Twin modeling method, device and equipment based on geographic process and medium

Abstract

The invention discloses a twin modeling method and device based on a geographic process, equipment and a medium. According to the method, firstly, a target object of the geographic process is constructed according to definition of twin objects in the geographic process, a graph topological structure is constructed according to the relation between the target object and geographic environment factors, and then the influence value of the geographic environment factors on the target object is extracted by utilizing a relation feature model; and finally, according to the influence value, carrying out quantitative reconstruction on the graph topological structure to obtain a geographical process evolution relation graph, and completing twin modeling of the geographical process. In the process, the target object is obtained by defining the twin object in the geographic process, then the relation feature model is used for mining the relation feature, namely the influence value, between the target object and the geographic environment factor, and finally the graph topological structure established by the target object and the geographic environment factor is subjected to quantitative reconstruction according to the influence value. Therefore, the geographical process influenced by various geographical environment elements can be coupled to realize twin modeling of the geographical process.

Description

Technical Field

[0001] The present application relates to the technical field of digital twin modeling, and in particular to a twin modeling method, device, equipment and medium based on geographic processes. Background Art

[0002] With the development of science and technology, digital twins have well realized the twin modeling (spatial position and geometric shape) of static objects in the field of earth sciences based on the digital twin modeling theory in the industrial field, such as BIM and City Information Model (CIM).

[0003] However, currently available patents on geographic twin modeling mostly focus on the expression and visualization of static geographic twin scenes or the construction of platform systems, ignoring the exploration of relationships between geographic objects. It is also difficult to depict the dynamic changes and interactions of geographic objects, that is, geographic processes. In this regard, this application provides a twin modeling method based on geographic processes. Summary of the invention

[0004] The present application provides a twin modeling method, device, equipment and medium based on geographic processes, which realizes twin modeling based on geographic processes by setting static and dynamic characteristics of geographic elements and exploring the evolutionary relationship between geographic elements and geographic environment elements.

[0005] In a first aspect, the present application provides a twin modeling method based on geographic processes, characterized in that the method comprises:

[0006] According to the definition of twin objects in geographic process, construct the target object of geographic process;

[0007] Constructing a graph topology structure according to the relationship between the target object and the geographical environment factors, wherein the geographical environment factors include geographical environment elements driving the evolution of geographical processes, the target object and the geographical environment factors in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factors in the graph topology structure is an edge;

[0008] The influence value of the geographical environment factor on the target object is extracted by using the relationship feature model, wherein the relationship feature model is a trained neural network model, and the training samples of the relationship feature model are the relationship feature values ​​between the target object and the geographical environment factor;

[0009] The graph topology is quantitatively reconstructed using the influence value to obtain a geographic process evolution relationship graph, which is a graph-based geographic process evolution driving relationship expression model.

[0010] Optionally, according to the definition of the twin object in the geographic process, the target object of the geographic process is constructed, including:

[0011] According to the definition of twin objects in the geographic process, determine the spatial location information, geometric morphology information and attribute information required by the target object;

[0012] Construct the target object based on spatial location information, geometric shape information and attribute information.

[0013] Optionally, the method further comprises:

[0014] Constructing a dynamic change evolution relationship of the target object at multiple times based on the spatial position information, geometric shape information and attribute information at multiple times;

[0015] In the spatial visualization window, the spatial position, geometric shape and attribute information of the target object at multiple times are displayed in the form of spatiotemporal snapshots.

[0016] Optionally, the influence value is used to quantitatively reconstruct the graph topology to obtain a relationship diagram of the evolution of geographical processes, including;

[0017] The influence values ​​are normalized, and the size piecewise function of the geographical environment factor nodes in the graph topology is established according to the normalized influence values;

[0018] Quantitatively display the geographical environment factor nodes according to the size piecewise function;

[0019] The attribute information of the target object, the attribute information of the geographical environment factor and the influence value are used to calculate and obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor;

[0020] The evolution relationship strength is used as the attribute of the edge, and the evolution relationship direction is represented by the edge style, so as to obtain the geographic process evolution relationship diagram.

[0021] Optionally, the attribute information of the target object, the attribute information of the geographical environment factor, and the influence value are used to perform calculations to obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor, including:

[0022] Substitute the attribute value of the target object and the attribute value of the geographical environment factor into the evolution relationship strength formula to obtain the evolution relationship strength between the target object and the geographical environment factor. The evolution relationship strength formula includes:

[0023]

[0024] in, is the evolution relationship strength between the target object and the geographical environment factor from time t to t+1; fabs() is an absolute value function to ensure that the relationship strength is positive; G t+1 (Value) and G t (Value) are the attribute information of geographical environment factors at time t+1 and time t respectively; t+1 (Value) and O t (Value) is the attribute information of the target object at time t+1 and time t, respectively, and t is the discrete moment of the entire research cycle or geographical process cycle;

[0025] Substitute the impact value into the evolution relationship direction formula to obtain the evolution relationship direction between the target object and the geographical environment factor. The evolution relationship direction formula includes:

[0026]

[0027] in, is the direction of the evolution relationship between the target object and the geographical environmental factors from time t to t+1; v is the mean value of the impact of the geographical environmental factors on the target object; T in The time series length parameter for time-dependent feature mining input; F v (t i ) is at t i The impact value of geographical environment factors on the target object at each moment.

[0028] Optionally, the process of building the relationship feature model includes:

[0029] Set the characteristic values ​​of the target object and the characteristic values ​​of the geographical environment factors;

[0030] Based on the spatial size of the characteristic value of the target object, the geographical environment factor is spatially resampled or interpolated so that the spatial size of the characteristic value of the geographical environment factor is consistent with the spatial size of the characteristic value of the target object;

[0031] Normalize the characteristic values ​​of the geographical environment factors and the characteristic values ​​of the target objects, and construct the characteristic values ​​of the relationship between the target objects and the geographical environment factors;

[0032] Input the relationship feature values ​​into the initial neural network model to conduct mining and training of spatial and temporal features between the target object and the geographical environment factors;

[0033] When the mining training of the initial neural network model is completed, the relational feature model is obtained.

[0034] Optionally, the method further comprises:

[0035] In the evolution relationship visualization window, the evolution relationship diagram of geographical processes at multiple times is displayed in the form of a graph structure.

[0036] In a second aspect, the present application provides a twin modeling device based on geographic processes, the device comprising:

[0037] A construction unit, used to construct a target object of a geographic process according to the definition of a twin object in the geographic process;

[0038] The construction unit is further used to construct a graph topology structure according to the relationship between the target object and the geographical environment factors, wherein the geographical environment factors include geographical environment elements that drive the evolution of geographical processes, the target object and the geographical environment factors in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factors in the graph topology structure is an edge;

[0039] An extraction unit is used to extract the impact value of the geographical environment factor on the target object by using a relationship feature model, wherein the relationship feature model is a trained neural network model, and the training sample of the relationship feature model is the relationship feature value between the target object and the geographical environment factor;

[0040] The obtaining unit is used to quantitatively reconstruct the graph topology structure by using the influence value to obtain a geographic process evolution relationship graph, wherein the geographic process evolution relationship graph is a graph-based geographic process evolution driving relationship expression model.

[0041] Optionally, the construction unit is used to construct the target object of the geographic process according to the definition of the twin object in the geographic process, specifically for:

[0042] According to the definition of twin objects in the geographic process, determine the spatial location information, geometric morphology information and attribute information required by the target object;

[0043] Construct the target object based on spatial location information, geometric shape information and attribute information.

[0044] Optionally, the device further comprises a display unit for:

[0045] Constructing a dynamic change evolution relationship of the target object at multiple times based on the spatial position information, geometric shape information and attribute information at multiple times;

[0046] In the spatial visualization window, the spatial position, geometric shape and attribute information of the target object at multiple times are displayed in the form of spatiotemporal snapshots.

[0047] Optionally, the obtaining unit comprises:

[0048] A processing subunit is used to normalize the impact value and establish a size piecewise function of the geographical environment factor node in the graph topology structure according to the normalized impact value;

[0049] The quantification subunit is used to quantify and display the geographical environment factor nodes according to the size piecewise function;

[0050] A calculation subunit, used to calculate using the attribute information of the target object, the attribute information of the geographical environment factor and the influence value, to obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor;

[0051] Subunits are obtained for taking the evolution relationship strength as the attribute of the edge and using the edge style to represent the evolution relationship direction, thereby obtaining a geographic process evolution relationship diagram.

[0052] Optionally, the computing subunit is specifically used for:

[0053] Substitute the attribute value of the target object and the attribute value of the geographical environment factor into the evolution relationship strength formula to obtain the evolution relationship strength between the target object and the geographical environment factor. The evolution relationship strength formula includes:

[0054]

[0055] in, is the evolution relationship strength between the target object and the geographical environment factor from time t to t+1; fabs() is an absolute value function to ensure that the relationship strength is positive; G t+1 (Value) and G t (Value) are the attribute information of geographical environment factors at time t+1 and time t respectively; t+1 (Value) and O t (Value) is the attribute information of the target object at time t+1 and time t, respectively, and t is the discrete moment of the entire research cycle or geographical process cycle;

[0056] Substitute the impact value into the evolution relationship direction formula to obtain the evolution relationship direction between the target object and the geographical environment factor. The evolution relationship direction formula includes:

[0057]

[0058] in, is the direction of the evolution relationship between the target object and the geographical environmental factors from time t to t+1; v is the mean value of the impact of the geographical environmental factors on the target object; T in The time series length parameter for time-dependent feature mining input; F v (t i ) is at t iThe impact value of geographical environment factors on the target object at each moment.

[0059] Optionally, the building block is also used to:

[0060] Set the characteristic values ​​of the target object and the characteristic values ​​of the geographical environment factors;

[0061] Based on the spatial size of the characteristic value of the target object, the geographical environment factor is spatially resampled or interpolated so that the spatial size of the characteristic value of the geographical environment factor is consistent with the spatial size of the characteristic value of the target object;

[0062] Normalize the characteristic values ​​of the geographical environment factors and the characteristic values ​​of the target objects, and construct the characteristic values ​​of the relationship between the target objects and the geographical environment factors;

[0063] Input the relationship feature values ​​into the initial neural network model to conduct mining and training of spatial and temporal features between the target object and the geographical environment factors;

[0064] When the mining training of the initial neural network model is completed, the relational feature model is obtained.

[0065] Optionally, the device further comprises a display unit for:

[0066] In the evolution relationship visualization window, the evolution relationship diagram of geographical processes at multiple times is displayed in the form of a graph structure.

[0067] In a third aspect, the present application provides an electronic device, the electronic device comprising a memory and a processor:

[0068] The memory is used to store computer programs;

[0069] The processor is used to execute the method provided by the first aspect according to the computer program.

[0070] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store a computer program, and the computer program is used to execute the method provided in the first aspect above.

[0071] It can be seen that this application has the following beneficial effects:

[0072] The present application provides a twin modeling method based on geographic process. First, according to the definition of twin objects in the geographic process, after constructing the target object of the geographic process, a graph topology is constructed according to the relationship between the target object and the geographic environment factor. The geographic environment factor includes the geographic environment factor that drives the evolution of the geographic process. The target object and the geographic environment factor in the graph topology are nodes, and the evolution relationship between the target object and the geographic environment factor in the graph topology is an edge. Then, the relationship feature model is used to extract the impact value of the geographic environment factor on the target object. The relationship feature model is a trained neural network model, and the training sample of the relationship feature model is the relationship feature value between the target object and the geographic environment factor. Finally, the graph topology is quantitatively reconstructed according to the impact value to obtain the geographic process evolution relationship graph, wherein the geographic process evolution relationship graph is a graph-based geographic process evolution driving relationship expression model. In this process, the target object of the geographic process is obtained by defining the twin object in the geographic process, and the relationship feature model is used to mine the relationship characteristics between the target object and the geographic environment factor, that is, the impact value. Finally, the graph topology established by the target object and the geographic environment factor is quantitatively reconstructed according to the impact value, so that the geographic process affected by multiple geographic environment elements can be coupled to realize the twin modeling of the geographic process. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0074] Figure 1 A schematic diagram of the implementation route of the technical solution of the twin modeling method based on geographic processes in the embodiment of the present application;

[0075] Figure 2 A schematic diagram of a flow chart of an embodiment of a twin modeling method based on geographic processes in an embodiment of the present application;

[0076] Figure 3 A schematic diagram of the process of mining a geographic evolution relationship model based on deep learning in an embodiment of the present application;

[0077] Figure 4 This is a schematic diagram of the evolution relationship of geographical processes in the embodiments of this application;

[0078] Figure 5 A schematic flow chart of another embodiment of a twin modeling method based on geographic processes in an embodiment of the present application;

[0079] Figure 6 A schematic diagram of visualization of a geographic process twin object in an embodiment of the present application;

[0080] Figure 7 Schematic diagram of the structure of a twin modeling device based on geographic process in an embodiment of the present application;

[0081] Figure 8 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0082] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0083] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0084] A geographic process is a characteristic of the spatial position, geometric form, evolutionary process, attribute characteristics, and interaction relationship of geographical phenomena / geographic objects (hereinafter referred to as geographic objects in this application) that changes over time and space. Therefore, the twin modeling of geographic processes required in this application is not only to realize the spatial simulation and modeling of the spatial position, geometric form, and attribute characteristics of geographic objects and the simulation and modeling of the spatial position, geometric form, and attribute characteristics of geographic objects that change over time, but also to realize the simulation and modeling of the evolutionary relationship that causes the dynamic change of geographic objects. The twin modeling elements of geographic processes include: (1) the spatial position, geometric form, and attribute characteristics of geographic objects; (2) the dynamic change of geographic objects; (3) the evolutionary relationship that causes the dynamic change of geographic objects.

[0085] At present, geographic process twin modeling is one of the core technologies for future prediction of digital twin earth and an important part of scenario simulation analysis. The reasons why digital twin modeling technology methods are difficult to achieve geographic process twin modeling are summarized as follows: 1) lack of definition and connotation of geographic process twin; 2) digital twin modeling methods do not consider the dynamic changes of objects and cannot carry out object modeling with change information; 3) the dynamic changes of geographic objects are affected by the comprehensive influence of multiple geographic environmental factors, and there is currently a lack of relevant research on how to carry out twin modeling of the interaction relationship between geographic objects.

[0086] In order to solve the above-mentioned technical problems, the embodiments of the present application conceive that the definition and connotation of geographic process twins can be first given, and then the driving factors of geographic dynamic changes and their evolutionary relationships can be explored from a system perspective. Finally, geographic dynamic changes (time-series geographic objects) and geographic dynamic evolution relationships can be coupled to realize geographic process twin modeling.

[0087] According to the above steps, the present invention proposes that Figure 1 The technical route of twin modeling of geographic processes shown in the figure is, first, to define geographic processes and their connotations, which is the basis of twin modeling of geographic processes. According to the definition of geographic process: the characteristics of the spatial position, geometric form, evolution process, attribute characteristics and interaction relationship of geographic objects over time and space, that is, the basic elements of geographic process include: (1) the spatial position, geometric form and attributes of geographic objects (static characteristics); (2) the changes in the spatial position, geometric form and attributes of geographic objects over time (dynamic characteristics); (3) the relationship between the dynamic changes of geographic objects and other geographic environmental elements (evolution characteristics). Therefore, the definition of geographic process twin is: to achieve accurate digital expression and modeling of geographic objects, changes in geographic objects and the relationship between driving the dynamic changes and evolution of geographic objects. Secondly, from the perspective of geographic system, develop a geographic "driver-response" relationship mining model based on deep learning, and establish a graphical expression method for the interaction relationship between geographic objects and geographic environmental elements. Thirdly, combine the discrete moment state geographic objects and geographic environmental elements, calculate the strength of the dynamic change evolution relationship of geographic objects, and reconstruct the evolution relationship between geographic objects and geographic environmental elements within the life cycle or research period. Finally, couple the moment state geographic objects and the evolution relationship between them to achieve the modeling and simulation of geographic process twin.

[0088] In a specific implementation, the method may include, for example: first, according to the definition of the twin object in the geographic process, after constructing the target object of the geographic process, constructing a graph topology structure according to the relationship between the target object and the geographic environment factors, wherein the geographic environment factors include the geographic environment elements that drive the evolution of the geographic process, the target object and the geographic environment factors in the graph topology structure are nodes, and the evolution relationship between the target object and the geographic environment factors in the graph topology structure is an edge, and then using the relationship feature model to extract the impact value of the geographic environment factor on the target object, wherein the relationship feature model is a trained neural network model, and the training samples of the relationship feature model are the relationship feature values ​​between the target object and the geographic environment factors, and finally using the impact value to quantitatively reconstruct the graph topology structure to obtain a geographic process evolution relationship graph, wherein the geographic process evolution relationship graph is a graph-based geographic process evolution driving relationship expression model.

[0089] It can be seen that the method provided by the implementation of this application obtains the target object of the geographical process by defining the twin object in the geographical process, and then uses the relationship feature model to mine the relationship characteristics between the target object and the geographical environmental factors, that is, the influence value. Finally, the graph topology structure established by the target object and the geographical environmental factors is quantitatively reconstructed according to the influence value, thereby coupling the geographical processes affected by multiple geographical environmental elements to realize twin modeling of the geographical process.

[0090] To facilitate understanding of the specific implementation of the twin modeling method based on geographic processes provided in the embodiments of the present application, an explanation will be given below with reference to the accompanying drawings.

[0091] It should be noted that the subject implementing the twin modeling method based on geographic process can be the twin modeling device based on geographic process provided in the embodiment of the present application, and the twin modeling device based on geographic process can be carried in an electronic device or a functional module of an electronic device. The electronic device in the embodiment of the present application can be any device capable of implementing the twin modeling method based on geographic process in the embodiment of the present application, for example, it can be an Internet of Things (IoT) device.

[0092] Figure 2 A schematic diagram of a process flow of a twin modeling method based on geographic process provided in an embodiment of the present application. The method can be applied to a twin modeling device based on geographic process, which can be, for example, Figure 7 The twin modeling device 700 based on geographic process shown in FIG. 1 may also be integrated into Figure 8 Functional modules in the electronic device 800 are shown.

[0093] like Figure 2 As shown, the method includes the following steps S201 to S204:

[0094] S201: Construct the target object of the geographic process according to the definition of the twin object in the geographic process.

[0095] In order to realize the twin modeling of geographic process, the embodiment of the present application first constructs the target object of the geographic process according to the definition of the twin object in the geographic process, and then constructs the graph topology structure according to the relationship between the target object and the geographic environment factor. The geographic environment factor includes the geographic environment elements that drive the evolution of the geographic process. The target object and the geographic environment factor in the graph topology structure are nodes, and the evolution relationship between the target object and the geographic environment factor in the graph topology structure is an edge; then the relationship feature model is used to extract the impact value of the geographic environment factor on the target object. The relationship feature model is a trained neural network model, and the training sample of the relationship feature model is the relationship feature value between the target object and the geographic environment factor; finally, the graph topology structure is quantitatively reconstructed using the impact value to obtain the geographic process evolution relationship graph, which is a graph-based geographic process evolution driving relationship expression model. Therefore, the embodiment of the present application constructs the target object of the geographic process through S201, which makes preliminary preparations for the subsequent construction of the graph topology structure.

[0096] As an example, S201 may include: determining the spatial position information, geometric morphology information, and attribute information required for the target object according to the definition of the twin object in the geographic process; and constructing the target object according to the spatial position information, geometric morphology information, and attribute information.

[0097] First, based on the definition of geographic process, the embodiment of the present application gives a definition of geographic process twin: the process of using digital technology to model and simulate the dynamic changes of geographic objects and their evolutionary relationships. Therefore, the elements of geographic process twin modeling include: 1) twin modeling of time-series geographic objects; 2) twin modeling that drives the dynamic changes and evolutionary relationships of geographic objects.

[0098] At the same time, according to the definition of geographic process twin, the definition of geographic process twin object is given as: the coupling body of the evolution relationship between the dynamic change of geographic objects and the dynamic change of geographic objects driven by them within a certain spatial range. Different from the components of digital twin objects: spatial position, geometric form and thematic attribute twin, the components of geographic process twin objects include: 1) spatial position and geometric form twin; 2) attribute twin, among which attribute twin is a twin involving quantitative data; 3) geographic object dynamic change twin; 4) evolution relationship twin between geographic objects.

[0099] The expression of the geographic process twin object is shown in formula (1).

[0100] E t t+1 <O t (S,A),O t+1 (S,A),T1≤t<T N Formula (1)

[0101] Among them, the above Ot (S, A) represents the twin object of the geographic process at time t, S and A represent the spatial information (spatial position and geometric form) and attribute information of the twin object at time t respectively; similarly, O t+1 (S, A) represents the twin object of the geographic process at time t+1, S and A represent the spatial information (spatial position and geometric form) and attribute information of the twin object at time t+1 respectively; T represents the modeling time range or the time range of the input data, T1 is the start time, and TN is the end time; Represents the evolution relationship between the defined geographic objects from time t to time t+1.

[0102] That is, the above is the expression of the geographic process twin object defined in the embodiment of the present application. Through the expression of the geographic process twin object, it can be determined that first the spatial position, geometric shape and attribute information of the twin object at different times are needed, and then the twin object of the geographic process at the corresponding time is obtained according to the spatial position, geometric shape and attribute information of the twin object at different times, that is, the target objects at different times are constructed.

[0103] In this process, the embodiment of the present application defines the concept, connotation and expression of the twin object of the geographic process through the definition of the geographic process, which promotes the progress of realizing the twin of the geographic dynamic process and realizes it from scratch.

[0104] It should be noted that after constructing the target objects at different times, the embodiment of the present application can also visualize the target objects at different times and intuitively display the twin objects of the geodynamic process. The specific process may include: constructing the dynamic change evolution relationship of the target object at multiple times according to the spatial position information, geometric morphology information and attribute information at multiple times, for example, obtaining the geographic objects at times t and t+1 ( t (S,A),O t+1 (S,A)) and the dynamic evolution relationship between them And in the spatial visualization window, the spatial position, geometric shape and attribute information of the target object at multiple times are displayed in the form of spatiotemporal snapshots.

[0105] S202: constructing a graph topology structure according to the relationship between the target object and the geographical environment factors, wherein the geographical environment factors include geographical environment elements that drive the evolution of geographical processes, the target object and the geographical environment factors in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factors in the graph topology structure is an edge.

[0106] In the embodiment of the present application, a graph topology is established to establish a model for expressing the driving relationship of the evolution of the geographical process. Since the twin modeling of the geographical process requires the coupling of multiple geographical environmental factors, the geographical environmental factors need to be used as nodes in the graph topology, where the geographical environmental factors are the geographical environmental factors that drive the evolution of the geographical process. Therefore, it is necessary to first set the nodes and edges in the graph topology. In the embodiment of the present application, the target object and the geographical environmental factors are abstracted as the nodes of the graph, and the evolution relationship between the target object and the geographical environmental factors is abstracted as the edges of the graph.

[0107] S203: Using a relationship feature model, extracting the impact value of the geographical environment factor on the target object, the relationship feature model is a trained neural network model, and the training samples of the relationship feature model are the relationship feature values ​​between the target object and the geographical environment factor.

[0108] In the embodiment of the present application, a driving relationship mining model between target objects and geographic environmental factors based on the Pytorch framework is set up, which includes three layers: data input layer, spatiotemporal feature learning and evolution relationship output layer. The geographic data input layer is mainly responsible for the data input and feature tensor construction of target objects and geographic environmental factors; the CNN and LSTM models constitute the learning layer of geographic spatiotemporal feature mining, in which CNN is responsible for processing the spatial characteristics of target objects and environmental factors, and LSTM is responsible for processing the time-dependent characteristics of target objects and environmental factors; the evolution relationship output layer is responsible for constructing a geographic process evolution model and establishing the driving relationship between geographic environmental factors and target objects. The specific technical process is as follows: Figure 3 shown.

[0109] It should be noted that the CNN and LSTM models in the above technical process are intended to realize the mining of the driving relationship between geographical environmental elements and target objects (geographic processes). Other deep learning models, such as GRU, UNet, AttentionLSTM, ResNet models, etc., can also be used and are not limited here.

[0110] In the embodiment of the present application, a convolutional neural network (CNN) and a long short-term memory network (LSTM) are coupled to develop a CNN-LSTM model to mine the evolution relationship of geographical processes (relationship feature model): that is, the "drive-response" relationship between geographical environmental factors and target objects.

[0111] First, the input data to be input into the relational feature model is set. That is, the geographic data input layer is mainly responsible for the input of target object and geographic environmental factor data and the construction of feature tensors, including: (1) setting the feature value of the target object, that is, setting the spatial size of the target object to: Row and Col; setting the feature value of the geographic environmental factor that drives the dynamic change of the target object, that is, setting the number of geographic environmental factors to M and the time series length to N; (2) based on the length Row and width Col of the target object, spatially resample or interpolate the geographic environmental factor data, reconstruct the length to Row and the width to Col, so that the length and width are consistent with the target object; (3) normalize the target object and the geographic environmental factor, and construct a three-dimensional feature tensor with a time series length of N as 3DTensor: Row*Col*(M+1), that is, obtain the relational feature value between the target object and the geographic environmental factor.

[0112] The relationship feature values ​​are then input into the initial neural network model to conduct mining and training of the spatial and temporal features between the target object and the geographic environmental factors.

[0113] Finally, when the mining training of the initial neural network model is completed, the relationship feature model is obtained.

[0114] The embodiment of the present application uses the CNN-LSTM model as the relationship feature model, that is, the CNN model is used to mine and train the spatial features between the target object and the geographical environment factors, and the LSTM model is used to mine and train the temporal features between the target object and the geographical environment factors. The training process of the CNN-LSTM model is described in detail below:

[0115] First, the relationship feature values ​​are input into the CNN model to mine and train the spatial features between the target object and the geographical environment factors. The specific training process may include:

[0116] (1) For each time t, set the number of convolution depth layers (K) to perform spatial feature mining;

[0117] (2) The convolution kernel size is set to 3*3, the output channel is set to 32 (the number of spatial features), and the spatial convolution operation is performed using formula (2). The set values ​​are not specifically limited here in order to implement the convolution operation. Formula (2) is as follows:

[0118] X conv1 =ReLU(W conv1 *X input +b convl ) Formula (2)

[0119] Among them, X conv1 is the output feature after convolution, W conv1 is the convolution kernel, Xinput is a three-dimensional feature tensor (3DTensor) or the output feature map of the previous layer, * is the convolution operation, b convl is the bias term of the convolutional layer, and ReLU is the activation function.

[0120] (3) Perform max pooling using formula (3) to achieve spatial dimensionality reduction, so that the size of the spatial features after pooling is: X convl size / pooling window size, where formula (3) is as follows:

[0121] X pool1 = MAXPool2D(X convl , 2×2) Formula (3)

[0122] where X pool1 is the max pooling result, X conv1 is the previous convolutional feature, MAXPool2D() is the max pooling operation function, and 2×2 is the size of the pooling window (the size of the spatial features becomes: Xconv1 size / 2).

[0123] (4) Repeat steps (2) and (3), perform K layers of depth spatial convolution, and fully connect the spatial features of all depth layers to achieve spatial dimensionality increase, and output spatial features with a size of Row*Col;

[0124] (5) Repeat steps (1) to (5) until all the spatial features of the interaction relationships between the target object and the geographical environment factors at all times are mined.

[0125] Then input the above output spatial features into the LSTM model to mine and train the time features between the target object and the geographical environment factors. The specific training process can include:

[0126] (1) Flatten the spatial features at each time step to form a spatial feature tensor: Row*Col;

[0127] (2) Determine the time series length Tin (Tin < N) of the input features, the number N of input samples, and the number (M + 1) of input features, and construct the input feature tensor structure InputShape: (Tin, N, M + 1);

[0128] (3) Determine the time series length Tout (Tout <= Tin) of the output features, the number (N) of input samples, and the dimension (hidden_size * num_directions) of the hidden layer, and construct the output feature tensor structure OutputShape: (Tout, N, hidden_size * num_directions);

[0129] (4) Based on the LSTM model in the PyTorch model library, formulas (4) to (7) are used to mine the temporal evolution characteristics of the target object and the geographical environment factors. Formulas (4) to (7) are as follows:

[0130] Initialize the input feature function:

[0131]

[0132] Constructing the evolving feature model:

[0133] LSYM_Evolution=PyTorch.Module.LSTM(InputShape,hidden_size,num_layers) formula (5)

[0134] Execute the forward propagation function:

[0135] OutputShape,(h_n,c_n)=PyTorch.Module.LSTM(InputShape) Formula (6)

[0136] Set up the full link layer:

[0137] OutputShape=PyTorch.Module.LSTM(hidden_size,outFeatureNum=1) formula (7)

[0138] In the above formulas (4) to (7), Tin and Tout represent the use of the time series features of the length of Tin to infer the time series features of the length of Tout. For example, the target object and the geographical environment factors of the first three moments are used to infer the target object of the next moment, then Tin and Tout are 3 and 1 respectively; num_layers is the number of depth layers of the LSTM model, and the embodiment of the present application adopts the default value 1; num_directions is the dimension of the propagation direction. When the LSTM model adopts bidirectional propagation, num_directions=2; when the LSTM model adopts unidirectional propagation, num_directions=1. The embodiment of the present application adopts a unidirectional propagation LSTM model, that is, num_directions=1; hidden_size is the size of the hidden layer, the number of features output at each moment, and the default setting of the embodiment of the present application is 4; outFeatureNum is the number of features finally output by the LSTM model. In order to construct the interaction relationship between the target object and the geographical environment factors, the final output features of the embodiment of the present application include the target object and the geographical environment factors. Therefore, outFeatureNum is M+1.

[0139] In this process, neural network models are used to extract high-level features from raw data to mine evolving relationships without relying on manually designed algorithms. This reduces reliance on domain expert knowledge and can discover complex patterns that are difficult to capture with traditional methods. In addition, thanks to the powerful representation capabilities of deep learning, it can capture complex and nonlinear relationships in data. At the same time, compared with traditional relationship mining methods, neural network models can better fit complex data distributions due to their deep structure and large number of parameters.

[0140] After the relationship feature model is constructed, S203 in the embodiment of the present application describes that the SHAP value of the geographical environment factor for the target object can be extracted according to the relationship feature model so as to perform subsequent reconstruction according to the SHAP value.

[0141] It should be noted that the embodiment of the present application uses SHAP analysis to extract and express the evolution relationship between the target object and the geographical environment factors. Other similar model interpretation analyses, such as Local Interpretable Model-agnostic Explanations (LIME), Deep Learning Important Features (DeepLIFT), Permutation Feature Importance (PFI), Partial Dependence Plots (PDP) and Gradient-based Methods, etc. can be used as alternatives to extract and express the evolution relationship between the target object and the geographical environment factors, and are not limited here.

[0142] S204: Quantitatively reconstructing the graph topology structure using the influence value to obtain a geographic process evolution relationship graph, wherein the geographic process evolution relationship graph is a graph-based geographic process evolution driving relationship expression model.

[0143] As an example, S204 may include: S2041, normalizing the influence value, and establishing a size piecewise function of the geographical environment factor nodes in the graph topology structure according to the normalized influence value; S2042, quantitatively displaying the geographical environment factor nodes according to the size piecewise function; S2043, calculating using the attribute information of the target object, the attribute information of the geographical environment factor and the influence value to obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor; S2044, using the evolution relationship strength as the attribute of the edge, and using the edge style to express the evolution relationship direction, to obtain a geographical process evolution relationship diagram.

[0144] The size piecewise function in S2042 above can be shown as the following formula (8):

[0145]

[0146] Among them, Size(NodeO) and Size(NodeF) are the size of the target object (O) node and the size of the geographical environment factor (F) node, respectively. v is the normalized SHAP value of the geographical environment factor (F).

[0147] The larger the node, the greater the contribution of the geographical environment factor to the relationship feature model, and the smaller the node, the smaller the contribution of the geographical environment factor to the evolution model. The quantitative display of the geographical environment factor nodes in the graph topology structure can be as follows: Figure 4 As shown, the contribution of the geographical environment factor node to the relationship feature model can be intuitively displayed.

[0148] The above S2043 describes that after obtaining the impact value of the geographical environment factor on the target object, the geographical process evolution relationship reconstruction can be performed. The geographical process evolution relationship reconstruction is performed after the geographical environment factor node is quantified ( Figure 2 ) is used to quantify the evolution relationship strength and the evolution relationship direction. The specific process may include: firstly, substituting the attribute value of the target object and the attribute value of the geographical environment factor into the evolution relationship strength formula to obtain the evolution relationship strength between the target object and the geographical environment factor, wherein the evolution relationship strength formula is shown in the following formula (9):

[0149]

[0150] in, is the evolution relationship strength between the target object and the geographical environment factor from time t to t+1; fabs() is an absolute value function to ensure that the relationship strength is positive; G t+1 (Value) and G t (Value) are the attribute information of geographical environment factors at time t+1 and time t respectively; t+1 (Value) and O t (Value) is the attribute information of the target object at time t+1 and time t respectively, and t is the discrete moment of the entire research cycle or geographical process cycle.

[0151] Then substitute the impact value into the evolution relationship direction formula to obtain the evolution relationship direction between the target object and the geographical environment factor, where the evolution relationship direction formula is shown in the following formula (10) and formula (11):

[0152]

[0153] in, is the direction of the evolution relationship between the target object and the geographical environmental factors from time t to t+1; v is the mean value of the impact of the geographical environmental factors on the target object; T in The time series length parameter for time-dependent feature mining input; F v (t i ) is at t i The impact value of geographical environment factors on the target object at each moment.

[0154] The above S2044 is specifically described as including updating the graph topology from time t to time t+1 based on the evolution relationship strength and the evolution relationship direction: taking the evolution relationship strength as the attribute of the edge between the geographical environment factor and the target object; using solid line edges to express positive correlation and dotted line edges to express negative correlation, repeating the above S2043 and S2044 until the graph topology structure within the entire research cycle or the geographical process cycle is constructed, and the geographical process evolution relationship graph is obtained. The geographical process evolution relationship graph is a graph-based geographical process evolution-driven relationship expression model, that is, the twin modeling based on the geographical process is completed.

[0155] After obtaining the geographic process evolution relationship diagram, the geographic process evolution relationship diagram can be visualized. The specific process may include: in the evolution relationship visualization window, displaying the geographic process evolution relationship diagram at multiple moments in the form of a graph structure, for example, traversing from t to t+1, until the entire research cycle or life cycle.

[0156] In this process, dynamic geographic twin modeling was realized by considering the changes and interactions between temporal target objects and temporal geographic environmental factors, and dynamic linkage visualization was performed through the process twin object visualization method that couples geographic temporal elements / geographic objects and their evolutionary relationships.

[0157] It can be seen that the embodiment of the present application obtains the target object of the geographical process by defining the twin object in the geographical process, and then uses the relationship feature model to mine the relationship characteristics between the target object and the geographical environmental factors, that is, the influence value. Finally, the graph topology structure established by the target object and the geographical environmental factors is quantitatively reconstructed according to the influence value, so that the geographical processes affected by multiple geographical environmental elements can be coupled to realize the twin modeling of the geographical process.

[0158] In order to make the method provided in the embodiment of the present application clearer and easier to understand, Figure 5 A specific example of the application of this method is described below.

[0159] like Figure 5 As shown, the implementation may include the following steps:

[0160] S501: According to the definition of the twin object in the geographic process, determine the spatial location information, geometric information and attribute information required by the target object to construct the target object.

[0161] In the embodiments of this application, based on the definition of geographic process, the definition and connotation of the twin of geodynamic process and the twin object of geographic process are innovatively given, which promotes the progress of realizing the twin of geodynamic process and realizes it from scratch. The definition and connotation of the specific twin of geodynamic process and the twin object of geographic process can be referred to the above embodiments, which will not be repeated here.

[0162] Through the definition of twin objects in the geographic process, it can be seen that to construct a target object, it is first necessary to determine the spatial position information, geometric shape information and attribute information required for the target object. After obtaining the corresponding spatial position information, geometric shape information and attribute information, the construction of the target object can be completed.

[0163] S502: Constructing a graph topology structure according to the relationship between the target object and the geographical environment factors, wherein the geographical environment factors include geographical environment elements that drive the evolution of geographical processes, the target object and the geographical environment factors in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factors in the graph topology structure is an edge.

[0164] During the evolution of geographical processes, the target object will be affected by a variety of geographical environmental factors. Therefore, when expressing the driving relationship expression model of the geographical process evolution, the target object needs to couple the influence of a variety of geographical environmental factors. That is, a graph topology structure can be constructed based on the relationship between the target object and the geographical environmental factors, in which the target object and the geographical environmental factors are defined as nodes, and the evolution relationship between the two is an edge.

[0165] S503: constructing a relationship feature model using the relationship feature values ​​between the target object and the geographical environment factors, wherein the relationship feature model is a trained neural network model.

[0166] For the graph topology structure established above, the edges in the graph topology structure are the evolutionary relationships between the target objects and the geographical environment factors. In order to assign attributes to the edges, the embodiment of the present application needs to mine the evolutionary relationships between the target objects and the geographical environment factors, that is, to construct a driving relationship mining model between the target objects and the geographical environment factors based on the Pytorch framework. This driving relationship mining model mainly includes three layers: data input layer, spatiotemporal feature learning and evolutionary relationship output layer.

[0167] The data input layer is the relationship feature value between the target object and the geographic environmental factors, and the spatiotemporal feature learning is the relationship feature model in the embodiment of the present application, that is, the learning layer (relationship feature model) of geographic spatiotemporal feature mining is formed by CNN and LSTM models, wherein CNN is responsible for processing the spatial features of the target object and the environmental factors, LSTM is responsible for processing the time-dependent features of the target object and the environmental factors, and the evolution relationship output layer is responsible for constructing the relationship feature model (CNN-LSTM model). The specific construction method is shown in the above embodiment and will not be repeated here. The evolution relationship output layer is also responsible for establishing the driving relationship between the geographic environmental factors and the target object, and the specific steps are shown in S504 to S506 below.

[0168] S504: Using the relationship feature model, extract the impact value of the geographical environment factor on the target object.

[0169] Since the graph topology has been established above, in order to use the graph structure (node ​​side) to establish a comprehensive geographical process evolution relationship expression, you can refer to Figure 4 The diagram of the evolution relationship of geographical processes is shown. It is necessary to numerically quantify the topological structure of the graph. That is, it is necessary to extract the impact value of geographical environmental factors on the target object from the relationship feature model, and then quantify the geographical environmental factor nodes according to the impact value.

[0170] S505: normalizing the influence values, and establishing a size piecewise function of the geographical environment factor nodes in the graph topology structure according to the normalized influence values.

[0171] Since the normalized data has the same dimension and range, it is convenient to compare and analyze different data and eliminate the influence caused by the dimension and range differences. Therefore, it is necessary to normalize the impact value and establish the size piecewise function of the geographical environment factor node in the graph topology structure according to the normalized impact value. The specific size piecewise function is shown in the above embodiment and will not be described in detail here.

[0172] S506: Quantitatively display the geographical environment factor nodes according to the size piecewise function.

[0173] Through the size piecewise function, the size of each geographic environment factor node can be determined, and then quantitatively displayed in the graph topology structure, so that the evolution relationship between the geographic process target object and the geographic environment factor node can be more intuitively expressed.

[0174] S507: Calculate using the attribute value of the target object, the attribute value of the geographical environment factor and the evolution relationship strength formula to obtain the evolution relationship strength between the target object and the geographical environment factor.

[0175] The reconstruction of the relationship between geographical process evolution is based on the relationship diagram of geographical process evolution ( Figure 4 ) is used to quantify the evolution relationship strength and the evolution relationship direction. That is, in the embodiment of the present application, the attribute value of the target object, the attribute value of the geographical environment factor and the evolution relationship strength formula are used to calculate the evolution relationship strength between the target object and the geographical environment factor. The specific calculation process is as shown in the above embodiment and will not be repeated here.

[0176] S508: Calculate using the influence value and evolution relationship direction formula to obtain the evolution relationship direction between the target object and the geographical environment factor.

[0177] In the embodiment of the present application, the influence value and evolution relationship direction formula are used to calculate the evolution relationship direction between the target object and the geographical environment factor. The specific calculation process is as shown in the above embodiment and will not be repeated here.

[0178] S509: Using the evolution relationship strength as the attribute of the edge and using the edge style to represent the evolution relationship direction, to obtain a geographic process evolution relationship diagram.

[0179] Since the geographical process evolution relationship diagram is to be reconstructed, it is necessary to use the evolution relationship strength as the attribute of the edge, and use the edge style to express the evolution relationship direction, so as to obtain the geographical process evolution relationship diagram, wherein the edge style can specifically be to use solid line edges to express positive correlation and dotted line edges to express negative correlation. It should be noted that the edge styles in the embodiments of the present application are only used as examples and are not specifically limited.

[0180] Finally, the steps of reconstructing the evolution relationship diagram of the geographical process at each moment are repeated until the evolution relationship diagram of the entire research cycle or the geographical process cycle is constructed.

[0181] S510: In the spatial visualization window, the spatial position, geometric shape and attribute information of the target object at multiple times are displayed in the form of spatiotemporal snapshots, and in the evolution relationship visualization window, the evolution relationship diagram of the geographical process at multiple times is displayed in the form of a graph structure.

[0182] The visualization of the twin objects of geodynamic processes is an intuitive display of the twin objects of geodynamic processes. Therefore, we can first design the spatial visualization and evolution relationship visualization windows, and then obtain the geographic objects (O t (S,A),O t+1 (S,A)) and the dynamic evolution relationship between them (E t t+1) After that, in the spatial visualization window, the spatial position, geometric shape and attribute information of the target object are displayed in the form of spatiotemporal snapshots, and in the evolution relationship visualization window, the geographic dynamic evolution relationship is displayed in the form of a graph structure. Finally, traverse from t to t+1 until the entire research cycle or life cycle. The schematic diagram of the linkage visualization of geographic process objects is as follows: Figure 6 shown.

[0183] In this process, dynamic geographic twin modeling was realized by considering the changes and interactions between temporal target objects and temporal geographic environmental factors, and dynamic linkage visualization was performed through the process twin object visualization method that couples geographic temporal elements / geographic objects and their evolutionary relationships.

[0184] This embodiment provides a twin modeling method based on geographic processes. By defining twin objects in geographic processes, the target object of the geographic process is obtained, and then the relationship characteristics between the target object and the geographic environmental factors, namely the influence value, are mined by using the relationship feature model. Finally, the graph topology structure established by the target object and the geographic environmental factors is quantitatively reconstructed according to the influence value, so that the influence of multiple geographic environmental elements can be coupled to realize the twin modeling of the geographic process.

[0185] See also Figure 7 , the embodiment of the present application provides a twin modeling device 700 based on geographic process, the device comprising:

[0186] A construction unit 701 is used to construct a target object of the geographic process according to the definition of the twin object in the geographic process;

[0187] The construction unit 701 is further used to construct a graph topology structure according to the relationship between the target object and the geographical environment factor, wherein the geographical environment factor includes the geographical environment elements driving the evolution of the geographical process, the target object and the geographical environment factor in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factor in the graph topology structure is an edge;

[0188] An extraction unit 702 is used to extract the impact value of the geographical environment factor on the target object by using a relationship feature model, wherein the relationship feature model is a trained neural network model, and the training samples of the relationship feature model are relationship feature values ​​between the target object and the geographical environment factor;

[0189] The obtaining unit 703 is used to quantitatively reconstruct the graph topology structure by using the influence value to obtain a geographic process evolution relationship graph, wherein the geographic process evolution relationship graph is a graph-based geographic process evolution driving relationship expression model.

[0190] Optionally, the construction unit 701 is used to construct the target object of the geographic process according to the definition of the twin object in the geographic process, specifically to:

[0191] According to the definition of twin objects in the geographic process, determine the spatial location information, geometric morphology information and attribute information required by the target object;

[0192] Construct the target object based on spatial location information, geometric shape information and attribute information.

[0193] Optionally, the device 700 further includes a display unit configured to:

[0194] Constructing a dynamic change evolution relationship of the target object at multiple times based on the spatial position information, geometric shape information and attribute information at multiple times;

[0195] In the spatial visualization window, the spatial position, geometric shape and attribute information of the target object at multiple times are displayed in the form of spatiotemporal snapshots.

[0196] Optionally, the obtaining unit 703 includes:

[0197] A processing subunit is used to normalize the impact value and establish a size piecewise function of the geographical environment factor node in the graph topology structure according to the normalized impact value;

[0198] The quantification subunit is used to quantify and display the geographical environment factor nodes according to the size piecewise function;

[0199] A calculation subunit, used to calculate using the attribute information of the target object, the attribute information of the geographical environment factor and the influence value, to obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor;

[0200] Subunits are obtained for taking the evolution relationship strength as the attribute of the edge and using the edge style to represent the evolution relationship direction, thereby obtaining a geographic process evolution relationship diagram.

[0201] Optionally, the computing subunit is specifically used for:

[0202] Substitute the attribute value of the target object and the attribute value of the geographical environment factor into the evolution relationship strength formula to obtain the evolution relationship strength between the target object and the geographical environment factor. The evolution relationship strength formula includes:

[0203]

[0204] in, is the evolution relationship strength between the target object and the geographical environment factor from time t to t+1; fabs() is an absolute value function to ensure that the relationship strength is positive; G t+1 (Value) and G t(Value) are the attribute information of geographical environment factors at time t+1 and time t respectively; t+1 (Value) and O t (Value) is the attribute information of the target object at time t+1 and time t, respectively, and t is the discrete moment of the entire research cycle or geographical process cycle;

[0205] Substitute the impact value into the evolution relationship direction formula to obtain the evolution relationship direction between the target object and the geographical environment factor. The evolution relationship direction formula includes:

[0206]

[0207] in, is the direction of the evolution relationship between the target object and the geographical environmental factors from time t to t+1; v is the mean value of the impact of the geographical environmental factors on the target object; T in The time series length parameter for time-dependent feature mining input; F v (t i ) is at t i The impact value of geographical environment factors on the target object at each moment.

[0208] Optionally, the construction unit 701 is further used for:

[0209] Set the characteristic values ​​of the target object and the characteristic values ​​of the geographical environment factors;

[0210] Based on the spatial size of the characteristic value of the target object, the geographical environment factor is spatially resampled or interpolated so that the spatial size of the characteristic value of the geographical environment factor is consistent with the spatial size of the characteristic value of the target object;

[0211] Normalize the characteristic values ​​of the geographical environment factors and the characteristic values ​​of the target objects, and construct the characteristic values ​​of the relationship between the target objects and the geographical environment factors;

[0212] Input the relationship feature values ​​into the initial neural network model to conduct mining and training of spatial and temporal features between the target object and the geographical environment factors;

[0213] When the mining training of the initial neural network model is completed, the relational feature model is obtained.

[0214] Optionally, the device 700 further includes a display unit configured to:

[0215] In the evolution relationship visualization window, the evolution relationship diagram of geographical processes at multiple times is displayed in the form of a graph structure.

[0216] It should be noted that the specific implementation method and effect of the twin modeling device 700 based on geographic process can be found in the above Figure 2or Figure 5 The relevant descriptions in the provided methods will not be repeated here.

[0217] The present application embodiment also provides an electronic device 800, such as Figure 8 As shown, the device 800 includes a memory 801 and a processor 802:

[0218] The memory 801 is used to store computer programs;

[0219] The processor 802 is used to execute the above Figure 2 or Figure 5 The method provided.

[0220] In addition, the present application also provides a computer-readable storage medium, the computer-readable storage medium is used to store a computer program, and the computer program is used to execute Figure 2 or Figure 5 The method provided.

[0221] Through the description of the above implementation methods, it can be known that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a read-only memory (ROM) / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network communication device such as a router) to execute the methods described in each embodiment of the present application or some parts of the embodiments.

[0222] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is only schematic, in which the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the goals of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0223] The above description is merely an exemplary embodiment of the present application and is not intended to limit the protection scope of the present application.

Claims

1. A twin modeling method based on geographic process, characterized in that: The method comprises: According to the definition of the twin object in the geographic process, construct the target object of the geographic process; Constructing a graph topology structure according to the relationship between the target object and the geographical environment factor, wherein the geographical environment factor includes the geographical environment elements driving the evolution of the geographical process, the target object and the geographical environment factor in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factor in the graph topology structure is an edge; The influence value of the geographical environment factor on the target object is extracted by using a relationship feature model, wherein the relationship feature model is a trained neural network model, and the training samples of the relationship feature model are relationship feature values ​​between the target object and the geographical environment factor; The graph topology is quantitatively reconstructed using the influence value to obtain a geographic process evolution relationship graph, which is a graph-based geographic process evolution driving relationship expression model.

2. The method according to claim 1, characterized in that The step of constructing the target object of the geographic process according to the definition of the twin object in the geographic process includes: According to the definition of the twin object in the geographic process, determining the spatial location information, geometric morphology information and attribute information required by the target object; The target object is constructed according to the spatial position information, geometric shape information and attribute information.

3. The method according to claim 2, characterized in that The method further comprises: Constructing a dynamic change evolution relationship of the target object at multiple times according to the spatial position information, geometric shape information and attribute information at multiple times; In the spatial visualization window, the spatial position, geometric shape and attribute information of the target object at multiple moments are displayed in the form of spatiotemporal snapshots.

4. The method according to claim 2, characterized in that: The method of quantitatively reconstructing the graph topology structure by using the influence value to obtain a geographic process evolution relationship graph includes: Normalizing the impact value, and establishing a size piecewise function of the geographical environment factor node in the graph topology structure according to the normalized impact value; Quantitatively display the geographical environment factor nodes according to the size piecewise function; Calculating using the attribute information of the target object, the attribute information of the geographical environment factor, and the influence value to obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor; The evolution relationship strength is used as the attribute of the edge, and the evolution relationship direction is represented by the style of the edge, so as to obtain the geographic process evolution relationship diagram.

5. The method according to claim 4, characterized in that The step of calculating using the attribute information of the target object, the attribute information of the geographical environment factor, and the influence value to obtain the evolution relationship strength and evolution relationship direction between the target object and the geographical environment factor includes: Substituting the attribute value of the target object and the attribute value of the geographical environment factor into the evolution relationship strength formula, the evolution relationship strength between the target object and the geographical environment factor is obtained. The evolution relationship strength formula includes: in, is the evolution relationship strength between the target object and the geographical environment factor from time t to t+1; fabs() is an absolute value function to ensure that the relationship strength is positive; G t+1 (Value) and G t (Value) are the attribute information of geographical environment factors at time t+1 and time t respectively; t+1 (Value) and O t (Value) is the attribute information of the target object at time t+1 and time t, respectively, and t is the discrete moment of the entire research cycle or geographical process cycle; Substitute the impact value into the evolution relationship direction formula to obtain the evolution relationship direction between the target object and the geographical environment factor. The evolution relationship direction formula includes: in, is the direction of the evolution relationship between the target object and the geographical environmental factors from time t to t+1; v is the mean value of the impact of the geographical environmental factors on the target object; T in The time series length parameter for time-dependent feature mining input; F v (t i ) is at t i The impact value of geographical environment factors on the target object at each moment.

6. The method according to claim 1, characterized in that The process of constructing the relationship feature model includes: Setting the characteristic value of the target object and the characteristic value of the geographical environment factor; Based on the spatial size of the characteristic value of the target object, spatially resample or interpolate the geographical environment factor so that the spatial size of the characteristic value of the geographical environment factor is consistent with the spatial size of the characteristic value of the target object; Normalizing the characteristic values ​​of the geographical environment factors and the characteristic values ​​of the target object to construct a relationship characteristic value between the target object and the geographical environment factors; Inputting the relationship feature value into an initial neural network model to perform mining training of spatial features and temporal features between the target object and the geographical environment factor; When the mining training of the initial neural network model is completed, the relationship feature model is obtained.

7. The method according to claim 1, characterized in that The method further comprises: In the evolution relationship visualization window, the evolution relationship diagram of the geographical process at multiple times is displayed in the form of a graph structure.

8. A twin modeling device based on geographic process, characterized in that: The device comprises: A construction unit, used to construct a target object of a geographic process according to the definition of a twin object of the geographic process; The construction unit is further used to construct a graph topology structure according to the relationship between the target object and the geographical environment factor, wherein the geographical environment factor includes the geographical environment elements driving the evolution of the geographical process, the target object and the geographical environment factor in the graph topology structure are nodes, and the evolution relationship between the target object and the geographical environment factor in the graph topology structure is an edge; An extraction unit, used to extract the impact value of the geographical environment factor on the target object by using a relationship feature model, wherein the relationship feature model is a trained neural network model, and the training samples of the relationship feature model are relationship feature values ​​between the target object and the geographical environment factor; An obtaining unit is used to quantitatively reconstruct the graph topology structure using the influence value to obtain a geographic process evolution relationship graph, wherein the geographic process evolution relationship graph is a graph-based geographic process evolution driving relationship expression model.

9. An electronic device, characterized in that: The device includes a memory and a processor, and the processor is used to execute a program stored in the memory to run the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store a computer program, and the computer program is used to execute the method according to any one of claims 1 to 7.

Images