A carbon storage prediction method, device, equipment and medium

Abstract

The present invention relates to the field of carbon storage technology, and more specifically, to a carbon storage prediction method, apparatus, device, and medium. The present invention first applies a prediction model to the characteristic values ​​of each grid. The prediction model learns the intrinsic correlation between each grid based on the characteristic values ​​of each grid, and predicts the future carbon storage of each grid based on this correlation, thereby obtaining an initial carbon storage prediction for that grid. Since grids of different scenario types have different effects on their carbon storage, after obtaining the initial prediction, the present invention adjusts the initial prediction based on the scenario type, thereby obtaining a final prediction with a higher accuracy. Furthermore, since the present invention takes into account the intrinsic correlation between each grid, the accuracy of the final prediction is further improved.

Description

Technical Field

[0001] The present invention relates to the field of carbon reserve technology, and in particular to a carbon reserve prediction method, device, equipment and medium. Background Art

[0002] Carbon storage refers to the total mass of carbon elements stored in an ecosystem. Carbon storage can be stored in different carbon reservoirs such as forests, wetlands, soil, oceans, atmosphere and fossil fuels. Predicting future carbon storage is conducive to predicting future climate change and planning environmental protection measures in advance. Existing technologies often use field measurements of carbon storage in different historical periods in various places, and artificially estimate the change pattern of carbon storage over time based on the carbon storage in different historical periods, and then predict the carbon storage in various places based on this change pattern. Due to the limited data that can be obtained from field measurements, the estimated change pattern is not enough to accurately reflect the change pattern of carbon storage. In addition, manual estimation cannot explore the intrinsic correlation between data that can affect carbon storage. Therefore, the carbon storage estimated by existing technologies based on field measurements is less accurate.

[0003] In summary, the carbon storage estimated by existing technologies is less accurate.

[0004] Therefore, the existing technology still needs to be improved and enhanced. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a carbon reserve prediction method, device, equipment and medium, which solves the problem of low accuracy of carbon reserves estimated by the prior art.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a carbon reserve prediction method, comprising:

[0008] Obtaining characteristic values ​​of each grid, and applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model, and obtaining an initial predicted amount of carbon storage for each grid;

[0009] A scenario type is obtained for each of the grids, where the scenario type is used to characterize measures that affect carbon storage, and based on the scenario type, an initial predicted amount of carbon storage for each of the grids is adjusted to obtain a final predicted amount of carbon storage for each of the grids.

[0010] In one implementation, applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model, and obtaining an initial predicted carbon storage of each grid includes:

[0011] determining a tree characteristic value and a forestland characteristic value among the characteristic values;

[0012] Applying the prediction model to the tree characteristic value and the forestland characteristic value of each of the grids, so as to predict the probability of each of the grids belonging to each carbon storage level by the prediction model, and obtain a predicted probability;

[0013] An initial predicted amount of carbon storage for each grid is obtained by using the prediction model based on the prediction probabilities and carbon storage levels corresponding to each grid.

[0014] In one implementation, the prediction model is a convolutional neural network model, which includes an embedding layer, a first convolutional layer, a first maximum pooling layer, a second convolutional layer, a second maximum pooling layer, a third convolutional layer, a global maximum pooling layer, a flattening layer, a first fully connected layer, and a second fully connected layer, which are cascaded in sequence.

[0015] In one implementation, adjusting the initial predicted carbon storage amount of each grid based on the scenario type to obtain the final predicted carbon storage amount of each grid includes:

[0016] When the scene type of the grid is a natural scene, the initial predicted carbon storage amount of the grid is used as the final predicted carbon storage amount;

[0017] Alternatively, when the scenario type of the grid is a planning scenario, zero is used as the final predicted carbon storage amount of the grid;

[0018] Alternatively, when the scenario type of the grid is an ecological restoration scenario, the carbon storage improvement rate of the grid under the action of ecological restoration measures is determined, and based on the carbon storage improvement rate and the initial carbon storage prediction, the final carbon storage prediction of the grid is obtained.

[0019] In one implementation, applied to the optimization scenario type, the carbon storage prediction method further includes:

[0020] Determining a target grid among the grids, wherein the target grid is a target grid selected from an area of ​​a scene type to be formulated;

[0021] Based on the final predicted amount of carbon storage in the target grid, the scenario type of the region is optimized to obtain a scenario optimization type.

[0022] In one implementation, the target grid is selected by:

[0023] Divide the area where the scene type is to be formulated into several grids;

[0024] A random selection algorithm is used to select the target grid from several grids.

[0025] In one implementation, the target grid is selected by:

[0026] Divide the area where the scene type is to be formulated into several grids;

[0027] Obtain the historical carbon storage and / or connectivity of several grids;

[0028] A target grid is selected from the plurality of grids according to the respective historical carbon storage and / or the connectivity of the plurality of grids.

[0029] In a second aspect, an embodiment of the present invention further provides a carbon storage prediction device, wherein the device includes the following components:

[0030] A first prediction module is configured to obtain characteristic values ​​of each grid, apply a prediction model to the characteristic values ​​of each grid, predict the carbon storage of each grid using the prediction model, and obtain an initial prediction of carbon storage for each grid;

[0031] The second prediction module is used to obtain a scenario type for each of the grids, where the scenario type is used to characterize measures that affect carbon storage, and adjust the initial carbon storage prediction for each of the grids based on the scenario type to obtain a final carbon storage prediction for each of the grids.

[0032] In a third aspect, an embodiment of the present invention further provides a terminal device, wherein the terminal device includes a memory, a processor, and a carbon reserve prediction program stored in the memory and executable on the processor, and when the processor executes the carbon reserve prediction program, the steps of the above-mentioned carbon reserve prediction method are implemented.

[0033] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a carbon reserve prediction program is stored. When the carbon reserve prediction program is executed by a processor, the steps of the above-mentioned carbon reserve prediction method are implemented.

[0034] Beneficial Effects: The present invention first applies a prediction model to the characteristic values ​​of each grid. The prediction model learns the intrinsic correlation between each grid based on the characteristic values ​​of each grid, and predicts the future carbon storage of each grid based on this correlation, obtaining an initial carbon storage prediction for that grid. Because grids of different scenario types have different effects on their carbon storage, the present invention adjusts the initial prediction based on the scenario type after obtaining the initial prediction, thereby obtaining a final prediction with a higher accuracy rate. Moreover, because the present invention takes into account the intrinsic correlation between each grid, the accuracy of the final prediction is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is the overall flow chart of the present invention;

[0036] Figure 2 A structural diagram of the carbon reserve prediction device provided by the present invention;

[0037] Figure 3 This is a block diagram of the internal structure of a terminal device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following is a clear and complete description of the technical solutions of the present invention in conjunction with the embodiments and the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] Research has found that carbon reserves refer to the total mass of carbon elements stored in an ecosystem. Carbon reserves can be stored in different carbon reservoirs such as forests, wetlands, soil, oceans, atmosphere and fossil fuels. Predicting future carbon reserves is conducive to predicting future climate change and planning environmental protection measures in advance. Existing technologies often use field measurements of carbon reserves in different historical periods in various places, and artificially estimate the change pattern of carbon reserves over time based on the carbon reserves in different historical periods, and then predict the carbon reserves in various places based on this change pattern. Due to the limited data that can be obtained from field measurements, the estimated change pattern is not enough to accurately reflect the change pattern of carbon reserves. In addition, manual estimation cannot explore the intrinsic correlation between data that can affect carbon reserves. Therefore, the carbon reserves estimated by existing technologies based on field measurements are less accurate.

[0040] To address the aforementioned technical issues, the present invention provides a carbon storage prediction method, apparatus, device, and medium, addressing the low accuracy of carbon storage estimates in existing technologies. In practice, a prediction model is first applied to the characteristic values ​​of each grid. The prediction model then predicts the carbon storage of each grid, yielding an initial carbon storage estimate for each grid. This initial carbon storage estimate is then adjusted based on the scenario type to yield a final carbon storage estimate for each grid.

[0041] For example, a city is divided into several grids, and the characteristic values ​​related to carbon storage in each grid are collected. Due to the mutual influence between the grids, the characteristic values ​​of all grids are input into the prediction model. Based on the characteristic values ​​of all grids, the prediction model can accurately predict the future carbon storage of each grid. Since the above is based on the carbon storage predicted under ideal conditions, that is, the carbon storage predicted without any human interference. However, in reality, the interference of various human factors will change the scene type of the grid, that is, change the carbon storage of the grid. Therefore, the present invention adjusts the initial carbon storage prediction based on the scene type to improve the carbon storage prediction accuracy of each grid.

[0042] Example 1: This embodiment provides a carbon reserve prediction method, which can be applied to a terminal device, which can be a terminal product with data processing function, such as a computer. Figure 1 As shown in , the carbon storage prediction method specifically includes the following steps:

[0043] S100, obtaining characteristic values ​​of each grid, applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model, and obtaining an initial predicted carbon storage amount of each grid;

[0044] S200 , obtaining a scenario type for each of the grids, where the scenario type is used to characterize measures that affect carbon storage, and adjusting an initial carbon storage prediction for each of the grids based on the scenario type to obtain a final carbon storage prediction for each of the grids.

[0045] The above-mentioned carbon storage prediction method of this embodiment can be applied to green construction planning, that is, the city is divided into various grids, and the above-mentioned carbon storage prediction method is used to predict the final predicted carbon storage amount of each grid in future years. If the predicted amount is lower than the carbon storage threshold, the green area that needs to be increased at the current time is reversely calculated based on the difference between the predicted amount and the carbon storage threshold, so that the final predicted carbon storage amount in future years can reach the carbon storage threshold.

[0046] In this embodiment, step S100 includes the following specific steps S101, S102, and S103:

[0047] S101, determining tree characteristic values ​​and forestland characteristic values ​​in the characteristic values.

[0048] Tree eigenvalues ​​include tree height, species type, and other eigenvalues ​​such as canopy density. Species type is a categorical variable, assigned a numerical label to represent its eigenvalue. Tree height and canopy density are continuous variables, with tree height used directly as the tree eigenvalue, and forest canopy density used directly as the forest eigenvalue. Canopy density is an important indicator of forest structure and environment, widely used in carbon storage estimation. It reflects the ratio of the total projected area of ​​the tree canopy under direct sunlight to the total forest area.

[0049] Forest land characteristic values ​​include forest land type, forest land ownership, and forest land slope. Forest land types include tropical dry forest, bamboo forest, poplar-birch forest, etc. The carbon storage potential of tropical dry forest and poplar-birch forest is higher than that of coniferous forest. The carbon storage of tropical dry forest is mainly concentrated in the trunk biomass.

[0050] S102 : Apply the prediction model to the tree characteristic value and the forestland characteristic value of each grid, so as to predict the probability of each grid belonging to each carbon storage level through the prediction model, and obtain a predicted probability.

[0051] This embodiment discretizes [0, 20] with a step size of 0.1, and obtains 201 carbon storage levels, that is, the carbon storage corresponding to the first carbon storage level is 0, the carbon storage corresponding to the second carbon storage level is 0.1, the carbon storage corresponding to the third carbon storage level is 0.2, and so on.

[0052] The probability of each grid belonging to each carbon storage class is predicted to obtain the predicted probability.

[0053] S103 , obtaining an initial predicted amount of carbon storage for each grid based on the predicted probabilities and carbon storage levels corresponding to each grid using the prediction model.

[0054] ;

[0055] in, , represents the initial prediction of carbon storage, Represents the grid belongs to The probability of a carbon storage level, Representative The carbon storage corresponding to each carbon storage level is Represents the total number of carbon storage levels. Carbon storage can be calculated from biomass. Biomass multiplied by the carbon conversion factor is carbon storage. The carbon conversion factor can be 0.45.

[0056] In this embodiment, the prediction model is a convolutional neural network model, which includes an embedding layer, a first convolutional layer, a first maximum pooling layer, a second convolutional layer, a second maximum pooling layer, a third convolutional layer, a global maximum pooling layer, a flattening layer, a first fully connected layer, and a second fully connected layer, which are cascaded in sequence. Among them, the global maximum pooling layer further processes the output of the third convolutional layer. The cascade means that the output of the previous layer is connected to the input of the next layer, that is, the output of the embedding layer is connected to the input of the first convolutional layer, the output of the first convolutional layer is connected to the input of the first maximum pooling layer, the output of the first maximum pooling layer is connected to the input of the second convolutional layer, the output of the second convolutional layer is connected to the input of the second maximum pooling layer, the output of the second maximum pooling layer is connected to the input of the third convolutional layer, and so on.

[0057] Each grid has 76 eigenvalues. The eigenvalues ​​of all grids are spliced ​​into a two-dimensional matrix. This two-dimensional matrix enters the embedding layer through the input layer of the convolutional neural network model. The embedding layer maps the discrete eigenvalues ​​to a continuous vector space. This continuous vector space not only retains the original eigenvalues ​​but also captures the potential relationship between the eigenvalues ​​of each grid.

[0058] The vector space output by the embedding layer is passed through the first, second, and third convolutional layers connected in series for convolutional feature extraction. Each convolutional layer is a one-dimensional convolution, which can further explore the feature correlations between grids. The convolution kernel length of each convolutional layer is 3, and the number of channels of the first convolutional layer is 32, the second convolutional layer is 64, and the third convolutional layer is 128. The output of the first convolutional layer is input into the first maximum pooling layer through a normalization layer. The output of the first maximum pooling layer is input into the second convolutional layer. The output of the second maximum pooling layer is input into the third convolutional layer. The output of the third convolutional layer is input into the global maximum pooling layer through a normalization layer. The output of the global maximum pooling layer is input into the flattening layer. The output of the flattening layer is input into the first fully connected layer. The output of the first fully connected layer is input into the second fully connected layer. The second fully connected layer outputs the initial carbon storage prediction for each grid.

[0059] The maximum pooling layer has a kernel size of 2 and a stride of 2. This layer compresses sequence length, reduces model complexity, and mitigates overfitting. The feature maps output by the convolutional modules are fed into a flattening layer and fed into an output module consisting of two fully connected layers. The first convolutional module contains 64 neurons and uses the ReLU (Rectified Linear Unit) activation function to extract high-order semantic features. To improve model generalization, a Dropout layer (a regularization technique used in deep learning models) is added between the two convolutional modules to achieve a 50% neuron deactivation rate. The output layer consists of 201 neurons, corresponding to the 201 carbon storage levels. The output is normalized using the Softmax function (a type of activation function) to produce a predicted distribution, representing the probability of each grid cell falling within each carbon storage level range.

[0060] The final fully connected layer in this embodiment outputs a 201-dimensional vector, representing the predicted probability distribution of carbon storage at 201 discrete levels. This vector is normalized by Softmax to obtain the probability distribution of carbon storage levels. This probability distribution represents the probability of the grid belonging to each carbon storage level. The final carbon storage prediction value is obtained by weighted summing the distribution. The result of the weighted summation is the above-mentioned .

[0061] This embodiment also constructs the data set required for training the convolutional neural network model, including:

[0062] The first step is to obtain the city's original forest patch data, which includes nearly 200 characteristic variables such as ecology, structure, and topography. For example, forest patch data can include forest species, forest category, canopy density, dominant tree species, and number of trees per hectare.

[0063] In the second step, the forest patch units were screened for uniqueness, and only grids with characteristic values ​​of each forest patch consistent with the carbon storage value were selected to construct the dataset to determine the consistency and validity of the data in the dataset.

[0064] The third step is to extract the characteristic variables of each grid in all years, retaining only the characteristic variables that are consistent in all years and the redundant characteristic variables that have not changed in six years. Finally, 76 characteristic variables are screened out to construct the data set required for training the convolutional neural network model.

[0065] The 76 feature variables mentioned above can be categorized into two types: categorical variables (such as forestland type, forest species, and ownership) and continuous variables of varying dimensions (such as tree height, canopy density, and slope). Directly inputting these into a convolutional model can lead to issues such as scale inconsistency and gradient instability. Therefore, this paper first employs an embedding mechanism to map each feature value into a vector of a specified length.

[0066] For categorical variables, discrete values ​​are directly mapped to vectors of a specified length (that is, categorical variables are assigned values ​​so that they are mapped to vectors of a specified length). For continuous variables, after ascending sorting, an equidistant binning strategy (10 intervals by default) is employed, with the bin numbers representing the original eigenvalues, which are then mapped to eigenvectors of a specified length. Ultimately, all 76 variables are converted to vector form, and an independent embedding representation is learned for each variable. This embedding scheme effectively adapts to the modeling requirements of single-grid-level samples, providing the basic input structure for subsequent convolutional layers to extract combined features of local variables, thereby improving model generalization and robustness.

[0067] In this embodiment, step S200 adjusts the predicted carbon storage amount in the following manner:

[0068] When the scene type of the grid is a natural scene, the initial predicted amount of carbon storage of the grid is used as the final predicted amount of carbon storage.

[0069] In natural scenarios, there is no human intervention, and the carbon storage of the grid is determined only by the ecosystem. That is, the future carbon storage is only related to the current characteristic value of the grid. Therefore, the carbon storage predicted by the model can be used as the final prediction of the grid.

[0070] When the scenario type of the grid is a planning scenario, zero is used as the final predicted amount of carbon storage in the grid.

[0071] When a grid is designated as planned construction land, the scenario type of the grid is a planning scenario. No vegetation that can produce carbon will be planted in the grid, so the final predicted amount of carbon storage in the future of the grid is zero.

[0072] ;

[0073] The above formula means: Grid When planning a scene, the grid The carbon storage in the year is 0, Represents the planning scenario, using Representative Grid In the annual carbon storage.

[0074] When the scenario type of the grid is an ecological restoration scenario, the carbon storage improvement rate of the grid under the action of ecological restoration measures is determined, and based on the carbon storage improvement rate and the initial carbon storage prediction, the final carbon storage prediction of the grid is obtained.

[0075] The grid is planned as an ecological restoration scenario, and human intervention is carried out to enhance the carbon storage of the grid based on ecological evolution.

[0076] ;

[0077] The above formula means: Grid For ecological restoration scenarios (using represents the ecological restoration scenario), the grid The initial carbon storage forecast for the year is , the grid is in carbon storage per year

[0078] The final prediction is , For the The carbon storage improvement rate of the grid is It is 10%.

[0079] This embodiment also calculates the Grid The neighboring grid (using Representative Grid The final predicted amount of carbon storage in the neighboring grids is Represents the final predicted amount of carbon storage in the adjacent grid:

[0080] ;

[0081] in, is the carbon storage improvement rate of the adjacent grid. 5%.

[0082] In the second embodiment, for some cities, based on existing knowledge, it is impossible to know the scenario type from 2035 to 2060, that is, it is unknown whether the city belongs to the natural scenario, the planned scenario, or the ecological restoration scenario from 2035 to 2060. First, it is assumed that the city belongs to one of the scenarios, and then the carbon storage of the city under this scenario is predicted. Based on the predicted carbon storage, it is determined whether the assumed scenario is conducive to the increase of carbon storage. If not, the scenario is optimized.

[0083] This embodiment includes the following specific steps S300, S400, and S500:

[0084] S300, selecting a target grid from an area of ​​a scenario type to be formulated, obtaining characteristic values ​​of the target grid, applying a prediction model to the characteristic values ​​of the target grid, predicting the carbon storage of each target grid using the prediction model, and obtaining an initial predicted carbon storage amount of each target grid.

[0085] This embodiment uses the same method as the first embodiment to predict the initial predicted amount of carbon storage in the target grid.

[0086] S400, presetting the scenario type of the target grid to obtain a scenario preset type, and adjusting the initial predicted amount of carbon storage of the target grid based on the scenario preset type of the target grid to obtain a final predicted amount of carbon storage of each target grid;

[0087] S500: Optimizing the scenario type of the region based on the final predicted carbon storage amount of the target grid to obtain an optimized scenario type.

[0088] In this embodiment, a random method is used to select a target grid, that is, the area of ​​the scene type to be formulated is divided into a plurality of grids; and a random selection algorithm is used to select a target grid from the plurality of grids.

[0089] ;

[0090] The above formula means: using random selection algorithm from N grids Select M target grids.

[0091] This embodiment may also select a target grid in the following manner: divide the area for the scenario type to be formulated into a plurality of grids; obtain the historical carbon storage and / or connectivity of each of the plurality of grids; and select a target grid from the plurality of grids based on the historical carbon storage and / or connectivity of each of the plurality of grids.

[0092] The connectivity degree represents the number of other grids to which each grid is connected. For example, if a grid is connected to three of the other grids, the connectivity degree of the grid is 3. Connection definition: If the Euclidean distance between the center points of two grids is less than a threshold, the two grids are considered connected; otherwise, no connection is considered between the two grids.

[0093] When the target grid is selected based on the descending order of historical carbon storage and / or connectivity, the following formula is used to select the target grid:

[0094] ;

[0095] The above formula means: sort the historical carbon storage or connectivity of N grids in descending order to obtain a sequence, and select from the head of the sequence target grid, Represents descending sort, Representative Grid historical carbon stocks or connectivity.

[0096] When the target grid is selected based on the ascending order of historical carbon storage and / or connectivity, the following formula is used to select the target grid:

[0097] ;

[0098] The above formula means: Sort the historical carbon storage or connectivity of N grids in ascending order to obtain a sequence, and select from the head of the sequence target grid, Represents ascending sort order.

[0099] In this embodiment, after the target grid is selected, step S500 predicts the final predicted carbon storage amount of each target grid based on the scenario preset type, and then calculates the predicted carbon storage amount of the area where all target grids are located based on the final predicted carbon storage amount of each target grid, and optimizes the scenario type based on the predicted carbon storage amount of the area. That is, if the scenario preset type makes the predicted carbon storage amount of the area less than the threshold, the scenario type of the area is reset.

[0100] ;

[0101] Where, The predicted carbon storage in the representative region, is the existing carbon stock in the region, For the region The final predicted amount of carbon storage in the target grid, Representing the region The target grid and the When the interaction relationship is 0, it means that there is no adjacency relationship between the grids; when the interaction relationship is 1, it means that there is an adjacency relationship between the grids. Representing the region The target grid and the The correlation size of the two grids is used to measure the degree of influence between the two grids.

[0102] In summary, this paper uses an "embedding mechanism + one-dimensional convolutional network" to process non-image urban green space data (e.g., 76 features such as tree species, tree height, and slope). Specifically, the high-dimensional non-image data is first vectorized into a two-dimensional matrix, and then local variables are identified through one-dimensional convolution, improving the model's adaptability and expressiveness for ecological survey data.

[0103] The present invention proposes a grade distribution prediction mechanism, which converts the prediction of carbon quantity into probability distribution modeling at 201 grades. The final prediction value is obtained by weighted summation of the distribution, which improves the robustness of the model to extreme samples and abnormal disturbances.

[0104] Based on natural growth trend predictions, this paper overlays urban planning and ecological restoration scenario intervention mechanisms to construct a multi-scenario simulation model combining "rule control + strategy evolution." By incorporating real-world scenario types into the temporal dimension and introducing multiple grid selection methods into the spatial dimension, this model simulates the long-term dynamic evolution of urban carbon storage systems driven by multiple sources.

[0105] This embodiment also provides a carbon storage prediction device, such as Figure 2 As shown, the device includes the following components:

[0106] The first prediction module 01 is used to obtain the characteristic value of each grid, apply the prediction model to the characteristic value of each grid, predict the carbon storage of each grid by the prediction model, and obtain the initial predicted carbon storage of each grid;

[0107] The second prediction module 02 is used to obtain a scenario type for each grid, where the scenario type is used to characterize measures that affect carbon storage, and adjust the initial carbon storage prediction for each grid based on the scenario type to obtain a final carbon storage prediction for each grid.

[0108] Based on the above embodiment, the present invention further provides a terminal device, whose principle block diagram can be shown as follows: Figure 3 As shown. The terminal device includes a processor, memory, a network interface, and a display screen connected via a system bus. The processor of the terminal device is used to provide computing and control capabilities. The memory of the terminal device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and computer program in the non-volatile storage medium to run. The network interface of the terminal device is used to communicate with an external terminal via a network connection. When executed by the processor, the computer program implements a carbon storage prediction method. The display screen of the terminal device can be a liquid crystal display or an electronic ink display.

[0109] Those skilled in the art will understand that Figure 3 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal device to which the solution of the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0110] In one embodiment, a terminal device is provided. The terminal device includes a memory, a processor, and a carbon storage prediction program stored in the memory and executable on the processor. When the processor executes the carbon storage prediction program, the following operating instructions are implemented:

[0111] Obtaining characteristic values ​​of each grid, and applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model, and obtaining an initial predicted amount of carbon storage for each grid;

[0112] A scenario type is obtained for each of the grids, where the scenario type is used to characterize measures that affect carbon storage, and based on the scenario type, an initial predicted amount of carbon storage for each of the grids is adjusted to obtain a final predicted amount of carbon storage for each of the grids.

[0113] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware using a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A carbon reserve prediction method, characterized in that: include: Obtaining characteristic values ​​of each grid, and applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model, and obtaining an initial predicted amount of carbon storage for each grid; Obtaining a scenario type for each of the grids, the scenario type being used to characterize measures that affect carbon storage, and adjusting an initial carbon storage prediction for each of the grids based on the scenario type to obtain a final carbon storage prediction for each of the grids; Applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model to obtain an initial predicted carbon storage amount of each grid, including: determining a tree characteristic value and a forestland characteristic value among the characteristic values; Applying the prediction model to the tree characteristic value and the forestland characteristic value of each of the grids, so as to predict the probability of each of the grids belonging to each carbon storage level by the prediction model, and obtain a predicted probability; Obtaining an initial predicted amount of carbon storage for each grid based on the predicted probabilities and carbon storage levels corresponding to each grid using the prediction model; Adjusting the initial predicted carbon storage amount of each grid based on the scenario type to obtain the final predicted carbon storage amount of each grid includes: When the scene type of the grid is a natural scene, the initial predicted carbon storage amount of the grid is used as the final predicted carbon storage amount; Alternatively, when the scenario type of the grid is a planning scenario, zero is used as the final predicted carbon storage amount of the grid; Alternatively, when the scenario type of the grid is an ecological restoration scenario, the carbon storage improvement rate of the grid under the action of ecological restoration measures is determined, and based on the carbon storage improvement rate and the initial carbon storage prediction, the final carbon storage prediction of the grid is obtained.

2. The carbon reserve prediction method according to claim 1, wherein: The prediction model is a convolutional neural network model, which includes an embedding layer, a first convolutional layer, a first maximum pooling layer, a second convolutional layer, a second maximum pooling layer, a third convolutional layer, a global maximum pooling layer, a flattening layer, a first fully connected layer, and a second fully connected layer, which are cascaded in sequence.

3. The carbon reserve prediction method according to any one of claims 1 to 2, characterized in that: Applied to the optimization scenario type, the carbon storage prediction method further includes: Determining a target grid among the grids, wherein the target grid is a target grid selected from an area of ​​a scene type to be formulated; Based on the final predicted amount of carbon storage in the target grid, the scenario type of the region is optimized to obtain a scenario optimization type.

4. The carbon reserve prediction method according to claim 3, wherein: The target grid selection method includes: Divide the area where the scene type is to be formulated into several grids; A random selection algorithm is used to select the target grid from several grids.

5. The carbon reserve prediction method according to claim 3, wherein: The target grid selection method includes: Divide the area where the scene type is to be formulated into several grids; Obtain the historical carbon storage and / or connectivity of several grids; A target grid is selected from the plurality of grids according to the respective historical carbon storage and / or the connectivity of the plurality of grids.

6. A carbon reserve prediction device, characterized in that: The device comprises the following components: A first prediction module is configured to obtain characteristic values ​​of each grid, apply a prediction model to the characteristic values ​​of each grid, predict the carbon storage of each grid using the prediction model, and obtain an initial prediction of carbon storage for each grid; a second prediction module, configured to obtain a scenario type for each of the grids, the scenario type being used to characterize measures that affect carbon storage, and adjust an initial carbon storage prediction for each of the grids based on the scenario type to obtain a final carbon storage prediction for each of the grids; Applying a prediction model to the characteristic values ​​of each grid, predicting the carbon storage of each grid using the prediction model to obtain an initial predicted carbon storage amount of each grid, including: determining a tree characteristic value and a forestland characteristic value among the characteristic values; Applying the prediction model to the tree characteristic value and the forestland characteristic value of each of the grids, so as to predict the probability of each of the grids belonging to each carbon storage level by the prediction model, and obtain a predicted probability; Obtaining an initial predicted amount of carbon storage for each grid based on the predicted probabilities and carbon storage levels corresponding to each grid using the prediction model; Adjusting the initial predicted carbon storage amount of each grid based on the scenario type to obtain the final predicted carbon storage amount of each grid includes: When the scene type of the grid is a natural scene, the initial predicted carbon storage amount of the grid is used as the final predicted carbon storage amount; Alternatively, when the scenario type of the grid is a planning scenario, zero is used as the final predicted carbon storage amount of the grid; Alternatively, when the scenario type of the grid is an ecological restoration scenario, the carbon storage improvement rate of the grid under the action of ecological restoration measures is determined, and based on the carbon storage improvement rate and the initial carbon storage prediction, the final carbon storage prediction of the grid is obtained.

7. A terminal device, characterized in that: The terminal device includes a memory, a processor, and a carbon storage prediction program stored in the memory and executable on the processor. When the processor executes the carbon storage prediction program, the steps of the carbon storage prediction method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a carbon storage prediction program, and when the carbon storage prediction program is executed by the processor, the steps of the carbon storage prediction method according to any one of claims 1 to 5 are implemented.

Images