A method for constructing a geographical element prediction model, a prediction method and related devices

Abstract

The present invention discloses a method for constructing a geographic element prediction model, a prediction method and related devices, which are applied to the field of geographic information technology, including: constructing a sample; the sample includes an environmental variable grid, a first distance grid and a geographic element of a sampling point in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved with a preset window and a window movement strategy; the preset geographic element prediction model is trained through the sample to obtain a trained geographic element prediction model. The beneficial effects of the present invention are: a machine learning model is used to fully integrate known environmental variable information and the first law of geography, and has stronger modeling capabilities and higher prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the field of geographic information technology, and in particular to a geographic element prediction model construction method, a geographic element prediction method, a geographic element prediction device and a computer-readable storage medium. Background Art

[0002] The first law of geography states that "everything is related, but things that are close are more closely related". Therefore, the classic regression model is not particularly suitable for modeling non-static spatiotemporal relationships. Therefore, sliding window regression, geographically weighted regression, hybrid geographically weighted regression, spatiotemporal geographically weighted regression, non-Euclidean geographically weighted regression, spatial attribute geographically weighted regression, multi-scale geographically weighted regression, geographical neural network weighted regression and other spatiotemporal local modeling methods came into being. Spatiotemporal local modeling methods can be used to model non-static spatiotemporal relationships and have been widely used in soil organic carbon prediction, modeling of the relationship between surface temperature and environmental factors, precipitation interpolation, temperature interpolation, radar-rain gauge joint precipitation estimation, multi-source data fusion quantitative precipitation estimation, numerical model forecast product interpretation, etc., and show characteristics that are superior to many traditional methods. However, when modeling, some existing spatiotemporal statistical methods require parameterized weight functions, and some only consider the environmental variable information of the sample location, which leads to the inability to fully integrate the known environmental variable information and the first law of geography.

[0003] Therefore, how to provide a geographical element prediction technology solution that can fully integrate known environmental variable information and the first law of geography is an urgent problem that technicians in this field need to solve. Summary of the invention

[0004] The purpose of the present invention is to provide a method for constructing a geographic element prediction model, which can fully integrate known environmental variable information and the first law of geography; another purpose of the present invention is to provide a geographic element prediction method, a geographic element prediction device and a computer-readable storage medium, which can fully integrate known environmental variable information and the first law of geography.

[0005] In order to solve the above technical problems, the present invention provides a method for constructing a geographic element prediction model, comprising:

[0006] Constructing a sample; the sample includes an environmental variable grid, a first distance grid and geographic elements of sampling points in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved according to the preset window and window movement strategy;

[0007] The preset geographic element prediction model is trained through the samples to obtain a trained geographic element prediction model.

[0008] Optionally, the construction sample includes:

[0009] Acquire data required for geographic element prediction, and construct the environmental variable grid according to the distribution data of environmental variables; the data required for geographic element prediction includes the distribution data of the environmental variables within the target range and the geographic elements of the sampling points;

[0010] Get the control point grid composed of control points within the target range;

[0011] The distances between the sampling points in the window and each control point in the window are determined, and a first distance grid is constructed.

[0012] Optionally, before determining the distances between the sampling points in the window and each control point in the window and constructing the first distance grid, the method further includes:

[0013] The preset window is moved according to a preset window moving strategy, and sampling points in the window that meet preset standards are selected.

[0014] Optionally, the distance includes any one or any combination of the following:

[0015] Distance based on space, distance based on time, distance based on geographical attributes.

[0016] Optionally, after obtaining the data required for geographic element prediction, the method further includes:

[0017] The data required for the prediction of the geographic elements is preprocessed.

[0018] Optionally, the preprocessing includes any one or any combination of the following:

[0019] Remove and / or replace abnormal data, encode the geographic elements or normalize or standardize the geographic elements, and encode the environmental variables.

[0020] Optionally, after constructing the environment variable grid, the following steps are further included:

[0021] The environmental variables are dimensionality reduced, and / or the environmental variables are normalized or standardized.

[0022] The present invention also provides a geographical element prediction method, which uses the geographical element prediction model trained as described in any one of the above items, comprising:

[0023] Constructing a model input; the model input includes an environmental variable grid and a second distance grid in the same window; the second distance grid is a grid constructed according to the distance between the target grid point in the window and each control point in the window when the window is moved according to the preset window and window movement strategy;

[0024] The geographic element of the target grid point is obtained according to the model input and the trained geographic element prediction model.

[0025] The present invention also provides a geographical element prediction device, comprising:

[0026] Memory for storing computer programs;

[0027] A processor is used to execute the computer program to implement the steps of the method for constructing a geographic element prediction model as described in any of the above items, and / or the steps of the method for predicting geographic elements as described in any of the above items.

[0028] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program is executed by a processor to implement the steps of the method for constructing a geographic element prediction model as described in any of the above items, and / or the steps of the geographic element prediction method as described in any of the above items.

[0029] The present invention provides a method for constructing a geographic element prediction model, comprising: constructing a sample; the sample comprises an environmental variable grid, a first distance grid and geographic elements of a sampling point in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved with a preset window and a window movement strategy; and training a preset geographic element prediction model through the sample to obtain a trained geographic element prediction model.

[0030] The beneficial effects of the present invention are: using a machine learning model to fully integrate known environmental variable information and the first law of geography, it has stronger modeling capabilities and higher prediction accuracy.

[0031] The present invention also provides a geographic element prediction method, a geographic element prediction device and a computer-readable storage medium, which also have the above-mentioned beneficial effects and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the following briefly introduces the drawings to be used in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 A flowchart of a method for constructing a geographic element prediction model provided by an embodiment of the present invention;

[0034] Figure 2 A flowchart of a specific method for constructing a geographic element prediction model provided by an embodiment of the present invention;

[0035] Figure 3 It is a schematic diagram of target range, sampling points, target grid and window;

[0036] Figure 4 It is the target range, control point grid, and window schematic diagram;

[0037] Figure 5 Schematic diagram of the solution process;

[0038] Figure 6 A structural block diagram of a device for constructing a geographic element prediction model provided by an embodiment of the present invention;

[0039] Figure 7 A flowchart of a geographic element prediction method provided by an embodiment of the present invention;

[0040] Figure 8 A flowchart of a specific geographical element prediction method provided by an embodiment of the present invention;

[0041] Fig. 9 A structural block diagram of a geographic element prediction device provided by an embodiment of the present invention;

[0042] Fig.10 A structural block diagram of a geographic element prediction device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The core of the present invention is to provide a method for building a geographic element prediction model. In the prior art, some existing spatiotemporal statistical methods require parameterized weight functions when modeling, while others only consider the environmental variable information of the sample location, which results in the inability to fully integrate the known environmental variable information and the first law of geography.

[0044] The present invention provides a method for constructing a geographic element prediction model, comprising: constructing a sample; the sample comprises an environmental variable grid, a first distance grid and geographic elements of a sampling point in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved with a preset window and a window movement strategy; and training a preset geographic element prediction model through the sample to obtain a trained geographic element prediction model.

[0045] The beneficial effects of the present invention are: using a machine learning model to fully integrate known environmental variable information and the first law of geography, it has stronger modeling capabilities and higher prediction accuracy.

[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention.

[0047] Please refer to Figure 1 , Figure 1 A flowchart of a method for constructing a geographic element prediction model provided by an embodiment of the present invention.

[0048] See also Figure 1 In an embodiment of the present invention, a method for constructing a geographic element prediction model includes:

[0049] S101: Build a sample.

[0050] In an embodiment of the present invention, the sample includes an environmental variable grid, a first distance grid and geographical elements of sampling points in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved with the preset window and window movement strategy.

[0051] The environmental variable values ​​corresponding to each grid point in the above environmental variable grid can be uniformly taken as the environmental variable values ​​at the corresponding positions, or can be uniformly taken as the statistical values ​​of the corresponding environmental variables within a certain time and space range of each grid point, such as the maximum value, minimum value, average value, median value, etc.

[0052] In the embodiment of the present invention, one environmental variable grid usually corresponds to one environmental variable, and the sample usually includes multiple environmental variable grids, and different environmental variable grids correspond to different environmental variables. It should be noted that in the embodiment of the present invention, different environmental variable grids can have different resolutions, which can be determined based on factors such as the geographic element prediction model to be adopted and the characteristics of the environmental variables. The specific types of the geographic elements of the above-mentioned sampling points are set according to actual needs and are not specifically limited here.

[0053] When constructing the first distance grid, the control point grid needs to be determined first. The control point grid is a grid constructed according to the preset control points, and the control points are points set by themselves in the target range. Specifically, the target range can be a spatiotemporal range of any dimension greater than or equal to one dimension. When calculating the first distance grid, a window needs to be pre-set, and the window needs to have a preset shape, size and a certain movement strategy. The movement strategy includes sequential movement along the grid direction, or movement according to the sampling point, or random movement, which is not specifically limited here. Specifically, in the embodiment of the present invention, the movement strategy of the above window can be: window by window, full coverage of the target range, minimum overlap and the window does not go out of the target range, which can be set according to the actual situation. The shape and size of the above window can be determined in combination with factors such as the model to be adopted, the size of the target range, the density of sampling points, the spatiotemporal range related to environmental variables and geographical elements, and the hardware conditions available.

[0054] When moving the window, the environmental variable grid in the window after each movement can be recorded, including the environmental variable values ​​corresponding to each grid point; and the distance between the sampling point in the window and each control point in the window can be calculated to construct a first distance grid. When constructing the first distance grid, usually one sampling point corresponds to more than one first distance grid. For the selection of sampling points, each sampling point in the window can be selected, and the corresponding first distance grid can be calculated for each sampling point; or sampling points that meet the preset requirements can be selected, for example, sampling points located in the center area of ​​the window are selected, and only the first distance grid corresponding to the sampling points located in the center area of ​​the window can be calculated.

[0055] The above distance specifically includes any one or any combination of the following: distance based on space, distance based on time, and distance based on geographical attributes. That is, when calculating the distance and constructing the first distance grid, the distance between the calculated sampling point and the control point can specifically include distance based on space, such as Euclidean distance, Manhattan distance, Chebyshev distance, Minkowski distance, and distance in a certain coordinate direction; it can also include distance based on time, such as periodic time distance and non-periodic time distance between the sampling point and the control point; it can also include distance based on geographical attributes, such as the distance between the sampling point and the control point calculated based on geographical attributes such as wind direction, etc. When calculating the distance by geographical attributes, the geographical attribute values ​​corresponding to the sampling point and the control point can be uniformly taken as the geographical attribute values ​​of the corresponding positions, or can be uniformly taken as the statistical values ​​of the corresponding geographical attributes within a certain time and space range of each sampling point or control point, such as maximum value, minimum value, average value, median value, etc. Generally, a sampling point can correspond to multiple distances, and a distance usually corresponds to a first distance grid, so that a sampling point can correspond to multiple first distance grids, and the difference between multiple first distance grids corresponding to the same sampling point is usually the type of distance. For example, a sampling point can correspond to two first distance grids, one first distance grid is constructed by the Euclidean distance between the sampling point and the control point in the same window, and the other first distance grid can be constructed by the distance calculated based on geographic attributes.

[0056] The above window moving strategy, sampling point selection criteria, and distance types can be set according to actual conditions and are not specifically limited here. Usually, a geographical element corresponding to a sampling point, one or more first distance grids, and one or more environmental variable grids are taken as a sample. When moving the window, multiple samples are generated to train the model, thereby ensuring the accuracy of the trained model.

[0057] S102: training a preset geographic element prediction model through samples to obtain a trained geographic element prediction model.

[0058] Before training, it is usually necessary to preset the geographic element prediction model to be adopted, including setting the training strategy, hyperparameters, etc. of the geographic element prediction model. For example, if the geographic element prediction model to be adopted is a convolutional neural network, the above training strategy includes but is not limited to parameter initialization strategy, parameter optimization strategy, learning rate strategy, loss function, one or a combination of overfitting judgment indicators, and the above hyperparameters include but are not limited to epoch maximum value, batch size, learning rate, initial learning rate, maximum learning rate, dropout probability, one or a combination of the upper limit of the overfitting judgment indicator.

[0059] Afterwards, the preset geographic element prediction model will be trained based on the samples constructed in S101 to obtain a trained geographic element prediction model. The above-mentioned geographic element prediction model can be a convolutional neural network (such as LeNet, AlexNet, VGGNet, GoogleLeNet, ResNet, DenseNet, NIN, CapsNet, etc.), a fully connected neural network, a random forest, AdaBoost, GBDT, XGBoost, LightGBM and other well-known machine learning models and their variants, which are not specifically limited here.

[0060] The above-mentioned geographic element prediction model can specifically be a neural network model, a regression model, or a random forest model, which is not specifically limited here. It should be noted that the above-mentioned environmental variable grid and the control point grid can have the same specifications or different specifications, which can be determined according to factors such as the geographic element prediction model to be adopted. When the geographic element prediction model is specifically a convolutional neural network model, the above-mentioned control point grid and the environmental variable grid usually need to have the same specifications, that is, each grid point in the environmental variable grid usually needs to have a corresponding control point. Preferably, the above-mentioned environmental variable grid and the control point grid can be completely identical in time and space position.

[0061] For the specific content of model training, please refer to the existing technology and will not be elaborated here.

[0062] A method for constructing a geographic element prediction model provided by an embodiment of the present invention includes: constructing a sample; the sample includes an environmental variable grid, a first distance grid and geographic elements of a sampling point in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved with a preset window and a window movement strategy; and training a preset geographic element prediction model through the sample to obtain a trained geographic element prediction model.

[0063] The beneficial effects of the present invention are: using a machine learning model to fully integrate known environmental variable information and the first law of geography, it has stronger modeling capabilities and higher prediction accuracy.

[0064] The specific contents of the method for constructing a geographic element prediction model provided by the present invention will be described in detail in the following invention embodiments.

[0065] Please refer to Figures 2 to 5 , Figure 2 A flowchart of a specific method for constructing a geographic element prediction model provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of target range, sampling points, target grid and window; Figure 4 It is the target range, control point grid, and window schematic diagram; Figure 5 Schematic diagram of the solution process.

[0066] See also Figure 2 In an embodiment of the present invention, a method for constructing a geographic element prediction model includes:

[0067] S201: Acquire data required for geographic element prediction, and construct an environmental variable grid based on the distribution data of environmental variables.

[0068] See also Figures 3 to 5 In the embodiment of the present invention, the data required for the geographic element prediction includes the distribution data of the environmental variables within the target range and the geographic elements of the sampling points. The type and source of the data obtained in this step can be set according to the actual situation and are not specifically limited here.

[0069] In this step, it is necessary to construct an environmental variable grid according to the obtained distribution data of the environmental variables. For details on how to construct the environmental variable grid, please refer to S101, which will not be described in detail here. The above sampling points are preset sampling points in the target range.

[0070] Specifically, after this step, the method may further include: preprocessing the data required for the prediction of the geographic elements. Specifically, the preprocessing includes any one or any combination of the following: removing and / or replacing abnormal data, encoding the geographic elements or normalizing or standardizing the geographic elements, and encoding the environmental variables.

[0071] The above-mentioned elimination and / or replacement of abnormal data can screen the data obtained in S201, and the identification criteria for abnormal data can be set according to actual conditions. When eliminating and / or replacing data, only a single data can be eliminated and / or replaced, or the entire group of data can be eliminated and / or replaced, which is not specifically limited here.

[0072] The purpose of normalizing or standardizing geographic elements is to facilitate model training and numerical calculation. The specific process of normalization or standardization can refer to the prior art and will not be described in detail here.

[0073] The above encoding of environmental variables can unify the types of environmental variables. For example, for categorized environmental variables, in an embodiment of the present invention, one-hot encoding or other encoding can be performed on them, so as to facilitate subsequent calculations. For categorized geographical elements, in an embodiment of the present invention, one-hot encoding can also be performed on their categories, so as to facilitate model training.

[0074] Furthermore, in an embodiment of the present invention, after constructing the environmental variable grid, the following may be included: reducing the dimension of the environmental variables, and / or normalizing or standardizing the environmental variables. That is, after constructing the environmental variable grid, the environmental variables may be reduced in dimension by principal component analysis, kernel principal component analysis, neural network (such as autoencoder), manifold learning (such as local linear embedding), and the like.

[0075] The purpose of normalizing or standardizing the environmental variables is to facilitate model training and numerical calculation. The specific process of normalization or standardization can refer to the prior art and will not be described here. Specifically, the normalization process, including the normalization process of environmental variables and geographical elements, can be performed by Min-Max normalization and other methods.

[0076] S202: Acquire a control point grid composed of control points within the target range.

[0077] Before this step, it is usually necessary to construct a control point grid, which usually includes a plurality of regularly distributed control points. Correspondingly, in this step, it is necessary to obtain the control point grid. It should be noted that for different types of distances, the control point grid obtained in this step may have different resolutions.

[0078] S203: Determine the distance between the sampling point in the window and each control point in the window, and construct a first distance grid.

[0079] Before this step, it is necessary to preset the window, including the shape, size and movement strategy of the window. In this step, the window will be moved according to the preset movement strategy. When moving the window, the distance between the sampling point in the window and each control point in the window will be calculated after each window movement to construct a first distance grid. The specific content of the first distance grid has been described in detail in the above-mentioned embodiment of the invention, and will not be repeated here.

[0080] Specifically, in this step, the preset window can be first moved with a preset window moving strategy, and sampling points in the window that meet the preset criteria can be selected. That is, the sampling points can be further screened in the step, and the screening criteria can be set according to actual conditions. For example, sampling points located in the edge area of ​​the window are excluded, and only sampling points located in the center area of ​​the window are processed, etc. It should be noted that in the embodiment of the present invention, the first distance grid can be further normalized or standardized, that is, the distance in each first distance grid is normalized or standardized, so as to facilitate the training of certain models. The specific content of normalization or standardization can refer to the prior art, and will not be repeated here.

[0081] Furthermore, in the embodiment of the present invention, the window can be set to be as large as the target range, so that the window does not need to be moved. When the target range is large, the target range can be divided into several sub-target ranges for modeling and prediction respectively.

[0082] Furthermore, in the embodiment of the present invention, the samples can be further expanded by window rotation and other methods, so as to improve the generalization ability of the trained geographic element prediction model. The specific content of window rotation can be referred to the prior art, and will not be repeated here.

[0083] S204: training a preset geographic element prediction model through samples to obtain a trained geographic element prediction model.

[0084] This step is basically the same as S102 in the above-mentioned invention embodiment. For details, please refer to the above-mentioned invention embodiment, which will not be repeated here. In the embodiment of the present invention, the model used can be a convolutional neural network, which usually includes a convolution layer, a pooling layer, a fully connected layer, a hidden layer, a batch normalization layer, etc.; further, the training strategies such as the He parameter initialization strategy, the small batch stochastic gradient descent parameter optimization strategy, the fixed learning rate strategy, and the mean square error loss function can be set; specifically, the maximum epoch value can be set to 10000, the batch size can be set to 8, the learning rate can be set to 0.1, and the dropout probability can be set to 0.8.

[0085] A method for constructing a geographic element prediction model provided by an embodiment of the present invention uses a machine learning model to fully integrate known environmental variable information and the first law of geography, thereby having stronger modeling capabilities and higher prediction accuracy.

[0086] A geographic element prediction model construction device provided in an embodiment of the present invention is introduced below. The geographic element prediction model construction device described below and the geographic element prediction model construction method described above can be referenced to each other.

[0087] Please refer to Figure 6 , Figure 6 This is a structural block diagram of a device for constructing a geographic element prediction model provided by an embodiment of the present invention. Figure 6 , the geographic element prediction model building device may include:

[0088] The sample module 100 is used to construct a sample; the sample includes an environmental variable grid, a first distance grid and geographic elements of sampling points in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved according to the preset window and window movement strategy;

[0089] The training module 200 is used to train the preset geographic element prediction model through the samples to obtain a trained geographic element prediction model.

[0090] Preferably, in the embodiment of the present invention, the sample module 100 includes:

[0091] The data acquisition unit is used to acquire the data required for geographic element prediction and construct the environmental variable grid according to the distribution data of environmental variables; the data required for geographic element prediction includes the distribution data of the environmental variables within the target range and the geographic elements of the sampling points.

[0092] A control point grid unit is used to obtain a control point grid composed of control points within a target range;

[0093] The first distance grid unit is used to determine the distance between the sampling point in the window and each control point in the window, and to construct a first distance grid.

[0094] Preferably, in the embodiment of the present invention, the sample module 100 further includes:

[0095] The sampling point selection unit is used to move the preset window according to a preset window movement strategy, and select sampling points in the window that meet preset standards.

[0096] Preferably, in the embodiment of the present invention, the distance includes any one or any combination of the following:

[0097] Distance based on space, distance based on time, distance based on geographical attributes.

[0098] Preferably, in the embodiment of the present invention, the sample module 100 further includes:

[0099] The preprocessing unit is used to preprocess the data required for the prediction of the geographic elements.

[0100] Preferably, in an embodiment of the present invention, the preprocessing includes any one or any combination of the following:

[0101] Remove and / or replace abnormal data, encode the geographic elements or normalize or standardize the geographic elements, and encode the environmental variables.

[0102] Preferably, in the embodiment of the present invention, the preprocessing unit is further used for:

[0103] The environmental variables are dimensionality reduced, and / or the environmental variables are normalized or standardized.

[0104] The geographic element prediction model construction device of this embodiment is used to implement the aforementioned geographic element prediction model construction method. Therefore, the specific implementation method of the geographic element prediction model construction device can be seen in the embodiment part of the geographic element prediction model construction method in the previous text. For example, the sample module 100 and the training module 200 are respectively used to implement steps S101 to S102 in the above-mentioned geographic element prediction model construction method. Therefore, its specific implementation method can refer to the description of the corresponding embodiments of each part, which will not be repeated here.

[0105] A geographic element prediction method will be provided below, and the specific content of the geographic element prediction method can be referenced to the content in the above-mentioned embodiment of the invention of the geographic element prediction model construction method.

[0106] Please refer to Figure 7 , Figure 7 A flowchart of a geographic element prediction method provided in an embodiment of the present invention. First of all, it should be noted that the geographic element prediction model trained in the above-mentioned geographic element prediction model construction method is required in the embodiment of the present invention. The specific content of the trained geographic element prediction model has been described in detail in the above-mentioned geographic element prediction model construction method invention embodiment, and will not be repeated here.

[0107] See also Figure 7 In an embodiment of the present invention, the geographical element prediction method includes:

[0108] S301: Construct model input.

[0109] In an embodiment of the present invention, the model input includes an environmental variable grid and a second distance grid in the same window; the second distance grid is a grid constructed according to the distance between the target grid point in the window and each control point in the window when the window is moved with the preset window and window movement strategy.

[0110] In the embodiment of the present invention, it is finally necessary to output the geographic elements corresponding to the target grid points. Specifically, the data that need to be input to the geographic element prediction model constructed in the embodiment of the present invention include the environmental variable grid in the window and the second distance grid. The specific content of the environmental variable grid in the window is the same as the environmental variable grid in the window introduced in the embodiment of the invention of the method for constructing the geographic element prediction model, and will not be repeated here.

[0111] In the embodiment of the present invention, the second distance grid input into the geographic element prediction model during prediction is similar to the first distance grid input into the geographic element prediction model during training in the above-mentioned embodiment of the invention of the method for constructing a geographic element prediction model. In the embodiment of the present invention, the second distance grid input into the geographic element prediction model during prediction specifically requires replacing the sampling points with the target grid points, and calculating the distance between each control point in the window and the target grid points.

[0112] Specifically, when constructing the second distance grid in the embodiment of the present invention, a window also needs to be preset. Generally, the shape and size of the window are the same as the window preset when constructing the sample in the embodiment of the invention of the method for constructing the geographic element prediction model, but the moving strategy can be different. When moving the window, the environmental variable grid in the window after each movement can be recorded, including the environmental variable values ​​corresponding to each grid point of the environmental variable grid in the window; and the distance between the target grid point in the window and each control point in the window is calculated to construct the second distance grid. When constructing the second distance grid, one target grid point can correspond to one or more second distance grids. The difference between multiple second distance grids corresponding to the same target grid point is usually different in the type of distance. The specific content of the above distance has been described in detail in the embodiment of the invention, and will not be repeated here. Generally, one or more second distance grids and one or more environmental variable grids corresponding to a target grid point are used as a model input. It should be emphasized that the type of distance, the resolution of the control point grid and the window used in the embodiment of the invention of the present invention are usually the same as the type of distance, the resolution of the control point grid and the window used in the construction of the first distance grid in the embodiment of the invention of the method for constructing the geographic element prediction model.

[0113] When constructing the model input and moving the window, its moving strategy can be to move sequentially along the grid direction, or to move according to the target grid point. Usually, when constructing the model input and predicting the geographic elements, the above-mentioned window will move according to the target grid point; that is, the window is moved so that the center of the window is as close as possible to the target grid point to be predicted and the window does not go out of the target range. The strategy of window movement in the embodiment of the present invention can be different from the strategy of window movement when constructing the sample in the above-mentioned invention embodiment.

[0114] Specifically, in the embodiment of the present invention, multiple target grid points can be set to form a target grid. Correspondingly, when constructing the model input, multiple groups of model inputs can be generated when the window is moved according to the distribution of the target grid points, so that the geographic element prediction model can output the geographic element distribution corresponding to the target grid.

[0115] It should be noted that when moving the window, when a target grid point is constructed with corresponding model inputs multiple times, for continuous geographic elements, the multiple prediction results of the target grid point can be processed by taking the average, maximum, minimum, mode, median, etc., or the multiple prediction results can be taken as a whole to evaluate the uncertainty of the target grid geographic element; for categorical geographic elements, the multiple prediction results of the target grid point can be processed by voting and other methods.

[0116] S302: Obtaining the geographic elements of the target grid points according to the model input and the trained geographic element prediction model.

[0117] In this step, the model input constructed in S301 is input into the trained geographic element prediction model to finally generate the geographic element of the target grid. When the target grid is pre-constructed, the geographic element prediction model can output the geographic element distribution corresponding to the target grid.

[0118] A geographic element prediction method provided by an embodiment of the present invention adopts a machine learning model to fully integrate known environmental variable information and the first law of geography, thereby having stronger modeling capabilities and higher prediction accuracy.

[0119] The specific contents of the geographical element prediction method provided by the present invention will be described in detail in the following invention embodiments.

[0120] Please refer to Figure 8 , Figure 8 The present invention provides a flowchart of a specific method for predicting geographic elements.

[0121] See also Figure 8 In an embodiment of the present invention, the geographical element prediction method includes:

[0122] S401: Acquire data required for geographic element prediction, and construct an environmental variable grid based on the distribution data of environmental variables.

[0123] For the specific content of constructing the environmental variable grid, please refer to the above-mentioned embodiment of the invention of the method for constructing a geographic element prediction model, which will not be repeated here.

[0124] Specifically, after this step, it is usually further included: pre-processing the data required for the geographic element prediction in the same manner as the above-mentioned embodiment of the invention of the method for building a geographic element prediction model, the difference being that in the embodiment of the invention, it is not necessary to process the geographic elements of the sampling points. The rest of the content can refer to the above-mentioned embodiment of the invention of the method for building a geographic element prediction model, including the process of processing the environmental variables after building the environmental variable grid, which will not be described in detail here.

[0125] Of course, in an embodiment of the present invention, the data corresponding to the above S201 and the environmental variable grid can also be directly obtained, that is, in an embodiment of the present invention, the above S401 can be omitted, and the environmental variable grid that has been constructed in the above-mentioned geographic element prediction model construction method invention embodiment can be directly called in subsequent steps.

[0126] S402: Acquire a control point grid composed of control points within the target range.

[0127] It should be noted that for different types of distances, the control point grids obtained in this step may have different resolutions. However, the resolution of each distance control point grid usually needs to be the same as the resolution of the corresponding distance control point grid in the above-mentioned embodiment of the invention of the method for constructing a geographic element prediction model.

[0128] Of course, in an embodiment of the present invention, the control point grid obtained in the above S202 can also be directly obtained, that is, in an embodiment of the present invention, the above S402 can be omitted, and the control point grid obtained in the above geographic element prediction model construction method invention embodiment can be directly called in subsequent steps.

[0129] S403: Determine the distance between the target grid point in the window and each control point in the window, and construct a second distance grid.

[0130] The above S402 to S403 are basically consistent with S202 to S203 in the above invention embodiment. In the embodiment of the present invention, the distance between the target grid point and each control point is specifically calculated to construct a second distance grid.

[0131] It should be noted that the above S403 can first move the preset window with a preset window movement strategy, and select the target grid points in the window that meet the preset standards. That is, the target grid points can be further screened in the step, and the screening criteria can be determined according to factors such as the window movement strategy. For example, when the window movement strategy is "moving the window so that the center of the window is as close as possible to the target grid point to be predicted and the window does not exceed the target range", only the second distance grid of the target grid point to be predicted is calculated. It should also be noted that in the embodiment of the present invention, the second distance grid can be further normalized or standardized, that is, each distance in the second distance grid is normalized or standardized. Of course, the normalized or standardized model and parameters should be consistent with the normalized or standardized model and parameters used in the above-mentioned embodiment of the invention of the method for constructing a geographic element prediction model. The specific content of normalization or standardization can refer to the prior art and will not be repeated here.

[0132] S404: Obtain the geographic elements of the target grid points according to the model input and the trained geographic element prediction model.

[0133] This step is basically the same as S302 in the above-mentioned embodiment of the invention. Please refer to the above-mentioned embodiment of the invention for details, and will not be described again here.

[0134] Specifically, in this step, when the output geographic element is a continuous geographic element, the value output by the geographic element prediction model is usually a normalized or standardized value; when the output geographic element is a categorized geographic element, the value output by the geographic element prediction model is usually a probability value for each category.

[0135] In other words, after the geographic elements are normalized or standardized in the embodiment of the present invention, the values ​​generated in this step are usually normalized or standardized values ​​of the geographic elements. Therefore, in order to facilitate user use, the generated normalized or standardized values ​​of the geographic elements can be further denormalized or destandardized after this step, so as to generate geographic element values ​​that are easy for users to understand; if the generated values ​​are probability values ​​of various categories of the geographic elements, the category to which the target grid point geographic element belongs can be further output according to the above probability values.

[0136] Furthermore, in the embodiment of the present invention, when there are multiple continuous geographic elements, after determining the weights of their respective errors, multiple continuous geographic elements can be modeled and predicted simultaneously. When there are multiple categorical geographic elements, the multiple categorical geographic elements can be combined into a new categorical geographic element, thereby achieving simultaneous modeling and prediction of multiple categorical geographic elements.

[0137] A geographic element prediction method provided by an embodiment of the present invention adopts a machine learning model to fully integrate known environmental variable information and the first law of geography, thereby having stronger modeling capabilities and higher prediction accuracy.

[0138] A geographic element prediction device provided by an embodiment of the present invention is introduced below. The geographic element prediction device described below and the geographic element prediction method described above can be referenced to each other.

[0139] Please refer to Fig. 9 , Fig. 9 This is a structural block diagram of a geographic element prediction device provided by an embodiment of the present invention. Fig. 9 , the geographic element prediction device may include:

[0140] The model input module 300 is used to construct a model input; the model input includes an environmental variable grid and a second distance grid in the same window; the second distance grid is a grid constructed according to the distance between the target grid point in the window and each control point in the window when the window is moved according to the preset window and window movement strategy;

[0141] The prediction module 400 is used to obtain the geographic element value or the category of the target grid point according to the model input and the trained geographic element prediction model.

[0142] Preferably, in the embodiment of the present invention, the model input module 300 includes:

[0143] The data acquisition unit is used to acquire the data required for geographic element prediction and construct the environmental variable grid according to the distribution data of environmental variables; the data required for geographic element prediction includes the distribution data of the environmental variables within the target range and the geographic elements of the sampling points.

[0144] A control point grid unit is used to obtain a control point grid composed of control points within a target range;

[0145] The second distance grid unit is used to determine the distance between the target grid point in the window and each control point in the window, and to construct a second distance grid.

[0146] Preferably, in the embodiment of the present invention, the model input module 300 further includes:

[0147] The target grid point selection unit is used to move the preset window according to the preset window movement strategy, and select the target grid points in the window that meet the preset standards.

[0148] Preferably, in the embodiment of the present invention, the distance includes any one or any combination of the following:

[0149] Distance based on space, distance based on time, distance based on geographical attributes.

[0150] Preferably, in the embodiment of the present invention, the model input module 300 further includes:

[0151] The preprocessing unit is used to preprocess the data required for the prediction of the geographic elements.

[0152] Preferably, in an embodiment of the present invention, the preprocessing includes any one or any combination of the following:

[0153] Abnormal data is removed and / or replaced, and the environment variables are encoded.

[0154] Preferably, in the embodiment of the present invention, the preprocessing unit is further used for:

[0155] The environmental variables are dimensionality reduced, and / or the environmental variables are normalized or standardized.

[0156] The geographic element prediction device of this embodiment is used to implement the aforementioned geographic element prediction method. Therefore, the specific implementation method of the geographic element prediction device can be seen in the embodiment part of the geographic element prediction method in the previous text. For example, the model input module 300 and the prediction module 400 are respectively used to implement steps S301 to S302 in the aforementioned geographic element prediction method. Therefore, its specific implementation method can refer to the description of the corresponding embodiments of each part, which will not be repeated here.

[0157] A geographic element prediction device provided by an embodiment of the present invention is introduced below. The geographic element prediction device described below and the geographic element prediction model construction method and geographic element prediction method described above can be referenced to each other.

[0158] Please refer to Fig.10 , Fig.10 A structural block diagram of a geographic element prediction device provided by an embodiment of the present invention.

[0159] Reference Fig.10 , the geographic element prediction device may include a processor 11 and a memory 12.

[0160] The memory 12 is used to store computer programs; the processor 11 is used to implement the specific content of the geographic element prediction model construction method described in the above-mentioned invention embodiment and / or the specific content of the geographic element prediction method when executing the computer program.

[0161] The processor 11 in the geographic element prediction device of this embodiment is used to install the geographic element prediction model construction device and / or the geographic element prediction device described in the above invention embodiments, and the processor 11 combined with the memory 12 can implement the geographic element prediction model construction method and / or the geographic element prediction method described in any of the above invention embodiments. Therefore, the specific implementation of the geographic element prediction device can be seen in the embodiment section of the foregoing text, and its specific implementation can refer to the description of the corresponding various parts of the embodiment, which will not be repeated here.

[0162] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, a method for constructing a geographic element prediction model and / or a geographic element prediction method described in any of the above-mentioned embodiments of the invention are implemented. The rest of the content can refer to the prior art and will not be described in detail here.

[0163] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0164] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0165] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0166] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0167] The above is a detailed introduction to a geographic element prediction model construction method, a geographic element prediction method, a geographic element prediction device and a computer-readable storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing a geographic element prediction model, characterized in that: include: Construct samples; The sample includes an environmental variable grid, a first distance grid and geographic elements of sampling points in the same window; the first distance grid is a grid constructed according to the distance between the sampling point in the window and each control point in the window when the window is moved according to the preset window and window movement strategy; The preset geographic element prediction model is trained through the samples to obtain a trained geographic element prediction model.

2. The method according to claim 1, characterized in that The sample build includes: Acquire data required for geographic element prediction, and construct the environmental variable grid according to the distribution data of environmental variables; the data required for geographic element prediction includes the distribution data of the environmental variables within the target range and the geographic elements of the sampling points; Get the control point grid composed of control points within the target range; The distances between the sampling points in the window and each control point in the window are determined, and a first distance grid is constructed.

3. The method according to claim 2, characterized in that Before determining the distances between the sampling points in the window and each control point in the window and constructing the first distance grid, the method further includes: The preset window is moved according to a preset window moving strategy, and sampling points in the window that meet preset standards are selected.

4. The method according to claim 3, characterized in that The distance includes any one or any combination of the following: Distance based on space, distance based on time, distance based on geographical attributes.

5. The method according to claim 2, characterized in that: After obtaining the data required for geographic element prediction, the following steps are also included: The data required for the prediction of the geographic elements is preprocessed.

6. The method according to claim 5, characterized in that The pretreatment includes any one or any combination of the following: Remove and / or replace abnormal data, encode the geographic elements or normalize or standardize the geographic elements, and encode the environmental variables.

7. The method according to claim 2, characterized in that: After building the environment variable grid, it also includes: The environmental variables are dimensionality reduced, and / or the environmental variables are normalized or standardized.

8. A method for predicting geographical elements, characterized in that: The trained geographic element prediction model obtained by applying the method according to any one of claims 1 to 7 comprises: Constructing a model input; the model input includes an environmental variable grid and a second distance grid in the same window; the second distance grid is a grid constructed according to the distance between the target grid point in the window and each control point in the window when the window is moved according to the preset window and window movement strategy; The geographic element of the target grid point is obtained according to the model input and the trained geographic element prediction model.

9. A geographical element prediction device, characterized in that: include: Memory for storing computer programs; A processor, used to execute the computer program to implement the steps of the method for constructing a geographic element prediction model as described in any one of claims 1 to 7, and / or the steps of the method for predicting geographic elements as described in claim 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the method for constructing a geographic element prediction model as described in any one of claims 1 to 7, and / or the steps of the method for predicting geographic elements as described in claim 8.

Images