may be affected by protection policies.

[0144] The output recommendations may include:

[0145] 1. Forest protection and restoration: According to the results analysis, the increase in forest area shows the importance of forest in land use. Therefore, it is recommended to strengthen the implementation of forest protection policies to maintain the existing forest area, promote forest ecological restoration projects, and promote the stability of forest ecosystems and the increase of carbon storage. For example, according to the analysis results, the forest area increased by about 224,007,828 square units between 2011 and 2019, which shows the effectiveness of forest protection policies, but still needs to be further strengthened.

[0146] 2. Water ecological protection: Considering the contribution of water bodies to carbon storage, it is recommended to strengthen the ecological environment protection of water bodies to prevent water pollution and ecosystem degradation. According to the results analysis, the change in water area is small, but its ecological function still contributes significantly to carbon storage. Therefore, promoting water ecological restoration projects is an important measure to maintain water carbon storage.

[0147] 3. Cultivated land management: Despite the reduction in cultivated land area, the contribution of cultivated land to carbon storage is still important. According to the results analysis, the transfer of cultivated land to urban and forest land is more significant. Therefore, it is recommended to adopt scientific farming techniques to increase soil organic carbon content and reduce land degradation and carbon emissions. For example, according to the analysis results, the cultivated land area has decreased by about 116,307,780 square units, which indicates the urgency and importance of cultivated land management.

[0148] 4. Urbanization planning: The impact of urbanization on carbon storage is mainly reflected in land use and energy consumption. According to the results, urbanization leads to a reduction in the area of ​​​​cultivated land and forest land. Therefore, in urbanization planning, attention should be paid to energy conservation and emission reduction and low-carbon development, promoting urban green construction and renewable energy utilization, reducing carbon emissions, and improving the quality of urban ecological environment. For example, according to the analysis results, urban land increased by about 2,245,202,052 square units, which highlights the urgency and importance of carbon emission reduction in urbanization planning.

[0149] Therefore, the embodiments of the present application can comprehensively consider the impact of land use changes on carbon storage, and then formulate reasonable land use planning recommendations and management measures, providing a reliable data foundation and strategic guarantee for effectively promoting sustainable land use and the realization of carbon neutrality.

[0150] Moreover, this application uses a large language model for analysis, prediction and recommendation, which reduces the professional requirements of the prediction users and does not require users to understand and master professional data processing technology. At the same time, the semantic analysis of the large language model can achieve standardization of data collection, prediction and analysis can achieve accurate data processing, and planning recommendations can achieve enriched data recommendations. It can assist users in finding the predicted carbon storage and land use planning strategies that meet their purposes during the prediction process, while understanding the impact of land use on carbon storage and the ecological environment, and provide a basis for the designation of green development policies.

[0151] It should be noted that although the operations of the method of the present invention are described in a particular order in the drawings, this does not require or imply that the operations must be performed in this particular order or that all illustrated operations must be performed to achieve desired results.

[0152] Figure 6 A block diagram of a device for predicting and simulating regional carbon storage based on a large language model provided in one embodiment of the present application is shown.

[0153] like Figure 6 As shown, the device 10 for predicting and simulating regional carbon storage based on a large language model proposed in an embodiment of the present application includes:

[0154] An acquisition module 11 is used to acquire historical land cover data of the simulation area using a large language model, and determine the initial carbon storage and initial land use transfer matrix of the simulation area based on the historical land cover data;

[0155] A correction module 12 is used to obtain a scenario factor corresponding to each scenario to be predicted, and use the scenario factor to perform scenario correction on the initial land use transfer matrix to obtain a scenario land transfer matrix;

[0156] A prediction module 13 is used to perform land use transfer planning prediction for each scenario to be predicted based on the scenario land transfer matrix using a Markov chain to obtain scenario land cover data;

[0157] The determination module 14 is used to determine the predicted carbon storage corresponding to the scenario land cover data based on the scenario land cover data.

[0158] In some embodiments, the acquisition module 11 is further used to:

[0159] Acquire initial input information, and use a large language model to perform semantic extraction on the initial input information to obtain initial input features;

[0160] Matching the initial input feature with preset regional land information, determining regional information corresponding to the initial input information, and using the area corresponding to the regional information as the simulation area;

[0161] Acquiring corresponding historical land cover data based on the regional information;

[0162] The initial carbon storage of the simulation area is determined based on the land cover data of the current year in the historical land cover data, and the initial land use transfer matrix is ​​determined based on the historical land cover data.

[0163] In some embodiments, the correction module 12 is further configured to:

[0164] Acquire input scenario restriction information, and use a large language model to perform semantic analysis on the scenario restriction information to obtain a scenario development type corresponding to the scenario restriction information; the scenario development type is a future development type corresponding to the simulation area;

[0165] Determining a scenario factor for scenario correction of the initial land use transfer matrix based on the scenario development type;

[0166] Based on the scenario factors, the initial land use transfer matrix is ​​scenario modified to obtain the scenario land transfer matrix.

[0167] In some embodiments, the correction module 12 is further configured to:

[0168] Acquire input scenario restriction information, and use a large language model to perform semantic analysis on the scenario restriction information to obtain a scenario type corresponding to the scenario restriction information;

[0169] Based on the scenario type, using the large language model to determine simulation parameters and driving factors corresponding to the scenario type, and based on the simulation parameters and driving factors to determine the scenario factor corresponding to the scenario type;

[0170] Based on the scenario factors, the initial land use transfer matrix is ​​scenario modified to obtain the scenario land transfer matrix.

[0171] In some embodiments, the determination module 14 is further configured to:

[0172] Determining the land use transfer data of the simulation area based on the historical land cover data and the corresponding initial carbon storage and the scenario land cover data and the corresponding predicted carbon storage; the land use transfer data at least includes the total transfer amount, transfer ratio and transfer direction of each type of land;

[0173] Based on the initial carbon stock, the predicted carbon stock and the land use transfer data, a land use transfer planning suggestion corresponding to the simulation area is generated.

[0174] In some embodiments, the determination module 14 is further configured to:

[0175] When the predicted carbon stock corresponding to any of the scenarios to be predicted is greater than the initial carbon stock, recommending the scenario land cover data corresponding to the scenario to be predicted as the land use transfer planning suggestion; or

[0176] When the predicted carbon stocks corresponding to all the scenarios to be predicted are less than or equal to the initial carbon stocks, feedback requests to adjust the scenario factor for at least one scenario to be predicted to obtain an adjusted scenario factor;

[0177] The corresponding predicted carbon storage is reacquired based on the adjusted scenario factors until the obtained predicted carbon storage is greater than the initial carbon storage.

[0178] It should be understood that the units or modules described in the apparatus 10 for predicting and simulating regional carbon storage based on a large language model are similar to those described in the reference Figure 2 The various steps in the described method correspond to each other. Therefore, the operations and features described above for the method are also applicable to the device 10 for predicting and simulating regional carbon reserves based on a large language model and the units contained therein, and will not be repeated here. The device 10 for predicting and simulating regional carbon reserves based on a large language model can be pre-implemented in a browser or other security application of an electronic device, or can be loaded into a browser or its security application of an electronic device by downloading or the like. The corresponding units in the device 10 for predicting and simulating regional carbon reserves based on a large language model can cooperate with the units in the electronic device to implement the solution of the embodiment of the present application.

[0179] For the several modules or units mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0180] The device for predicting and simulating regional carbon storage based on a large language model provided in the embodiment of the present application can determine the land transfer matrix consistent with the scenario to be predicted based on different scenarios to be predicted, and then perform planning and prediction in combination with a Markov chain with good convergence for random prediction, so that more accurate scenario land cover data can be obtained under the scenario to be predicted, and carbon storage calculation is performed based on the scenario land cover data to obtain predicted carbon storage, thereby realizing the carbon storage prediction of the scenario to be predicted and effectively improving the reliability of the carbon storage prediction.

[0181] Reference below Figure 7 , Figure 7 A schematic diagram of the structure of a computer system of an electronic device or server suitable for implementing an embodiment of the present application is shown.

[0182] like Figure 7 As shown, the computer system includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 702 or the program loaded from the storage part 708 to the random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation instructions of the system are also stored. The CPU 701, the ROM 702 and the RAM 703 are connected to each other through a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0183] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, etc.; an output section 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, a modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed, so that a computer program read therefrom is installed into the storage section 708 as needed.

[0184] In particular, according to an embodiment of the present application, the above reference flow chart Figure 2The described process can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program includes a program code for executing the method shown in the flow chart. In such an embodiment, the computer program includes a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 709, and / or installed from the removable medium 711. When the computer program is executed by the central processing unit (CPU) 701, the above-mentioned functions defined in the system of the present application are executed.

[0185] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium such as a computer-readable storage medium that can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0186] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operating instructions of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the aforementioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, the boxes represented by two connections can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operating instruction, or can be implemented with a combination of dedicated hardware and computer instructions.

[0187] The units or modules involved in the embodiments described in the present application may be implemented by software or by hardware. The described units or modules may also be set in a processor. For example, they may be described as: a processor includes an acquisition module, a correction module, a prediction module, and a determination module. Among them, the names of these units or modules do not constitute a limitation on the units or modules themselves under certain circumstances. For example, the acquisition module may also be described as "obtaining historical land cover data of the simulation area, and determining the initial carbon storage and initial land use transfer matrix of the simulation area based on the historical land cover data".

[0188] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device. The above computer-readable storage medium stores one or more programs, and when the above programs are used by one or more processors to execute the method for predicting and simulating regional carbon storage based on a large language model described in the present application.

[0189] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A method for predicting and simulating regional carbon storage based on a large language model, characterized in that: include: Acquiring historical land cover data of the simulation area using a large language model, and determining an initial carbon stock and an initial land use transfer matrix of the simulation area based on the historical land cover data; For each scenario to be predicted, a scenario factor corresponding to the scenario to be predicted is obtained, and the scenario factor is used to perform scenario correction on the initial land use transfer matrix to obtain a scenario land transfer matrix; Based on the scenario land transfer matrix, using Markov chain to predict the land use transfer planning for each scenario to be predicted, to obtain scenario land cover data; Based on the scenario land cover data, a predicted carbon storage corresponding to the scenario land cover data is determined.

2. The method for predicting and simulating regional carbon storage based on a large language model according to claim 1, characterized in that: The step of obtaining historical land cover data of the simulation area and determining the initial carbon storage and initial land use transfer matrix of the simulation area based on the historical land cover data includes: Acquire initial input information, and use a large language model to perform semantic extraction on the initial input information to obtain initial input features; Matching the initial input feature with preset regional land information, determining regional information corresponding to the initial input information, and using the area corresponding to the regional information as the simulation area; Acquiring corresponding historical land cover data based on the regional information; The initial carbon storage of the simulation area is determined based on the land cover data of the current year in the historical land cover data, and the initial land use transfer matrix is ​​determined based on the historical land cover data.

3. The method for predicting and simulating regional carbon storage based on a large language model according to claim 1, characterized in that: For each scenario to be predicted, the initial land use transfer matrix is ​​scenario modified to obtain a scenario land transfer matrix, including: Acquire input scenario restriction information, and use a large language model to perform semantic analysis on the scenario restriction information to obtain a scenario development type corresponding to the scenario restriction information; the scenario development type is a future development type corresponding to the simulation area; Determining a scenario factor for scenario correction of the initial land use transfer matrix based on the scenario development type; Based on the scenario factors, the initial land use transfer matrix is ​​scenario modified to obtain the scenario land transfer matrix.

4. The method for predicting and simulating regional carbon storage based on a large language model according to claim 1, characterized in that: For each scenario to be predicted, the initial land use transfer matrix is ​​scenario modified to obtain a scenario land transfer matrix, including: Acquire input scenario restriction information, and use a large language model to perform semantic analysis on the scenario restriction information to obtain a scenario type corresponding to the scenario restriction information; Based on the scenario type, using the large language model to determine simulation parameters and driving factors corresponding to the scenario type, and based on the simulation parameters and driving factors to determine the scenario factor corresponding to the scenario type; Based on the scenario factors, the initial land use transfer matrix is ​​scenario modified to obtain the scenario land transfer matrix.

5. The method for predicting and simulating regional carbon storage based on a large language model according to claim 1, characterized in that: Also includes: Determining land use transfer data of the simulation area based on the historical land cover data and the corresponding initial carbon storage and the scenario land cover data and the corresponding predicted carbon storage; The land use transfer data at least includes the total transfer amount, transfer ratio and transfer direction of each type of land; Based on the initial carbon stock, the predicted carbon stock and the land use transfer data, a land use transfer planning suggestion corresponding to the simulation area is generated.

6. The method for predicting and simulating regional carbon storage based on a large language model according to claim 5, characterized in that: The generating of land use transfer planning suggestions corresponding to the simulation area based on the initial carbon stock, the predicted carbon stock and the land use transfer data comprises: When the predicted carbon stock corresponding to any of the scenarios to be predicted is greater than the initial carbon stock, recommending the scenario land cover data corresponding to the scenario to be predicted as the land use transfer planning suggestion; or When the predicted carbon stocks corresponding to all the scenarios to be predicted are less than or equal to the initial carbon stocks, feedback requests to adjust the scenario factor for at least one scenario to be predicted to obtain an adjusted scenario factor; The corresponding predicted carbon storage is reacquired based on the adjusted scenario factors until the obtained predicted carbon storage is greater than the initial carbon storage.

7. A device for predicting and simulating regional carbon storage based on a large language model, characterized in that: include: An acquisition module, used for acquiring historical land cover data of a simulation area using a large language model, and determining an initial carbon storage and an initial land use transfer matrix of the simulation area based on the historical land cover data; A correction module, for performing scenario correction on the initial land use transfer matrix for each scenario to be predicted, to obtain a scenario land transfer matrix; A prediction module, for performing land use transfer planning prediction for each scenario to be predicted based on the scenario land transfer matrix using a Markov chain to obtain scenario land cover data; A determination module is used to determine the predicted carbon storage corresponding to the scenario land cover data based on the scenario land cover data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for predicting and simulating regional carbon storage based on a large language model as described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for predicting and simulating regional carbon storage based on a large language model as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting and simulating regional carbon storage based on a large language model as described in any one of claims 1 to 6 is implemented.

Images