Unified earth information model construction method

Abstract

The application provides a unified earth information model construction method, which comprises the following steps: constructing a unified earth information model basic framework, wherein the unified earth information model basic framework is composed of an earth spherical surface grid subdivision system, an underground space three-dimensional grid subdivision system, spatial attributes and time composition; determining the global spherical surface grid subdivision system; determining the underground space three-dimensional grid subdivision system; modeling and assigning values to a three-dimensional grid model based on the global spherical surface grid subdivision system and the underground space three-dimensional grid subdivision system; performing numerical simulation on the three-dimensional grid model; and establishing a global unified information model application system based on the unified earth information model basic framework. The application can meet the requirements of non-uniform, dynamic geological information expression and aboveground and underground integrated data organization, and provides support for establishing a global unified information model application system based on the unified earth information model basic framework.

Description

Technical Field

[0001] The present invention relates to the technical field of geological model building, and in particular to a method for constructing a unified earth information model. Background Art

[0002] Spatiotemporal big data not only promotes knowledge discovery in earth science but also effectively supports major national strategic needs such as resource exploration, environmental protection, and land management. Spatiotemporal big data contains rich spatiotemporal information, encompassing multiple layers of the Earth's interior and surface, and covering millions of years of geological evolution. Its data volume typically exceeds petabytes. Furthermore, different earth science fields generate data types with distinct semantics and formats. Consequently, the immense scale, rich variety, and complex structure of spatiotemporal big data make its organization and management even more complex and challenging.

[0003] The holographic geological model for geological spatiotemporal big data should explore the inherent characteristics of geological data to carry out pioneering research, build a unified earth information model, and form a unified expression model for holographic digital earth spatiotemporal big data. It should not only conform to the current trend of global grid subdivision strategy for spatial information integration in geography and surveying and mapping fields, but also uniformly describe the underground distribution of geological structures and multi-attribute fields, and consider the integrated organization and analysis of above-ground and underground data.

[0004] Therefore, how to construct a unified Earth Information Model (UGIM) and realize a unified expression model of spatiotemporal data of the holographic digital earth is a major fundamental issue for building a holographic geological digital infrastructure and realizing the calculation and sharing services of geological big data. Summary of the Invention

[0005] The purpose of the present invention is to provide a unified earth information model construction method to form a unified expression model of multi-source heterogeneous and spatiotemporal data under the holographic digital earth grid subdivision system.

[0006] In order to achieve the above-mentioned purpose of the invention, the specific technical solutions provided by the present invention are as follows:

[0007] A method for constructing a unified earth information model, the method comprising:

[0008] Constructing a unified earth information model basic framework, which consists of an earth spherical grid subdivision system, an underground space three-dimensional grid subdivision system, spatial attributes and time;

[0009] Determine the global spherical grid subdivision system;

[0010] Determine the three-dimensional grid subdivision system of underground space;

[0011] Modeling and assigning values ​​to three-dimensional grid models based on the global spherical grid subdivision system and the underground space three-dimensional grid subdivision system;

[0012] Perform numerical simulation on three-dimensional grid models;

[0013] Establish a global unified information model application system based on the unified earth information model basic framework.

[0014] Furthermore, the basic framework of the unified earth information model is constructed, including the following operations:

[0015] Determine the coordinate system and projection method;

[0016] Determine the formal expression of the three-dimensional grid model;

[0017] Define the data model for the unified Earth Information Model.

[0018] Furthermore, the three-dimensional mesh model is represented as a tuple<p,A,f(p)> , which is used to describe the attribute value of attribute A at position p. The dimension of p is three-dimensional or four-dimensional. The unified earth information model three-dimensional grid is used as the spatial discrete domain of the three-dimensional grid model. The semantic information of the earth space objects and the spatial distribution information of the attribute parameters are used as the field attributes of the three-dimensional grid model. The three-dimensional grid model is formally defined as:

[0019] UGIMGrid= <p,M,f m (p)>

[0020] Among them, UGIMGrid represents a three-dimensional grid model, M = {m0, m1, m2, ...} is the set of all spatial object attributes m that need to be integrated, f m (p) is a mapping function used to transform the current position p into the three-dimensional grid space R 3 The spatial position within is mapped to the value of attribute m.

[0021] Furthermore, the three-dimensional grid model adopts a global spherical grid subdivision system in the horizontal direction and is divided in the vertical direction according to the discrete requirements of the underground space, so that the grid shape in the horizontal direction is mainly a quadrilateral type in the form of a global spherical grid, and a regular grid or an irregular grid in the vertical direction toward the center of the earth, that is, a triangular prism or a quadrangular prism. The basic form of the interface between two grid units in the same prism is a quadrilateral, and the unit interface is manifested as a geological boundary or a fault plane. The spatial relationship set R of the unit interface is expressed as:

[0022] R={r ij}

[0023] r oj = <F i ,F j>

[0024] r ij Represents the interface F between two different units i and F j The spatial relationship of F i and spatial relations r ij is defined as follows:

[0025] ① In the same spherical subdivision unit, the unit interface F i is a single-valued surface, F i The piecewise function expression is F i =z(x,y), explicit function z i (x,y) is the interpolation function on the discrete unit, describing F i The z value at the (x,y) position;

[0026] ②Spatial relationship r ij Define the topological relationship of whether the interfaces of two units intersect and the directional relationship of the interfaces of two units in the Z direction of the earth's radial direction;

[0027] According to the above definition, we can get:

[0028]

[0029] r ij According to the geological conditions, the use of F i With F j The geological cutting relationship and formation sequence are set.

[0030] Furthermore, the global spherical meshing system is determined, which specifically includes the following operations:

[0031] Determine the grid origin;

[0032] Construct a multi-level scale spherical meshing scheme;

[0033] Setting multi-level resolutions that are compatible with the multi-level scale spherical meshing scheme;

[0034] The spherical grid is encoded according to a multi-level scale spherical grid partitioning scheme.

[0035] Furthermore, the underground space three-dimensional grid subdivision system is determined, which specifically includes the following operations:

[0036] Determine the starting point of the three-dimensional grid;

[0037] Construct a multi-level three-dimensional grid division scheme;

[0038] Setting multi-level resolutions that are compatible with the multi-level three-dimensional grid division scheme;

[0039] The three-dimensional grid is encoded according to a multi-level three-dimensional grid partitioning scheme.

[0040] Furthermore, the three-dimensional grid model based on the global spherical grid subdivision system and the underground space three-dimensional grid subdivision system is modeled and assigned values, which specifically includes the following operations:

[0041] Acquire multi-dimensional geological data, establish a three-dimensional geological structure model based on the multi-dimensional geological data, and generate a three-dimensional geoscientific attribute model of the underground three-dimensional space or the interior of the geological body under the boundary constraints of the three-dimensional geological structure model;

[0042] Integrate 3D geological structure models and 3D geoscientific attribute model data from different sources and convert them into a unified Earth Information Model data model;

[0043] Integrate underground space artificial building model data from different sources and convert them into a unified Earth Information Model data model;

[0044] Based on the data model of the unified earth information model, the three-dimensional grid model is given spatiotemporal semantic attributes, and the grid attributes and attribute values ​​are managed.

[0045] Furthermore, when integrating 3D geological structure models, 3D geoscientific attribute models, or underground space artificial building model data from different sources, the data is classified and identified through the following operations:

[0046] Use the training data set to pre-train the large model, then input the model integration scene classification data set into the large model, adjust the parameters of the large model, and output the large model with adjusted parameters;

[0047] Calculate clustering loss based on the feature distribution of the large model and the small model for the same data instance in the model integration scene classification dataset;

[0048] The small model matches the true label of the data instance through the standard cross entropy loss function to obtain the standard cross entropy loss;

[0049] The overall loss function is determined based on the clustering loss and the standard cross entropy loss. The parameters of the small model are updated through back propagation until the overall loss function converges and the small model is output.

[0050] Classify and identify data through small models.

[0051] Furthermore, the clustering loss is calculated based on the feature distribution of the outputs of the large model and the small model for the same data instance in the model integration scene classification dataset, which specifically includes the following operations:

[0052] For the same data instance in the model integration scene classification dataset, after it is trained with the large model and the small model respectively, the first feature distribution and the second feature distribution are output respectively;

[0053] The first feature distribution and the second feature distribution are respectively mapped to different reproducing kernel Hilbert spaces, and the cluster centers are respectively solved for the feature distributions output when the large model and the small model process multiple data instances with the same label to obtain the first cluster center and the second cluster center;

[0054] Calculate the distance between the first cluster center and the second cluster center, sum the distances between the first cluster center and the second cluster center under all labels, and obtain the clustering loss.

[0055] Furthermore, the numerical simulation of the three-dimensional grid model includes geological process simulation, time attribute coding and extension coding.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The unified earth information model construction method provided by the present invention is aimed at multi-dimensional heterogeneous geological big data. First, a unified earth information model basic framework is constructed. By respectively determining a global spherical grid subdivision system and an underground space three-dimensional grid subdivision system, three-dimensional grid models based on the global spherical grid subdivision system and the underground space three-dimensional grid subdivision system are modeled, assigned values ​​and numerically simulated. The organization and coding theory of the earth's ground and underground space information with multi-attribute fields, multi-scales and multi-level data is developed to meet the needs of non-uniform and dynamic geological information expression and integrated ground and underground data organization, providing support for the establishment of a global unified information model application system based on the unified earth information model basic framework. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only preferred embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0059] Figure 1 The figure is a schematic diagram of the overall process of a method for constructing a unified earth information model provided by an embodiment of the present invention.

[0060] Figure 2 Schematic diagram of a global spherical grid partitioning system provided by an embodiment of the present invention.

[0061] Figure 3 It is a schematic diagram of the underground space three-dimensional grid subdivision system provided by an embodiment of the present invention.

[0062] Figure 4 This is a schematic diagram of grid division of the same earth information model provided by an embodiment of the present invention.

[0063] Figure 5 This is a schematic diagram of a unified global grid coding format provided by an embodiment of the present invention.

[0064] Figure 6 This is a schematic diagram of the expression structure of the same earth information model three-dimensional grid provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The principles and features of the present invention are described below with reference to the accompanying drawings. The enumerated embodiments are only used to explain the present invention and are not used to limit the scope of the present invention.

[0066] Reference Figure 1 This embodiment provides a method for constructing a unified earth information model, which includes the following operations:

[0067] S101. Construct a unified earth information model basic framework, wherein the unified earth information model basic framework is composed of an earth spherical grid subdivision system, an underground space three-dimensional grid subdivision system, spatial attributes, and time.

[0068] S102. Determine the global spherical grid subdivision system.

[0069] S103. Determine the three-dimensional grid division system of the underground space.

[0070] S104. Model and assign values ​​to a three-dimensional grid model based on a global spherical grid subdivision system and an underground space three-dimensional grid subdivision system.

[0071] S105. Perform numerical simulation on the three-dimensional grid model.

[0072] S106: Establish a global unified information model application system based on the unified earth information model basic framework. In step S101, the basic framework of the unified earth information model can be expressed as:

[0073] UGIM=(G,Z,A,T)

[0074] Among them, G represents the global spherical grid subdivision system, which complies with the existing global grid subdivision and spatial data integration requirements, and complies with the relevant standards in the current surveying and mapping and geographic information fields in my country, such as the relevant standards based on GeoSOT. The optional scheme of the global spherical grid subdivision system can also include the spherical subdivision grid based on the longitude and latitude grid. Figure 2 shown.

[0075] Reference Figure 3, Z represents the three-dimensional grid subdivision system of underground space, and Z is the regular or irregular unit division of the spherical grid unit corresponding to the underground space below the earth's surface along the radial direction of the earth. The three-dimensional underground space can be a natural geological space or an artificial underground space building. Since the earth is considered to be a sphere in the global grid subdivision, the geoid is regarded as approximately mapped to the sphere, and the reference reference of Z is the geoid of elevation or the center of the sphere. The spatial discretization of Z along the radial direction of the earth can reflect the geological characteristics or the characteristics of artificial buildings to form the data organization of a unified spatiotemporal data model.

[0076] Reference Figure 4 G and Z together form a global unified underground space grid division, forming a discretized spatial grid model, in which each grid cell is filled with characteristic attributes (geological attributes or other attributes). The attribute values ​​of spatial grid cells can change over time.

[0077] A represents spatial attributes (such as geological attributes or geological semantic information).<id,parent_id,P> , considering the integration of multiple semantic information and the consistency of semantic information and spatial grid, the parent-child association relationship of geological semantic ID is established<id,parent_id> , where P is a specific attribute value or description. Based on semantic associations, spatial grids can be discretized within semantic constraints, ensuring consistency between some geological semantics and spatial representation. For example, stratigraphic ages can be spatially discretized based on the hierarchical structure of boundaries and groups, achieving a multi-level association between semantics and spatial grids.

[0078] T represents time, which specifies the current time instant in the Earth's spatial discretization and the filling of multiple Earth attributes. The internal attribute values ​​of grid cells can change over time.

[0079] In this embodiment, the earth grid based on the basic framework of the unified earth information model needs to continue to be uniformly coded. In the horizontal direction of the earth's surface, the spatial information and semantic attributes of each entity are uniformly coded through the location code under the global spherical grid subdivision system, which facilitates information association, tracking and sharing. In the vertical direction of the underground space, the spatial information of each underground space unit (such as a geological unit or an underground space structure unit) is encoded, and a relatively rough understanding of the entity information at a certain location in the underground space can be obtained through encoding and decoding. The unified coding in the underground space field can be used to: ① characterize the source and migration limitations of multi-factor natural resources; ② as a data encoding format for information sharing, through decoding, a simple and rough understanding of the information is obtained, such as the geographical location, the distribution range in the Z direction, the semantic information of the main underground space units, etc.

[0080] like Figure 5As shown, the global grid unified encoding includes spatial location and range domains, corresponding to the global location code and the underground space code. The underground space code defines the spatial range of attribute distribution within the grid cell. This can be approximated using the z-value range of the irregular top and bottom surfaces of the geological unit or defined by a regular grid. The semantic identification domain is primarily used to encode underground space semantics (such as geological semantics or underground space building semantics) and time. The extension domain code is used to provide specialized extensions to the location code.

[0081] The method provided in this embodiment provides a new perspective and approach for research in the field of earth science by establishing a multi-dimensional, spatiotemporal, unified earth information model basic framework. This basic framework covers the comprehensive consideration of multiple spatial dimensions and time variables, aiming to enhance the characterization, analysis, prediction, early warning, and forecasting of the complexity and dynamic changes of the earth system. Based on the basic framework, a systematic and complete technical process is formed, providing strong support for subsequent applications. The method can be generally divided into the following important steps:

[0082] First, a comprehensive and unified basic framework of the Earth Information Model should be developed to ensure theoretical consistency and scientificity throughout the research process.

[0083] Secondly, based on the horizontal spatial distribution characteristics on a global scale, a suitable global spherical grid division system is designed and determined to achieve accurate expression of the earth's surface characteristics.

[0084] Next, based on the complex structure of underground space, a three-dimensional grid division system for underground space is constructed to fully reflect the multi-level characteristics of the earth's interior.

[0085] At the same time, three-dimensional grid modeling and numerical assignment are carried out, and advanced three-dimensional modeling technology and three-dimensional gridding methods are used to improve the accuracy and reliability of the model.

[0086] Subsequently, numerical simulations based on three-dimensional grid models were carried out to explore the responses and changes of the Earth system under different scenarios.

[0087] Finally, based on the above research results, a global unified information model application system will be established to provide support and guidance for natural resource management, environmental dynamic monitoring, and various scientific research.

[0088] This technical route not only opens up new directions for research in the field of earth information science, but also lays a solid foundation for promoting the development of related applications.

[0089] As a possible implementation method, building a unified earth information model basic framework includes the following operations:

[0090] S201. Determine the coordinate system and projection method.

[0091] In the basic framework of the unified earth information model, the choice of coordinate system and projection method directly determines the accuracy, expression and application effect of geospatial information.

[0092] In this implementation, the unified Earth Information Model coordinate system uses the 2000 National Geodetic Coordinate System. This coordinate system accurately displays global spatiotemporal information, including information on surface cover layers such as topography, landforms, features, and geology, as well as subsurface layers. Users can quickly find their target location by entering coordinates or performing location positioning, and obtain detailed spatial semantic information and location relationships.

[0093] In terms of projection, this implementation uses spherical projection. Spherical projection is a method of projecting the Earth's surface onto a flat surface, maintaining its continuity and integrity while minimizing projection errors. The Unified Earth Information Model uses the Web Mercator projection to divide the Earth's surface into a series of grids, each projected onto a portion of the sphere. This projection method not only achieves full coverage of the Earth's surface, but also ensures the accuracy and clarity of spatiotemporal information.

[0094] S202: Determine a formal expression of the three-dimensional grid model.

[0095] In this embodiment, the formal expression of the three-dimensional grid model is used to uniformly describe spatial objects and continuous field data, and uniformly define the association between spatiotemporal positions and attributes. Specifically, the three-dimensional grid model can be represented as a tuple<p,A,f(p)> , used to describe the attribute value of attribute A at position p, where the dimension of p is three or four dimensions.

[0096] The 3D grid model describes different spatial objects and attribute distributions by combining them into entity units. In essence, the 3D grid model aggregates objects or attribute fields in a spatially discrete domain. Using the unified Earth Information Model 3D grid as the spatially discrete domain of the 3D grid model, and the semantic information of Earth spatial objects and the spatial distribution information of attribute parameters as the field attributes of the 3D grid model, the 3D grid model is formalized as follows:

[0097] UGIMGrid= <p,M,f m (p)>

[0098] Among them, UGIMGrid represents a three-dimensional grid model, M = {m0, m1, m2, ...} is the set of all spatial object attributes m that need to be integrated, f m (p) is a mapping function used to transform the current position p into the three-dimensional grid space R 3 The spatial position within is mapped to the value of attribute m.

[0099] In this implementation, the unified earth information model adopts a grid division strategy under the global spherical grid system. The global spherical grid division system is adopted horizontally, and the underground space (geological interface or geological attribute) is divided vertically according to the discrete requirements, so that the shape of the grid in the horizontal direction is mainly a quadrilateral type in the form of a global spherical grid, and a regular grid or an irregular grid is formed vertically toward the center of the earth, that is, a triangular prism or a quadrangular prism. The subdivision within the grid takes into account geological semantics and spatial boundaries, etc., and different types of grid units can be gradually formed. The global spherical grid system follows the current standards in the field of surveying and mapping and geographic information, or is based on a spherical division grid strategy of a longitude and latitude grid, and the three-dimensional grid is designed as a regular grid or an irregular grid, such as Figure 6 As shown, a spatial multi-scale global grid data model is formed, and segmentation or multi-attribute fusion is performed under semantic constraints.

[0100] In the global spherical grid subdivision system, the basic form of the interface between two grid cells in the same prism is a quadrilateral. The cell interface is manifested as a geological boundary or fault plane. The spatial relationship set R of the cell interface is expressed as:

[0101] R={r ij}

[0102] r ij = <F i ,F j >

[0103] r ij Represents the interface F between two different units i and F j The spatial relationship of F i and spatial relations r ij is defined as follows:

[0104] ① In the same spherical subdivision unit, the unit interface F i is a single-valued surface, F i The piecewise function expression is F i =z(x,y), explicit function z i (x,y) is the interpolation function on the discrete unit, describing F i The z value at the (x,y) position;

[0105] ②Spatial relationship r ij Define the topological relationship of whether the interfaces of two units intersect and the directional relationship of the interfaces of two units in the Z direction of the earth's radial direction;

[0106] According to the above definition, we can get:

[0107]

[0108] r ijAccording to the geological conditions, the use of F i With F j The geological cutting relationships and formation order of the faults can be set. For example, angular unconformities, rock intrusions, and the relationship between primary and secondary faults can be set. Based on spatial relationships, all geological boundaries or fault planes can be linked together to form a spatial relationship binary tree.

[0109] S203: Define a data model of the unified earth information model.

[0110] In this embodiment, the unified earth information model three-dimensional grid is formed in the form of a quadrangular prism, which essentially expresses the interface and spatial relationship of the three-dimensional grid units, thereby realizing the material structure and material composition of the underground space. The unified earth information model grid data organization and unified operation can be achieved by designing SurfaceGrid. SurfaceGrid is a grid model formed along the irregular, discontinuous three-dimensional space division interface of the geological layered structure and fractures in the radial three-dimensional space of the earth. It is mainly manifested in the description of the three-dimensional space interface (such as the stratigraphic level or the fault section). The three-dimensional space grid unit is spliced ​​out through the three-dimensional space interface, and the three-dimensional space grid unit is manifested as a polyhedron as a whole.

[0111] The quadrilateral unit interface object formed in the quadrangular prism is called SurfaceCell. The geological interface and the fault plane F are composed of SurfaceCell, that is:

[0112] F={SurfaceCell ij}={ <z i,j ,z i+1,j ,z i+1,j+1 ,z i,j+1 ,id,treeIndex>}

[0113] A SurfaceCell consists of six attributes: the z-values ​​of the four grid points, the SurfaceCell's geological attribute ID, and the index of its spatial binary tree node. The SurfaceGrid object can be used to retrieve the boundaries and internal properties of each spatial object.

[0114] Determining a global spherical gridding system is a crucial foundation for building efficient Earth system models, involving numerous technical and theoretical details. First, determining the grid origin is a key step in this system. This requires considering the distribution of global geographic coordinates to ensure that the selected origin effectively covers the entire sphere and achieves a uniformly distributed grid layout. Second, the design of a multi-level gridding scheme is crucial, providing flexible solutions tailored to diverse research needs and supporting applications ranging from global to regional scales. For example, for meteorological, oceanographic, or geological research, appropriate gridding levels can be selected to achieve computational efficiency while enhancing model detail. Furthermore, the selection of multiple levels of resolution should be coordinated with the needs of specific applications. Different levels of resolution can help researchers better capture the characteristics of specific phenomena and provide more accurate data input for numerical simulations. Finally, spherical grid encoding is a crucial technology for efficient computation and data exchange. It involves storing and transmitting three-dimensional spherical grid information in an appropriate encoding format to facilitate subsequent data processing and analysis. Through in-depth research and development in the above aspects, a comprehensive, efficient and adaptable global spherical grid division system can be realized, laying a solid foundation for related scientific research and practical applications.

[0115] Based on this, as another possible implementation, determining a global spherical meshing system specifically includes the following operations:

[0116] S301. Determine the grid origin.

[0117] When constructing a unified Earth Information Model, properly determining the grid origin is a crucial step. This process not only requires considering the Earth's geometric characteristics as a near-spherical shape, but also requires in-depth analysis of the specific distribution of global geographic coordinates.

[0118] Determining the grid origin requires comprehensive consideration of the global geographic coordinate distribution characteristics to ensure that the selected origin position can comprehensively and effectively cover the entire surface of the earth, thereby achieving a uniformly distributed grid layout.

[0119] The ideal grid origin selection should meet the following requirements:

[0120] Comprehensive coverage: The origin should be located in a position that ensures that the gridding rules used from that point will cover the entire Earth's surface without omission. This means avoiding locations where it would be difficult to include parts of the Earth in the grid system, such as the polar regions or other geographical anomalies.

[0121] Uniform distribution: To ensure that the generated grid is evenly distributed across the Earth's surface and minimize density variations due to geographical location, the origin should be chosen based on the Earth's natural topography (such as topography and landforms) and the distribution of human activities. Ideally, the area of ​​each grid cell should be as uniform as possible, which improves data processing efficiency and accuracy and facilitates subsequent spatial analysis and modeling.

[0122] Computational simplicity: From a mathematical and computational perspective, choosing an origin that simplifies coordinate transformations and mesh generation algorithms is also crucial. A good origin location can help optimize the computational process, reduce error accumulation, and improve overall system performance.

[0123] Compatibility and standardization: Considering that different countries and regions may use different geographic coordinate systems, choosing an internationally recognized reference point as the grid origin helps enhance system interoperability and data sharing capabilities. This usually means following international standards such as WGS84 (World Geodetic System 1984) to ensure compatibility with other geographic information systems or datasets.

[0124] Determining the grid origin is a multi-factor process that directly impacts the quality of subsequent grid layout and its effectiveness. Careful design and scientific justification, choosing the right origin location, can lay a solid foundation for creating an efficient, accurate, and highly compatible unified Earth Information Model.

[0125] Determining the grid origin in the global spherical grid system primarily involves longitude and latitude, which can be customized by the user. For example, two possible origin options are: one is to select 0 degrees east longitude and 0 degrees latitude; the other is to select 0 degrees east longitude and 90 degrees south latitude.

[0126] S302. Construct a multi-level scale spherical grid division scheme.

[0127] In this implementation, a multi-scale spherical meshing scheme is constructed to support diverse applications, from global to regional scales, with global coverage. For example, for surface and geological research, the appropriate meshing level can be selected to achieve a more detailed model while maintaining computational efficiency.

[0128] Exemplarily, the grid division levels are planned to be divided into 27 levels, from level 0 to level 26.

[0129] The Earth's longitude ranges from -180 degrees to 180 degrees, and its latitude ranges from -90 degrees to 90 degrees.

[0130] Level 0 grid:

[0131] Longitude ranges from 0 degrees to 180 degrees (i.e., from the prime meridian to the east), and is divided into a grid every +8 degrees. Longitude ranges from 0 degrees to -180 degrees (i.e., from the prime meridian to the west), and is divided into a grid every -8 degrees. Latitude ranges from 0 degrees to 90 degrees (i.e., from the equator to the north), and is divided into a grid every +4 degrees. Latitude ranges from 0 degrees to -90 degrees (i.e., from the equator to the south), and is divided into a grid every -4 degrees.

[0132] Level 1 to 3 grids: Level 1 grid is based on the grid size of level 0 in the longitude and latitude directions, with the grid size divided by 2 respectively. Levels 1 to 3 are divided by 2 in equal proportions. At level 5, the grid size in the longitude direction is 1 degree, and the grid size in the latitude direction is 1 degree.

[0133] Level 4 grid: longitude from the prime meridian to the east is divided into one grid every +32 minutes, longitude from the prime meridian to the west is divided into one grid every -32 minutes; latitude from the equator to the north is divided into one grid every +16 minutes, latitude from the equator to the south is divided into one grid every -16 minutes.

[0134] Level 5 to 9: The level 5 grid is based on the level 4 grid with the grid sizes in the longitude and latitude directions divided by 2 respectively. For levels 5 to 9, the grid sizes are divided by 2 in equal proportions. At level 5, the grid size in the longitude direction is 1 minute, and the grid size in the latitude direction is 1 minute.

[0135] Level 10 grid: longitude from the prime meridian to the east is divided into 1 grid every +32 seconds, longitude from the prime meridian to the west is divided into 1 grid every -32 seconds; latitude from the equator to the north is divided into 1 grid every +16 seconds, latitude from the equator to the south is divided into 1 grid every -16 seconds.

[0136] Level 11 to 15: The grid size of level 11 is based on the grid size of level 10 with the grid size in the longitude and latitude directions divided by 2 respectively. For levels 11 to 15, the grid size is divided by 2 in geometric proportions. At level 17, the grid size in the longitude direction is 1 second, and the grid size in the latitude direction is 1 second.

[0137] Level 16 to 26: The 16-level grid is based on the 15-level grid, with the grid sizes in the longitude and latitude directions divided by 2 respectively. For levels 16 to 26, the grid sizes are divided by 2 in geometric proportions. At level 25, the grid size in the longitude direction is 1 / 2048 second, and the grid size in the latitude direction is 1 / 2048 second.

[0138] S303: Setting a multi-level resolution that is compatible with a multi-level scale spherical grid division scheme.

[0139] In this implementation, the specification of multiple levels of resolution is aligned with a multi-scale spherical meshing scheme. Different levels of resolution help researchers better store, manage, and characterize three-dimensional spatiotemporal features, providing more accurate data input for numerical simulations. Based on this multi-level meshing scheme, the resolutions of the 27 levels of meshing are shown in the table below.

[0140] Table 1

[0141]

[0142]

[0143] S304 , encoding the spherical grid according to a multi-level scale spherical grid division scheme.

[0144] In this step, the spherical grid code consists of 27 bits, and the value of each bit is a quaternary value 0, 1, 2, 3. The coding order is Z-sequence.

[0145] When determining a 3D gridding system for underground space, multiple professional factors must be comprehensively considered to ensure scientific and practical results. First, the starting point of the 3D grid, the foundation of the 3D gridding system, must be precisely selected based on the specific characteristics of the underground space and application requirements. For example, natural factors such as topography, geomorphology, and geology, as well as artificial factors such as above- and underground structures, should be considered. A reasonable coordinate system should be established based on this to ensure the accuracy and consistency of subsequent 3D gridding. Second, the design of a multi-level gridding scheme should be guided by functional requirements and data accuracy, allowing for autonomous adjustment of gridding at different levels to accommodate various applications, such as urban planning, underground resource development, and environmental monitoring. Furthermore, the appropriate setting of the multi-level resolution within the 3D gridding scheme is crucial. This requires precise control at different levels to accommodate analyses of underground spaces of varying scale and complexity without sacrificing data availability. Finally, the systematic design of 3D grid coding will enhance data manageability and interactivity. Through unified coding rules, efficient storage, retrieval, and sharing of 3D grid data can be achieved, thereby promoting the deep integration of various related research and practical applications. Therefore, systematic research and integration of all the above aspects will provide a solid foundation for the scientific management of underground space.

[0146] As another possible implementation, determining the underground space three-dimensional grid subdivision system specifically includes the following operations:

[0147] S401: Determine the starting point of the three-dimensional grid.

[0148] The starting point of the three-dimensional grid is controlled by the Earth's surface topography and determined by the global spherical grid system. The grid resolution is set according to the 27-level spherical multi-level division scheme and is divided radially toward the Earth. Each level of the grid is horizontally consistent with the level and resolution of the global spherical grid. The terrain data uses 27-level DEM data, and the accuracy of the DEM data is consistent with the resolution of the 27-level global spherical grid.

[0149] S402: Construct a multi-level three-dimensional grid division scheme.

[0150] The design of the multi-level 3D gridding scheme is guided by functional requirements and data accuracy. By autonomously adjusting the gridding at different levels, it adapts to various applications, such as geological resource management, geological environmental protection, geological disaster prevention and control, and underground space utilization. For example, the multi-level 3D gridding scheme considers the Earth's internal gridding, which is primarily controlled by the Earth's radius, which is 6,378,137 meters along its major axis. The 3D gridding level is the same as the global spherical grid level, divided into 27 levels, from level 0 to level 26.

[0151] S403: Setting a multi-level resolution that is compatible with the multi-level three-dimensional grid division scheme.

[0152] Properly setting multiple levels of resolution can adapt to underground space applications of different scales and complexities without sacrificing data availability. For example, the multi-level resolution setting scheme of the underground space three-dimensional grid subdivision system in this embodiment is shown in Table 2.

[0153] Table 2

[0154]

[0155]

[0156] S404: Encode the three-dimensional grid according to a multi-level three-dimensional grid division scheme.

[0157] The underground space three-dimensional grid code consists of 27 bits according to the multi-level three-dimensional grid division scheme, and the value of each bit is a binary value 0 or 1.

[0158] Within the unified Earth Information Model framework, after the Earth is divided into three-dimensional grids, specialized applications require the assignment of spatiotemporal semantic attributes within these grids, such as geological age, stratigraphy, structure, lithology, geochemical properties, geophysical properties, hydrogeological parameters, engineering geological parameters, and ore grade. Three-dimensional geological modeling is the primary step in the grid modeling and assignment process. Its goal is to construct an efficient model reflecting subsurface geological characteristics through the collection and analysis of geological data. This model must not only comprehensively consider multiple geological elements, such as stratigraphy, structure, and ore bodies, but also utilize advanced modeling software and algorithms to achieve three-dimensional visualization of spatial data, enabling researchers and engineers to more intuitively understand the geological environment. Next, geological model integration involves effectively integrating multiple independent geological models to ensure data consistency and integrity. This process emphasizes interdisciplinary collaboration and data sharing to maximize the applicability and reliability of the model. Furthermore, the integration of other models is crucial, including models of underground structures related to the geological model space, to enable multi-dimensional analysis and decision support. Finally, attribute assignment and management are crucial. Their mission is to assign specific physical, chemical, and biological properties to each cell in the 3D grid model. This not only enhances the model's realism but also provides a solid foundation for subsequent simulations and predictions. The technical approach to attribute assignment involves developing multi-source, multi-scale voxelization technology to convert 3D vector models into 3D raster models within the Unified Earth Information Model framework. Accurately organizing and efficiently managing attribute data throughout this process ensures the scientific and practical validity of the model, ultimately serving diverse fields such as resource assessment, environmental protection, and engineering design.

[0159] As another possible implementation, modeling and assigning values ​​to a three-dimensional grid model based on a global spherical grid subdivision system and an underground space three-dimensional grid subdivision system may specifically include the following operations:

[0160] S501. Acquire multi-dimensional geological data, establish a three-dimensional geological structure model based on the multi-dimensional geological data, and generate a three-dimensional geoscientific attribute model of the underground three-dimensional space or the interior of the geological body under the boundary constraints of the three-dimensional geological structure model.

[0161] For example, the 3D geological structure model can be constructed using either a knowledge-driven explicit modeling approach or a data-driven implicit modeling approach. The 3D geoscientific attribute model is generated by converting multi-dimensional geological data into a potential field in space to generate a high-precision 3D geoscientific attribute model.

[0162] S502: Integrate the three-dimensional geological structure model and three-dimensional geoscientific attribute model data from different sources and convert them into a unified earth information model data model.

[0163] This step focuses on 3D geological model data exchange, enabling the reading, parsing, extraction, conversion, validation, and generation of models from multiple sources, and the unified conversion to a unified Earth Information Model (EIM) data model. Supported data formats include OBJ, GOCAD, Petrel, EVS, and Geo3DML. Dataset geometry types include drillholes, sections, surfaces, vector volumes, and attribute models.

[0164] S503. Integrate underground space artificial building model data from different sources and convert them into a unified earth information model data model.

[0165] For example, underground artificial building models primarily include BIM models of underground infrastructure, underground pipelines, and subways. This step allows for the reading, parsing, extraction, conversion, verification, and generation of underground artificial building models, converting them into a unified Earth Information Model data model. Supported data formats include IFC.

[0166] S504: Based on the data model of the unified earth information model, the three-dimensional grid model is given spatiotemporal semantic attributes, and the grid attributes and attribute values ​​are managed. This step mainly includes the following operations:

[0167] 1) The three-dimensional grid model gives spatiotemporal semantic attributes

[0168] Time dimension: Adding a time axis to a 3D grid model can be used to track how specific geological features change over time, such as the growth of sedimentary layers or erosion processes.

[0169] Spatial semantics: Each cell (or voxel) in a 3D grid model is given information describing its location, size, and relationship to other cells. This helps to more accurately locate geological objects and understand their interactions.

[0170] 2) Grid attributes and attribute value management

[0171] Define properties: First, determine which properties are important for your research objectives. These might include physical properties like rock type, permeability, and porosity; or environmental conditions like chemical composition, temperature, and pressure.

[0172] Assignment method: Assign specific attribute values ​​to grid cells based on actual measurement data or inference based on existing knowledge. This process often involves advanced technical means such as statistical analysis and the application of machine learning algorithms.

[0173] Dynamic Updates: As new data is acquired, the attribute values ​​in the model need to be adjusted to reflect the latest state of knowledge. These changes can be managed and maintained by establishing an effective dynamic update mechanism.

[0174] 3) Spatiotemporal semantic code

[0175] The semantic identification domain is encoded, mainly encoding the underground space semantics (such as geological semantics or underground space building semantics).

[0176] As a further possible implementation, when integrating 3D geological structure models, 3D geophysical attribute models, or underground space artificial building model data from different sources, the data is classified and identified through the following operations:

[0177] S601. Use the training data set to pre-train the large model, then input the model integration scene classification data set into the large model, adjust the parameters of the large model, and output the large model after parameter adjustment.

[0178] S602. Calculate clustering loss based on the feature distribution output by the large model and the small model for the same data instance in the model integration scene classification dataset.

[0179] Large models offer superior performance compared to small models, but are also larger and more complex. Small models, while inferior in performance, are more compact and easier to deploy on various platforms.

[0180] S603. The small model matches the true label of the data instance through the standard cross entropy loss function to obtain the standard cross entropy loss.

[0181] S604: Determine the overall loss function based on the clustering loss and the standard cross entropy loss, update the parameters of the small model through back propagation until the overall loss function converges, and output the small model.

[0182] S605: Classify and identify data using a small model.

[0183] In this implementation, an overall loss function based on clustering loss and standard cross entropy loss is established to transfer the excellent feature extraction capability of the large model to the small model, so that the small model can efficiently and accurately identify and classify data from different sources in the model integration scenario, thereby improving the efficiency of model integration.

[0184] The clustering loss is calculated based on the feature distribution of the outputs of the large model and the small model for the same data instance in the model integration scene classification dataset, which specifically includes the following operations:

[0185] S701. For the same data instance in the model integrated scene classification dataset, after it is trained with a large model and a small model respectively, a first feature distribution and a second feature distribution are output respectively.

[0186] S702. Map the first feature distribution and the second feature distribution to different reproducing kernel Hilbert spaces respectively, and solve the cluster centers of the feature distributions output when the large model and the small model process multiple data instances with the same label respectively to obtain the first cluster center and the second cluster center.

[0187] S703: Calculate the distance between the first cluster center and the second cluster center, sum the distances between the first cluster center and the second cluster center under all labels, and obtain the clustering loss.

[0188] After assigning values ​​to the three-dimensional grid model, finite element technology and artificial intelligence technology can be used to perform data simulation on the multi-scale model to reflect the dynamic evolution of complex geological processes.

[0189] As another possible implementation, a numerical simulation is performed on the three-dimensional grid model, including geological process simulation, time attribute coding, and extension coding.

[0190] First, in terms of geological process simulation, the three-dimensional grid model can be used to simulate the mineralization process. By accurately modeling the formation and evolution of ore bodies and their interaction with the surrounding geological environment, it helps researchers to deeply understand the distribution patterns of mineral resources and their genesis mechanisms.

[0191] In addition, geothermal simulation is another key application area. The three-dimensional grid model can simulate the heat conduction, fluid flow and its interaction with geological structures of the geothermal system, providing a scientific basis for the development and utilization of geothermal resources.

[0192] At the same time, groundwater simulation is also one of the important applications of the three-dimensional grid model. Through accurate simulation of groundwater flow, pollutant migration and hydrogeological characteristics, it can provide strong support for water resources management and environmental protection.

[0193] In this simulation process, the introduction of temporal attribute coding is crucial. It not only integrates time into the model, making simulation results more timely and accurate, but also helps researchers analyze the dynamics of geological processes at different time scales, providing a more comprehensive basis for decision-making in geological research and resource development. This implementation incorporates temporal attribute coding into the unified global grid code.

[0194] Extended coding is used to provide professional extensions to position coding.

[0195] Against the backdrop of increasing globalization and informatization, building a globally unified information model application system that enables cross-regional and cross-industry data sharing and collaboration is crucial for promoting the effective integration and utilization of information resources. The design and implementation of this system aims to break down information silos, improve decision-making efficiency, and enhance inter-organizational collaboration through the collaborative operation of a series of subsystems. Specifically, the globally unified information model application system includes the following key components:

[0196] 1. Model storage system

[0197] As the foundation of a globally unified information model application system, the model storage system is responsible for the efficient and secure storage and management of various information models. This system not only needs to possess robust data storage capabilities to accommodate the storage and access needs of massive amounts of data, but also provide flexible data access interfaces, supporting the import and export of multiple data formats to ensure data consistency and integrity. The system manages large-scale, multi-scale, multi-precision, multi-parameter, vector-grid integrated, high-precision three-dimensional geological attribute models, focusing on key technical issues such as distributed large-scale grid data storage, a distributed grid data computation framework, data resource access interfaces, computational service access interfaces, and lightweight data processing.

[0198] The model storage system provides grid database management and grid dataset management. Grid database management includes functions such as establishing a grid database, connecting to a grid database, modifying a grid database, and deleting a grid database. Grid dataset management includes functions such as creating a new grid model dataset group, reading a model dataset group, deleting a grid model dataset group, renaming a grid model dataset group, creating a new grid model dataset, reading a grid model dataset, writing a grid model dataset, deleting a grid model dataset, updating a grid model dataset, renaming a grid model dataset, creating a new grid model dataset field, deleting a grid model dataset field, and editing a grid model dataset field.

[0199] 2. Model publishing system

[0200] The model publishing system bridges the gap between model developers and end users. Through standardized 3D service interfaces, it enables unified service management, thematic scenario customization, service publishing, rapid deployment, and widespread dissemination of the unified Earth Information Model. This system implements service registration, service application, service approval, service publishing, and service management. It features a robust user experience, simplifies the model publishing process, and lowers the barrier to entry for even non-professionals.

[0201] The Unified Earth Information Model 3D Service Interface (UGIM3D API) follows the RESTful design specification and can be called through a 3D client.

[0202] The interface contains the following:

[0203] UGIM3DData: Data information acquisition service, used to obtain the data metadata information of the service, describing the basic information of the data, version number, data name, data type, spatial reference system and other information.

[0204] UGIM3DSharedResources: Public resource acquisition service, used to obtain public resource data under UGIM3D data, including material, texture, and feature information.

[0205] UGIM3DRootNodeInfo: Root node information acquisition service, used to obtain metadata information of the root node under UGIM3D data, describing information such as node name, node LOD level, node outer sphere range, etc.

[0206] UGIM3DNodeInfo: Node description information acquisition service, used to obtain metadata information of child nodes under the root node under the data, describing information such as node name, node LOD level, node outer sphere range, etc.

[0207] UGIM3DNodeData: Node data information acquisition service, used to obtain data information associated with a node, including geometry, attributes, material, and texture data.

[0208] UGIM3DGeometryData: Geometry information acquisition service, used to obtain feature geometry information associated with a node.

[0209] UGIM3DTextureData: Texture information acquisition service, used to obtain the texture information of the specified texture on the node.

[0210] UGIM3DAttributeData: Attribute information acquisition service, used to obtain feature attribute information associated with a node.

[0211] 3. Model analysis system

[0212] The model analysis system focuses on deep mining and intelligent analysis of a unified multidimensional spatiotemporal Earth information model. Leveraging machine learning and deep learning data analysis technologies, it implements virtual geological drilling analysis, virtual geological profile analysis, virtual geological excavation analysis, virtual geological block analysis, point-based holographic analysis, line-based holographic analysis, surface-based holographic analysis, scenario simulation, and forecasting and early warning functions, providing users with data insights and decision-making support. The system automatically identifies key information within the model, discovers potential data correlations, and assists users in developing more scientific and rational strategic plans. Through customized report generation, the model analysis system also presents complex analysis results to users in an intuitive and easy-to-understand format, enhancing the transparency and accuracy of decision-making.

[0213] 4. Professional application system

[0214] Specialized application systems are targeted at specific industries or fields, providing highly customized solutions based on unified Earth information models at multiple scales, including national, regional, provincial, municipal, and site-level. These applications encompass key regional digital mineralization belts, digital basins, digital watersheds, digital coastal zones, underground space, and digital oceans. Working closely with model storage, publishing, and analysis systems, these systems leverage existing information resources through multi-domain demonstration applications, achieving deep integration and integrated services with the Real 3D China and national land and space basic information platforms to meet the needs of diverse user groups. These systems typically offer diverse interactive interfaces and service models, providing crucial support for 3D natural resource surveys and oversight, deep mineral resource evaluation, and underground space development and utilization.

[0215] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for constructing a unified earth information model, characterized in that: The method comprises: Constructing a unified earth information model basic framework, which consists of an earth spherical grid subdivision system, an underground space three-dimensional grid subdivision system, spatial attributes and time; Determine the global spherical grid subdivision system; Determine the three-dimensional grid subdivision system of underground space; Modeling and assigning values ​​to three-dimensional grid models based on the global spherical grid subdivision system and the underground space three-dimensional grid subdivision system; Perform numerical simulation on three-dimensional grid models; Establish a global unified information model application system based on the unified earth information model basic framework; Constructing a unified earth information model basic framework includes the following operations: Determine the coordinate system and projection method; Determine the formal expression of the three-dimensional grid model; Define the data model for the unified Earth Information Model; The three-dimensional mesh model is represented as a tuple<p,A,f(p)> , which is used to describe the attribute value of attribute A at position p. The dimension of p is three-dimensional or four-dimensional. The unified earth information model three-dimensional grid is used as the spatial discrete domain of the three-dimensional grid model. The semantic information of the earth space objects and the spatial distribution information of the attribute parameters are used as the field attributes of the three-dimensional grid model. The three-dimensional grid model is formally defined as: UGIMGrid= <p,M,f m (p)> Among them, UGIMGrid represents a three-dimensional grid model, M = {m0, m1, m2, ...} is the set of all spatial object attributes m that need to be integrated, f m (p) is a mapping function used to transform the current position p into the three-dimensional grid space R 3 The spatial position within is mapped to the value of attribute m; The three-dimensional grid model adopts the global spherical grid subdivision system in the horizontal direction and is divided in the vertical direction according to the discrete requirements of the underground space. The horizontal grid shape is mainly a quadrilateral type in the form of the global spherical grid. In the vertical direction, a regular grid or an irregular grid is formed toward the center of the earth, that is, a triangular prism or a quadrangular prism. The basic form of the interface between two grid cells in the same prism is a quadrilateral. The cell interface is manifested as a geological boundary or a fault plane. The spatial relationship set R of the cell interface is expressed as: R＝{r ij} Represents the interface F between two different units i and F j The spatial relationship of F i and spatial relationships is defined as follows: ① In the same spherical subdivision unit, the unit interface F i is a single-valued surface, F i The piecewise function expression is F i =z(x,y), explicit function z i (x,y) is the interpolation function on the discrete unit, describing F i The z value at the (x,y) position; ②Spatial relationship Define the topological relationship of whether the interfaces of two units intersect and the directional relationship of the interfaces of two units in the Z direction of the earth's radial direction; According to the above definition, we can get: r ij According to the geological conditions, the use of F i With F j The geological cutting relationship and formation sequence are set.

2. A method for constructing a unified earth information model according to claim 1, characterized in that: Determine the global spherical meshing system, including the following operations: Determine the grid origin; Construct a multi-level scale spherical meshing scheme; Setting multi-level resolutions that are compatible with the multi-level scale spherical meshing scheme; The spherical grid is encoded according to a multi-level scale spherical grid partitioning scheme.

3. A method for constructing a unified earth information model according to claim 2, characterized in that: Determine the underground space three-dimensional grid subdivision system, which specifically includes the following operations: Determine the starting point of the three-dimensional grid; Construct a multi-level three-dimensional grid division scheme; Setting multi-level resolutions that are compatible with the multi-level three-dimensional grid division scheme; The three-dimensional grid is encoded according to a multi-level three-dimensional grid partitioning scheme.

4. A method for constructing a unified earth information model according to claim 1, characterized in that: Modeling and assigning values ​​to the three-dimensional grid model based on the global spherical grid subdivision system and the underground space three-dimensional grid subdivision system includes the following operations: Acquire multi-dimensional geological data, establish a three-dimensional geological structure model based on the multi-dimensional geological data, and generate a three-dimensional geoscientific attribute model of the underground three-dimensional space or the interior of the geological body under the boundary constraints of the three-dimensional geological structure model; Integrate 3D geological structure models and 3D geoscientific attribute model data from different sources and convert them into a unified Earth Information Model data model; Integrate underground space artificial building model data from different sources and convert them into a unified Earth Information Model data model; Based on the data model of the unified earth information model, the three-dimensional grid model is given spatiotemporal semantic attributes, and the grid attributes and attribute values ​​are managed.

5. A method for constructing a unified earth information model according to claim 4, characterized in that: When integrating 3D geological structure models, 3D geoscientific attribute models, or underground space artificial building model data from different sources, the following operations are performed to classify and identify the data: Use the training data set to pre-train the large model, then input the model integration scene classification data set into the large model, adjust the parameters of the large model, and output the large model with adjusted parameters; Calculate clustering loss based on the feature distribution of the large model and the small model for the same data instance in the model integration scene classification dataset; The small model matches the true label of the data instance through the standard cross entropy loss function to obtain the standard cross entropy loss; The overall loss function is determined based on the clustering loss and the standard cross entropy loss. The parameters of the small model are updated through back propagation until the overall loss function converges and the small model is output. Classify and identify data through small models.

6. A method for constructing a unified earth information model according to claim 5, characterized in that: The clustering loss is calculated based on the feature distribution of the large model and the small model for the same data instance in the model integration scene classification dataset. The specific operations include the following: For the same data instance in the model integration scene classification dataset, after it is trained with the large model and the small model respectively, the first feature distribution and the second feature distribution are output respectively; The first feature distribution and the second feature distribution are respectively mapped to different reproducing kernel Hilbert spaces, and the cluster centers are respectively solved for the feature distributions output when the large model and the small model process multiple data instances with the same label to obtain the first cluster center and the second cluster center; Calculate the distance between the first cluster center and the second cluster center, sum the distances between the first cluster center and the second cluster center under all labels, and obtain the clustering loss.

7. A method for constructing a unified earth information model according to claim 1, characterized in that: The numerical simulation of the three-dimensional grid model includes geological process simulation, time attribute coding and expansion coding.

Images