Natural resource asset management and territorial space planning system fusion method and system

Abstract

The invention relates to a natural resource asset management and territorial space planning system fusion method and system. The method comprises the following steps: encoding territorial space planning vector data and natural resource asset registration data to generate a space-time binding code; constructing a spatial topological graph and a time sequence diagram based on the codes, respectively extracting planning constraint feature vectors and asset dynamic feature vectors, and fusing the vectors to generate a planning asset coupling scoring matrix; recognizing a conflict unit set based on the planning asset coupling scoring matrix; based on the conflict unit set and a preset business rule base, performing classification processing on each conflict unit in the conflict unit set to obtain a conflict type; and based on the conflict type, generating a service execution instruction set through a preset reinforcement learning decision model. According to the method, through space-time unified coding and feature fusion processing, the accuracy of data association and the scientificity of decision making are improved, so that collaborative governance of territorial space planning and natural resource asset management is realized.

Description

Technical Field

[0001] This invention belongs to the field of computer technology, and in particular relates to a method and system for integrating natural resource asset management and land spatial planning. Background Technology

[0002] In the collaborative governance of natural resource asset management and territorial spatial planning, with the continuous increase in the demand for refined territorial spatial development and protection, and coordinated ecological protection and economic development, achieving deep integration and intelligent decision-making between the two types of data has become a core industry requirement. However, currently, territorial spatial planning vector data and natural resource asset registration data belong to different management systems, with differences in data formats and identification standards, and a lack of a unified spatiotemporal correlation coding mechanism. This results in the formation of "information silos" between the two types of data, making it impossible to achieve precise binding of spatial location and asset attributes and cross-scenario reuse.

[0003] Furthermore, in the feature extraction stage, traditional methods often analyze the spatial attributes of planning data or the static attributes of asset data separately, failing to simultaneously capture the topological constraints of national land space and the dynamic flow patterns of natural resource assets. This results in extracted features with limited dimensions, making it difficult to comprehensively reflect the coupling relationship between planning and assets. At the conflict resolution and decision-making level, existing conflict resolution technologies largely rely on manually preset simple rules to identify the compatibility conflicts between planning and assets, lacking a coupled scoring system based on quantitative indicators. This leads to low accuracy and efficiency in identifying conflict units. Moreover, the decision-making process lacks the support of intelligent optimization models, making it difficult to generate optimal solutions by combining multiple dimensions such as asset value levels and planning priorities. This often results in inefficient resource allocation and an imbalance between ecological protection and economic development. In addition, existing fusion methods have not formed a technical chain of data encoding-feature fusion-conflict identification-intelligent decision-making. The connections between each stage are loose, failing to achieve fully automated processing from data association to decision execution, and thus failing to meet the actual needs of efficient, accurate, and intelligent fusion in the new era of national land space governance. Summary of the Invention

[0004] Therefore, it is necessary to provide methods and systems for integrating natural resource asset management and territorial spatial planning to address the aforementioned technical issues, aiming to improve the accuracy and efficiency of data integration and enhance the scientific and intelligent level of decision-making.

[0005] Firstly, this application provides a method for integrating natural resource asset management with the national spatial planning system, including:

[0006] Spatiotemporal unified coding processing is performed on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes;

[0007] Based on spatiotemporal binding coding, a spatial topology map and a temporal sequence map are constructed. Planning constraint feature vectors and asset dynamic feature vectors are extracted from the spatial topology map and the temporal sequence map, respectively. The planning constraint feature vectors and asset dynamic feature vectors are then fused to generate a planning-asset coupling scoring matrix.

[0008] The set of conflicting units is identified based on the planning asset coupling scoring matrix; each conflicting unit in the set of conflicting units is classified based on the set of conflicting units and the preset business rule base to obtain the conflict type; based on the conflict type, a set of business execution instructions is generated through a preset reinforcement learning decision model.

[0009] In one embodiment, spatiotemporal unified coding processing is performed on land spatial planning vector data and natural resource asset registration data to generate a unified spatiotemporal binding code, including:

[0010] The GeoSOT-3D subdivision algorithm was used to perform three-dimensional mesh subdivision processing on the land spatial planning vector data to obtain the subdivided mesh.

[0011] Based on the coordinates of the center points of land parcels in the subdivided grid and the vector data of the national land spatial planning, a spatial reference code is generated. The spatial reference code contains longitude grid chains, latitude grid chains, elevation identifiers and timestamp information.

[0012] The data on natural resource asset registration is analyzed and processed to obtain resource type, ownership identifier, and value level;

[0013] The resource type, ownership identifier, and value level are encoded to generate an asset attribute code, which includes administrative division code, resource classification code, ownership identifier, and value level information.

[0014] Spatial location relationship analysis is performed on the spatial entities corresponding to the spatial reference code and the asset entities corresponding to the asset attribute code. A unique spatial relationship identifier is assigned to each spatial location relationship to determine the correspondence between the spatial reference code and the asset attribute code. Spatial location relationships include inclusion, intersection, or adjacency.

[0015] Based on the correspondence, the spatial reference code, asset attribute code and spatial relationship identifier are bound together to generate a spatiotemporal binding code.

[0016] In one embodiment, a spatial topology map and a temporal sequence map are constructed based on spatiotemporal binding coding. Planning constraint feature vectors and asset dynamic feature vectors are extracted from the spatial topology map and the temporal sequence map, respectively. The planning constraint feature vectors and asset dynamic feature vectors are then fused to generate a planning-asset coupling scoring matrix, including:

[0017] Using spatiotemporal binding codes as the first node, a spatial topology map is constructed based on the adjacency relationships of land parcels in the land spatial planning vector data. The edges of the spatial topology map indicate that the land parcels corresponding to the first node have an adjacency relationship.

[0018] The spatial topology graph is processed by graph convolutional network to extract features, and the planning constraint feature vector is obtained. The planning constraint feature vector contains land use type, plot ratio and ecological identification information.

[0019] Based on the timestamp information carried by the spatiotemporal binding code, asset data at different time points corresponding to the spatiotemporal binding code is obtained, and the asset data at different time points are sorted in chronological order to form a historical state sequence. The historical state sequence is used as the second node to construct a time sequence diagram in chronological order.

[0020] Feature extraction processing of the time series graph is performed by a gated graph neural network to obtain the asset dynamic feature vector, which contains information on land price, utilization rate and pollution index.

[0021] The planning constraint feature vector and the asset dynamic feature vector are weighted by a cross-channel attention mechanism to obtain the fusion weight.

[0022] Based on the fusion weight, the planning constraint feature vector and the asset dynamic feature vector are weighted and summed to obtain the coupling score value. The coupling score value is then arranged in the order of the corresponding spatiotemporal binding codes to obtain the planning asset coupling score matrix.

[0023] In one embodiment, a set of conflicting units is identified based on a planning asset coupling scoring matrix; each conflicting unit in the set is classified based on the set of conflicting units and a preset business rule base to obtain a conflict type; based on the conflict type, a set of business execution instructions is generated through a preset reinforcement learning decision model, including:

[0024] The coupling score value of each element in the planning asset coupling score matrix is ​​compared with the preset score threshold, and the spatial-asset unit corresponding to the element whose coupling score value is lower than the preset score threshold is identified as the conflict unit.

[0025] Each conflicting unit is combined to form a conflicting unit set;

[0026] Obtain and parse the spatiotemporal binding code corresponding to each conflict unit in the conflict unit set. Based on the association between the spatiotemporal binding code and the planning attribute and asset attribute, obtain the planning attribute and asset attribute corresponding to each conflict unit.

[0027] Based on the preset business rule base and planning attributes and asset attributes, each conflict unit is classified and processed to obtain the conflict type. The preset business rule base contains the correspondence between planning attributes, asset attributes and conflict types.

[0028] Based on the conflict type, the asset value level in the asset attribute, and the planning priority in the planning attribute, a decision state vector corresponding to each conflict unit is constructed.

[0029] The value of each candidate optimization action is obtained by performing value calculation on the decision state vector through a pre-set reinforcement learning decision model.

[0030] The action with the highest value among the candidate optimization actions is selected as the optimal optimization action. The candidate optimization actions include adjusting the planning boundary, initiating asset swap, and implementing mixed development.

[0031] Based on the optimal action corresponding to each conflict unit, a set of business execution instructions is generated.

[0032] In one embodiment, the method further includes:

[0033] The system executes a set of business execution instructions to update the planning database and asset status table, resulting in an updated planning database and an updated asset status table. The planning database is used to store vector data for land and space planning, and the asset status table is used to store data on natural resource asset registration.

[0034] Based on the planning database and the updated planning database, the asset status table and the updated asset status table, the asset value difference calculation and ecological compliance assessment are performed respectively to obtain the asset value change and ecological compliance index.

[0035] Based on changes in asset value and ecological compliance indicators, the actual reward value is obtained through calculation using a preset reward function.

[0036] Based on the actual reward value and the execution time of the business execution instruction set, the timestamp attribute of the spatiotemporal binding code corresponding to the conflict unit is updated to obtain the updated spatiotemporal binding code.

[0037] In one embodiment, the coupling score is calculated using the following formula:

[0038]

[0039] in, For the first The coupling score of each space-asset unit, with a value range of [0,1]; The contribution weight of the planning constraint features; The number of planning constraint features; The quantity of dynamic characteristics of assets; For the first The first space-asset unit Original values ​​of each planning constraint characteristic; For each space-asset unit within the preset area, the first The maximum value of each planning constraint characteristic; For each space-asset unit within the preset area, the first Minimum value of each planning constraint characteristic; For the first The first space-asset unit Original values ​​of dynamic characteristics of each asset; For each space-asset unit within the preset area, the first The maximum value of the dynamic characteristics of an asset; For each space-asset unit within the preset area, the first Minimum value of dynamic characteristics of an asset; This is the penalty coefficient for conflict risk; For the first Spatial conflict risk index for each space-asset unit.

[0040] Secondly, this application also provides an integrated system for natural resource asset management and territorial spatial planning, including:

[0041] The data encoding and spatiotemporal binding module is used to perform unified spatiotemporal encoding processing on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes;

[0042] The feature fusion and coupling scoring module is used to construct spatial topology maps and time series maps based on spatiotemporal binding coding, extract planning constraint feature vectors and asset dynamic feature vectors from the spatial topology maps and time series maps respectively, and fuse the planning constraint features and asset dynamic features to generate a planning-asset coupling scoring matrix.

[0043] The conflict identification and instruction generation module is used to identify a set of conflicting units based on the planning asset coupling scoring matrix; based on the set of conflicting units and a preset business rule base, it classifies each conflicting unit in the set of conflicting units to obtain the conflict type; based on the conflict type, it generates a set of business execution instructions through a preset reinforcement learning decision model.

[0044] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the first aspect.

[0045] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the first aspect.

[0046] The aforementioned method and system for integrating natural resource asset management and territorial spatial planning firstly generates spatiotemporal binding codes by uniformly encoding territorial spatial planning vector data and natural resource asset registration data. This resolves the information silo problem caused by the two types of data belonging to different systems and having inconsistent identification standards, improving the accuracy of data fusion and cross-scenario reusability. Secondly, based on the spatiotemporal binding codes, spatial topology maps and time-series maps are constructed to extract planning constraints and asset dynamic features and fuse them to generate a coupled scoring matrix. This addresses the problem of traditional methods having a single feature extraction dimension and failing to comprehensively reflect coupled relationships, improving the comprehensiveness of feature extraction and the depth of coupled analysis. Finally, based on the coupled scoring matrix, conflicts are identified and classified. A business execution instruction set is generated through a pre-set reinforcement learning decision model, solving the problems of low accuracy in traditional manual rule recognition and lack of intelligent optimization in decision-making, improving the efficiency of conflict identification and the intelligent adaptability of decision-making. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating a method for integrating natural resource asset management and territorial spatial planning systems, provided as an exemplary embodiment of the present invention;

[0049] Figure 2 A flowchart of a method for generating spatiotemporal binding codes is provided as an exemplary embodiment of the present invention;

[0050] Figure 3 A schematic diagram of the integrated system structure of natural resource asset management and territorial spatial planning system provided as an exemplary embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0052] In one embodiment, such as Figure 1 As shown, a method for integrating natural resource asset management and territorial spatial planning systems is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0053] S101: Perform spatiotemporal unified coding processing on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes.

[0054] Specifically, land spatial planning vector data includes geometric and planning information such as spatial location, planning boundaries, and land use type, while natural resource asset registration data focuses on asset attribute information such as resource ownership, value level, and resource type. These two types of data belong to the two major management systems of planning and assets, respectively, and have inherent defects such as inconsistent data identification standards and loose spatial and attribute correlations. Therefore, this embodiment constructs a three-in-one identification system of space, asset, and spatiotemporal dimensions to achieve the correlation between the two types of data. For example, the land spatial planning vector data is first spatially benchmarked to extract key spatial parameters such as the coordinates of the plot center point and the boundary range, ensuring the accuracy of spatial information. Simultaneously, the natural resource asset registration data undergoes attribute structure analysis, filtering core asset parameters such as resource type, ownership identifier, and value level, and eliminating redundant fields to ensure data validity. Subsequently, spatial parameters are converted into spatial dimension codes and asset parameters into asset dimension codes using preset coding rules, and a timestamp field is embedded to record the data generation time. Finally, a spatial location matching algorithm confirms the unique correspondence between the spatial dimension code and the asset dimension code, binding them to form a spatiotemporally bound code. This encoding not only carries the precise spatial location information of the land parcel but also links it to the core attribute information of the asset. At the same time, it enables data traceability through timestamps, fundamentally solving the information silo problem between the two types of data. It provides a unified data carrier for subsequent feature fusion and conflict analysis, significantly improving the efficiency of data reuse across scenarios and the accuracy of association.

[0055] S102: Construct a spatial topology map and a time series map based on spatiotemporal binding coding. Extract planning constraint feature vectors and asset dynamic feature vectors from the spatial topology map and the time series map respectively. Then, fuse the planning constraint feature vectors and asset dynamic feature vectors to generate a planning-asset coupling scoring matrix.

[0056] Specifically, the integrated analysis of planning and assets needs to consider both the static correlation of spatial constraints and the dynamic changes in asset status. For example, a spatial topology map can be constructed using spatiotemporal binding codes as unique nodes, based on the spatial relationships of adjacency, inclusion, and intersection of land parcels in the land spatial planning vector data. The presence or absence of edges in this map corresponds to the existence of spatial constraint relationships between nodes (i.e., the spatial-asset units corresponding to the spatiotemporal binding codes), which can fully characterize the spatial association rules at the planning level. Through the analysis of the spatial topology map, planning constraint feature vectors can be extracted. These feature vectors reflect the topological constraint relationships of land spatial planning, such as the spatial layout and mutual influence between planning units. Furthermore, based on the time-series characteristics of natural resource asset registration data, a time-series graph reflecting the dynamic flow of assets can be constructed by analyzing the state changes of natural resource assets at different points in time. In this graph, each node represents the asset state at a point in time, and the edges represent the transition relationships of asset states over time. Through the analysis of the time-series graph, dynamic asset feature vectors can be extracted. These feature vectors reflect the dynamic changes of natural resource assets, such as asset circulation, appreciation, and depreciation. Finally, the topological constraints of territorial spatial planning can be combined with the dynamic flow patterns of natural resource assets. This involves performing a weighted summation of the planning constraint feature vector and the asset dynamic feature vector to generate a planning-asset coupling scoring matrix that comprehensively reflects the coupling relationship between the two. Each element in this matrix corresponds to a quantitative value of the coupling fit of a spatial-asset unit, providing a quantitative basis for subsequent conflict identification and enhancing the comprehensiveness of feature expression and the depth of coupling analysis.

[0057] S103: Identify a set of conflict units based on the planning asset coupling scoring matrix; classify each conflict unit in the set of conflict units based on the set of conflict units and the preset business rule base to obtain the conflict type; generate a set of business execution instructions based on the conflict type through a preset reinforcement learning decision model.

[0058] Specifically, after generating the planning asset coupling scoring matrix, this embodiment can further utilize this matrix to identify a set of conflicting units between land spatial planning and natural resource assets. Conflicting units refer to units where there is a mismatch between land spatial planning and natural resource assets. These conflicts may manifest as inconsistencies between planning objectives and asset protection requirements, or contradictions between resource development and ecological protection. By analyzing the planning asset coupling scoring matrix, units with low coupling scores or obvious conflicts can be identified, thus forming a set of conflicting units. In the conflict classification stage, a pre-stored business rule base stores the correspondence between "planning attributes - asset attributes - conflict types," such as "industrial land planning + cultivated land asset attributes" corresponding to "conflict between planning use and asset type." Therefore, by parsing the spatiotemporal binding code corresponding to the conflicting unit, its associated planning attributes and asset attributes can be extracted and matched with the pre-stored business rule base to achieve accurate classification of conflict types, providing a basis for subsequent targeted decision-making. After determining the conflict type, a business execution instruction set can be generated through a pre-set reinforcement learning decision model. The reinforcement learning decision model is an intelligent decision-making model based on machine learning. It can generate the optimal set of business execution instructions through learning and optimization processes based on the input conflict type and relevant data. This process realizes the transformation of conflict handling from human experience-based decision-making to data-driven intelligent decision-making, which not only ensures the pertinence and scientific nature of the decision, but also improves the efficiency of transforming decision results into business execution, effectively solving the problems of inefficient resource allocation and insufficient balance of multiple objectives in traditional decision-making.

[0059] The aforementioned method first breaks down information silos between land spatial planning and natural resource asset data through unified spatiotemporal coding, achieving not only precise data correlation and cross-scenario reuse but also improving the efficiency and accuracy of data fusion. Second, it extracts planning constraints and asset dynamic features based on spatial topology maps and temporal series diagrams, generating a coupled scoring matrix that comprehensively reflects the coupling relationship between planning and assets, enhancing the comprehensiveness and accuracy of feature extraction. Finally, it generates a business execution instruction set through a reinforcement learning decision-making model, addressing the issues of low precision in conflict handling and lack of intelligent optimization support for decision-making. This enhances the scientific and intelligent level of decision-making, improves resource allocation efficiency, and achieves a dynamic balance between ecological protection and economic development.

[0060] In one embodiment, such as Figure 2 As shown, spatiotemporal unified coding processing is performed on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes, including:

[0061] S201: The GeoSOT-3D subdivision algorithm is used to perform three-dimensional mesh subdivision processing on the land spatial planning vector data to obtain the subdivided mesh;

[0062] S202: Based on the coordinates of the center points of land parcels in the subdivided grid and the vector data of the national land spatial planning, a spatial reference code is generated. The spatial reference code contains longitude grid chain, latitude grid chain, elevation identifier and timestamp information.

[0063] S203: Parse and process the natural resource asset registration data to obtain the resource type, ownership identifier, and value level; encode the resource type, ownership identifier, and value level to generate an asset attribute code, which includes administrative division code, resource classification code, ownership identifier, and value level information;

[0064] S204: Perform spatial location relationship analysis on the spatial entities corresponding to the spatial reference code and the asset entities corresponding to the asset attribute code, assign a unique spatial relationship identifier to each spatial location relationship, and determine the correspondence between the spatial reference code and the asset attribute code. Spatial location relationships include inclusion, intersection, or adjacency.

[0065] S205: Based on the correspondence, the spatial reference code, asset attribute code and spatial relationship identifier are bound together to generate a spatiotemporal binding code.

[0066] Specifically, the GeoSOT-3D subdivision algorithm can achieve seamless, multi-layered 3D subdivision of the entire Earth, adapting to planning scenarios with varying precision requirements and solving problems such as boundary fragmentation and uneven precision found in traditional subdivision algorithms. Therefore, the subdivision precision level can be determined first, and subdivision levels can be preset according to the scale of the planning area, with the level division following a progressive logic of "Earth, Hemisphere, Degree Zone, Sub-Zone, Second Zone, Sub-Second Zone". Subsequently, coordinate normalization processing can be performed on the land spatial planning vector data, converting the latitude, longitude, and elevation coordinates of each plot in the vector data into standard coordinates under the WGS84 coordinate system, eliminating coordinate deviations from different data sources. Based on the normalized 3D coordinates, the core function of the GeoSOT-3D subdivision algorithm can be called to recursively subdivide the coordinate space according to the preset levels. Each subdivision level generates a 3D grid of fixed size, and finally outputs the subdivision grid to which each plot belongs. Each subdivision grid is assigned a unique level identifier, which includes the subdivision level, longitude interval number, latitude interval number and elevation interval number, which can ensure the global uniqueness of the grid.

[0067] Furthermore, based on the coordinates of the center points of plots in the subdivided grid and land spatial planning vector data, a unique spatial identifier, or spatial reference code, can be constructed for each plot, linking spatial location and temporal information to provide a spatial reference for subsequent association with asset data. For example, the coordinates of the plot center point can be calculated first, and then the centroid of the polygon can be calculated for the boundary vector of each plot using vector data geometric calculation tools to obtain the latitude, longitude, and elevation coordinates of the center point. The calculation formula is as follows:

[0068]

[0069]

[0070] Where n represents the number of vertices of the land parcel boundary. This represents the longitude coordinates of the i-th vertex. This represents the latitude coordinate of the i-th vertex. Represents the longitude coordinates of the center point of the plot. The latitude coordinates represent the center point of the land parcel, while the elevation coordinates can be taken as the average elevation value within the land parcel area. Subsequently, the hierarchical identifiers of the subdivided grids in S201 can be extracted and decomposed into longitude and latitude grid chains. The longitude grid chain is formed by concatenating the longitude interval numbers of each subdivision level in descending order, and the latitude grid chain is similarly formed, for example, a concatenation of degree zone number - sub-zone number - second zone number. The elevation identifier can be generated by mapping the center point elevation H to the number of a preset elevation interval, with each elevation interval corresponding to a fixed elevation range and a unique number. The timestamp information can use UTC time format, obtained by extracting the final approval time of the land spatial planning vector data and converting it into a numerical code corresponding to "year, month, day, hour, minute, second". Finally, the longitude grid chain, latitude grid chain, elevation identifier, and timestamp information are concatenated in a fixed order, with each part separated by a preset separator, to generate a spatial reference code, thus ensuring that each land parcel corresponds to a unique spatial reference code.

[0071] Furthermore, core attributes can be extracted from unstructured or semi-structured asset registration data and standardized with encoding to construct unique attribute identifiers for assets, providing an attribute foundation for association with spatial reference codes. For example, database parsing tools combined with regular expression matching can be used to extract fields from natural resource asset registration data (such as XML-formatted registration files or registration record tables in relational databases), filtering out three core fields: resource type, ownership identifier, and value level, while eliminating invalid data such as duplicate registration numbers and missing field values. The resource type can be determined according to preset standards, the ownership identifier can be obtained by extracting the real estate certificate number or the unique code for natural resource asset ownership registration from the registration data, and the value level can be determined based on the assessment results in the asset valuation report, divided into five levels from level one to level five. Subsequently, the resource type, ownership identifier, and value grade can be coded. The administrative division code can use a six-digit code from the "Administrative Division Code of the People's Republic of China," directly matching the administrative division to which the asset belongs. The resource classification code can be converted from the resource type to a four-digit code according to national standards. The ownership identifier can retain the core characters of the original code and add a fixed prefix to ensure uniform code length. The value grade can be divided into five levels, corresponding to two-digit codes from 01 to 05. Finally, the administrative division code, resource classification code, ownership identifier code, and value grade code are concatenated in a fixed order, with each part separated by a preset separator, to generate an asset attribute code. Each asset corresponds to a unique asset attribute code.

[0072] Specifically, the spatial extent data of the spatial entity corresponding to the spatial reference code and the asset entity corresponding to the asset attribute code can be determined first. For example, the spatial entity can use the boundary vector data of the grid in S201, and the asset entity can use the ownership boundary vector data in the natural resource asset registration data. Both types of vector data can be uniformly converted to the WGS84 coordinate system to eliminate coordinate system differences. Then, a spatial overlay analysis algorithm can be used to determine the relationship, that is, calling the intersection analysis, containment analysis, and proximity analysis functions in the GIS spatial analysis tool to determine the spatial relationship between the two types of entities in turn. For example, when the boundary vector of the asset entity is completely located inside the boundary vector of the spatial entity, and the area overlap rate is 100%, it is determined to be a containment relationship. When the boundary vectors of the two types of entities partially overlap, and the area overlap rate is greater than 0 and less than 100%, it is determined to be an intersection relationship. When the boundary vectors of the two types of entities do not overlap, but the shortest distance is less than a preset proximity threshold (e.g., 5 meters), it is determined to be an adjacent relationship. Once the spatial relationship between the two types of entities is determined, a unique spatial relationship identifier can be assigned to each relationship. For example, containment relationship corresponds to 01, intersection relationship corresponds to 02, and adjacency relationship corresponds to 03. Simultaneously, auxiliary information such as the overlap area ratio of intersection relationships and the shortest distance of adjacency relationships can be recorded. Finally, a correspondence table between spatial reference codes and asset attribute codes is output. The table contains spatial reference codes, asset attribute codes, spatial relationship identifiers, and auxiliary information, clarifying that one spatial reference code can correspond to multiple asset attribute codes (e.g., multiple assets are contained within a single mesh), and one asset attribute code can also correspond to multiple spatial reference codes (e.g., an asset spans multiple meshes).

[0073] Furthermore, based on the correspondence table generated in S204, data can be grouped by spatial reference code. Each group contains all asset attribute codes associated with that spatial reference code and their corresponding spatial relationship identifiers. Then, the sequence "spatial reference code + spatial relationship identifier + asset attribute code" can be used, with preset non-repeating characters as separators between each part to avoid encoding ambiguity. For cases where one spatial reference code is associated with multiple asset attribute codes, an independent spatiotemporal binding code can be generated for each association combination. In addition, uniqueness verification can be performed after code generation. For example, the MD5 hash algorithm can be used to calculate the hash value of each spatiotemporal binding code and compare it with a preset code uniqueness verification library. If the hash value does not exist, the verification passes; if it exists, the separator is readjusted or a sequence number is added to ensure uniqueness. The code that passes verification is the final spatiotemporal binding code. This code simultaneously carries the spatial location information of the land parcel, the core attribute information of the asset, and the spatial relationship between the two, providing a unified data carrier for subsequent feature extraction, conflict analysis, and other steps.

[0074] In one embodiment, a spatial topology map and a temporal sequence map are constructed based on spatiotemporal binding encoding. Planning constraint feature vectors and asset dynamic feature vectors are extracted from the spatial topology map and the temporal sequence map, respectively. The planning constraint feature vectors and asset dynamic feature vectors are then fused to generate a planning-asset coupling scoring matrix, including:

[0075] Using spatiotemporal binding codes as the first node, a spatial topology map is constructed based on the adjacency relationships of land parcels in the land spatial planning vector data. The edges of the spatial topology map indicate that the land parcels corresponding to the first node have an adjacency relationship.

[0076] The spatial topology graph is processed by graph convolutional network to extract features, and the planning constraint feature vector is obtained. The planning constraint feature vector contains land use type, plot ratio and ecological identification information.

[0077] Based on the timestamp information carried by the spatiotemporal binding code, asset data at different time points corresponding to the spatiotemporal binding code is obtained, and the asset data at different time points are sorted in chronological order to form a historical state sequence. The historical state sequence is used as the second node to construct a time sequence diagram in chronological order.

[0078] Feature extraction processing of the time series graph is performed by a gated graph neural network to obtain the asset dynamic feature vector, which contains information on land price, utilization rate and pollution index.

[0079] The planning constraint feature vector and the asset dynamic feature vector are weighted by a cross-channel attention mechanism to obtain the fusion weight.

[0080] Based on the fusion weight, the planning constraint feature vector and the asset dynamic feature vector are weighted and summed to obtain the coupling score value. The coupling score value is then arranged in the order of the corresponding spatiotemporal binding codes to obtain the planning asset coupling score matrix.

[0081] Specifically, the spatiotemporal binding code serves as a unique identifier for a spatial-asset unit. Using this code as a node ensures that each node corresponds to a unique spatial-asset entity, preventing node confusion. Illustratively, boundary vector data for each plot can be extracted from the land spatial planning vector data. Then, the adjacency determination function in a GIS spatial analysis tool can be called to determine whether any two plots share a boundary segment. If so, an undirected edge can be established between the spatiotemporal binding codes (i.e., the first node) corresponding to these two plots; otherwise, no edge is established, thus constructing a spatial topology map. In this spatial topology map, nodes represent spatial-asset units, and edges represent adjacency constraints between units. This map comprehensively depicts the spatial association rules at the planning level, providing a topological foundation for subsequent extraction of planning constraint features.

[0082] Specifically, Graph Convolutional Networks (GCNs) can capture the constraints in spatial topology by aggregating features from neighboring nodes, avoiding the limitations of traditional feature extraction that only focuses on the attributes of a single unit. Therefore, the planning attributes of the spatial-asset unit corresponding to each first node can be converted into numerical features. For example, land use type can be encoded using one-hot encoding (e.g., residential land corresponds to [1,0,0], industrial land corresponds to [0,1,0]), plot ratio can be taken as the original value, and ecological identifiers can be converted into binary features according to "1 if located within the ecological red line, otherwise 0". After conversion, these numerical features can be used as the input node features of the GCN, while the adjacency matrix of the spatial topology graph represents the connection relationships between nodes. Subsequently, GCN can integrate the planning features of neighboring nodes into the features of the current node through inter-layer feature aggregation operations. For example, if the unit corresponding to a certain node is industrial land and its neighboring nodes are mostly residential land, then the land use type feature of the node after aggregation can reflect the constraint attribute of "adjacent to residential land". The final output planning constraint feature vector is a numerical vector containing land use type compatibility, plot ratio adaptability, and ecological constraint degree, where the vector dimension is consistent with the number of planning features.

[0083] Specifically, the timestamps in spatiotemporal binding codes enable temporal traceability of asset data. Therefore, constructing a time series graph based on these timestamps can capture the dynamic changes in asset states. For example, based on the timestamp field in the spatiotemporal binding code, asset registration data at different points in time corresponding to the same code can be retrieved, and three core asset parameters—land price, utilization rate, and pollution index—can be selected. These parameters can then be arranged in ascending order of timestamps, forming a sequence of historical asset states for the spatial-asset unit. Using the asset state corresponding to each point in time in this sequence as a second node, directed edges are established between nodes at adjacent time points in chronological order, with the edges pointing from earlier to later time points, thus constructing a time series graph. In this time series graph, nodes represent the asset state at a certain moment, and edges represent the temporal transition relationships of states, providing a temporal structural foundation for subsequent extraction of dynamic asset features.

[0084] In illustrative terms, the gating unit of a gated graph neural network (GGNN) can control the weighting of historical features, making it suitable for capturing dynamic patterns in time-series data. Therefore, the asset state parameters (land price, utilization rate, pollution index) corresponding to each second node can be used as input node features of the GGNN, with the temporal connection relationships between nodes represented by an adjacency matrix of the time-series graph. The GGNN can then selectively transmit historical asset features by recursively calculating the hidden state of each node. For example, if the land price of a unit shows a continuous upward trend for three consecutive time points, the gating unit can increase the weighting of this trend feature. Finally, the asset dynamic feature vector output by the GGNN can be a numerical vector containing land price fluctuation trends, utilization rate changes, and pollution index evolution patterns, where the vector dimension is consistent with the number of asset features.

[0085] Specifically, the cross-channel attention mechanism can measure the contribution of planning constraint features and asset dynamic features to the coupling analysis, avoiding the imbalance of feature importance caused by equal weighting. Therefore, the planning constraint feature vector and the asset dynamic feature vector can be input into two independent fully connected layers to convert the feature dimensions to the same dimension. Then, the two types of features after conversion can be concatenated and input into the attention calculation module, which can calculate the attention score for each feature through linear transformation and the LeakyReLU activation function. By normalizing the attention scores using softmax, the fusion weight corresponding to each feature can be obtained. For example, in ecological protection zones, the fusion weight of planning constraint features will be higher than that of asset dynamic features, thus reflecting the priority of planning constraints. Based on the obtained fusion weights, the planning constraint feature vector and the asset dynamic feature vector can be weighted and summed to obtain the coupling score. The coupling score combines planning, asset features, and conflict risk, quantifying the degree of coupling adaptation between spatial and asset units. The calculation formula is as follows:

[0086]

[0087] in, For the first The coupling score of each space-asset unit, with a value range of [0,1]; The contribution weights of planning constraint features can be preset according to the business scenario; In this embodiment, to determine the number of planning constraint features, =3, for application site type, plot ratio, and ecological labeling; In this embodiment, the number of dynamic asset characteristics is specified. =3, corresponding to land price, utilization rate, and pollution index; For the first The first space-asset unit Original values ​​of each planning constraint characteristic; For each space-asset unit within the preset area, the first The maximum value of each planning constraint characteristic; For each space-asset unit within the preset area, the first Minimum value of each planning constraint characteristic; For the first The first space-asset unit Original values ​​of dynamic characteristics of each asset; For each space-asset unit within the preset area, the first The maximum value of the dynamic characteristics of an asset; For each space-asset unit within the preset area, the first Minimum value of dynamic characteristics of an asset; The conflict risk penalty coefficient can be preset according to the risk level; For the first The spatial conflict risk index of each spatial-asset unit is calculated from the conflict status of adjacent units in the spatial topology diagram.

[0088] This is just an illustration; in actual calculations, each planning constraint feature can be normalized first (i.e., The normalized results of all planning features are summed, averaged, and then multiplied by . Simultaneously, the dynamic characteristics of each asset can be normalized (i.e., Summing, averaging, and then multiplying by (1- Add the two results together and then subtract the result. and The product of , we get the first Coupling score of each unit .

[0089] Finally, the coupling score values ​​of all space-asset units are calculated. Arranged sequentially according to the spatiotemporal binding codes, a planning asset coupling scoring matrix can be formed, where each row (or column) of the matrix corresponds to a different spatial-asset unit, and the matrix elements are the coupling score values ​​of the corresponding units. This matrix can intuitively reflect the degree of planning-asset coupling adaptation of each unit, providing a quantitative basis for subsequent conflict identification.

[0090] In one embodiment, a set of conflicting units is identified based on a planning asset coupling scoring matrix; each conflicting unit in the set is classified based on the set of conflicting units and a preset business rule base to obtain a conflict type; based on the conflict type, a set of business execution instructions is generated through a preset reinforcement learning decision model, including:

[0091] The coupling score value of each element in the planning asset coupling score matrix is ​​compared with the preset score threshold, and the spatial-asset unit corresponding to the element whose coupling score value is lower than the preset score threshold is identified as the conflict unit.

[0092] Each conflicting unit is combined to form a conflicting unit set;

[0093] Obtain and parse the spatiotemporal binding code corresponding to each conflict unit in the conflict unit set. Based on the association between the spatiotemporal binding code and the planning attribute and asset attribute, obtain the planning attribute and asset attribute corresponding to each conflict unit.

[0094] Based on the preset business rule base and planning attributes and asset attributes, each conflict unit is classified and processed to obtain the conflict type. The preset business rule base contains the correspondence between planning attributes, asset attributes and conflict types.

[0095] Based on the conflict type, the asset value level in the asset attribute, and the planning priority in the planning attribute, a decision state vector corresponding to each conflict unit is constructed.

[0096] The value of each candidate optimization action is obtained by performing value calculation on the decision state vector through a pre-set reinforcement learning decision model.

[0097] The action with the highest value among the candidate optimization actions is selected as the optimal optimization action. The candidate optimization actions include adjusting the planning boundary, initiating asset swap, and implementing mixed development.

[0098] Based on the optimal action corresponding to each conflict unit, a set of business execution instructions is generated.

[0099] Specifically, preset scoring thresholds can be determined by combining business scenario requirements with historical integration cases. For example, one can first count the lowest coupling scores of historically compliant planning-asset adaptation units within a preset area, and then adjust this value based on the area's functional positioning, such as if the threshold for an ecological protection zone is higher than that for an urban development boundary zone. Finally, the adjusted threshold can be stored in the system configuration file. Subsequently, based on this threshold, each element of the planning-asset coupling scoring matrix can be traversed, and the corresponding coupling score value can be read. Calling the numerical comparison function will Compare with the preset scoring threshold. If If the value is below this threshold, the unique identifier (i.e., spatiotemporal binding code) of the corresponding spatial-asset unit can be marked as a conflict unit, and the coupling score of the unit can be recorded for reference in subsequent classification. Combining these conflict units forms a conflict unit set. This conflict unit set can be stored in a structured list format, where each element contains core fields such as the spatiotemporal binding code and coupling score of the conflict unit. This list is stored in a temporary relational database table, with the spatiotemporal binding code as the primary key, ensuring that each conflict unit is recorded only once. By combining all the spatiotemporal binding codes marked as conflict units and their corresponding... By inserting values ​​sequentially into this temporary table, a set of conflicting units can be obtained.

[0100] Specifically, the association between spatiotemporal binding codes and planning and asset attributes is pre-stored in an association mapping table. The table fields include the spatiotemporal binding code, planning attribute ID, and asset attribute ID. Therefore, the code parsing interface can be called, inputting the spatiotemporal binding code of the conflicting unit. This interface can then separate the spatial reference code and asset attribute code according to the coding structure. Subsequently, through the association mapping table, the planning attributes (land use type, plot ratio, ecological identifier) ​​corresponding to the spatial reference code and the asset attributes (resource type, value level, ownership identifier) ​​corresponding to the asset attribute code can be retrieved. By converting these attributes into structured dictionary data, each conflicting unit corresponds to a dictionary containing both planning and asset attributes, ensuring the integrity and retrievability of the attribute information.

[0101] As an illustration, the pre-defined business rule base can be constructed using a production rule format and stored in a rule engine such as Drools. Each rule includes a condition section and a conclusion section. The condition section can be a combination of planning attributes and asset attributes, such as "planning attribute.land use type = industrial land AND asset attribute.resource type = arable land". The conclusion section can be the corresponding conflict type, such as a conflict between planned use and asset type. Therefore, by inputting the planning attribute and asset attribute dictionary of each conflict unit into the rule engine, the engine can traverse the rules in the rule base, match the condition sections, and when the conditions of a rule are fully met, it can output the conflict type corresponding to that rule and store it in the attribute dictionary of the conflict unit, ensuring that each conflict unit corresponds to a unique conflict type.

[0102] Specifically, the dimension of the decision state vector is determined by the number of input parameters. In this embodiment, the vector dimension can be 3, corresponding to conflict type, asset value level, and planning priority, respectively. Illustratively, during the construction of this vector, each parameter can first be numerically encoded. That is, the conflict type can be converted to integers according to a preset mapping table, such as "planning use conflict" corresponding to 1, "asset inefficient utilization" corresponding to 2, asset value level corresponding to 1 to 5 according to "level one to five", and planning priority corresponding to 3, 2, and 1 according to "high, medium, and low". Then, the three encoded values ​​are arranged sequentially to form the decision state vector, and each conflict unit corresponds to a unique decision state vector. Further, a preset reinforcement learning decision model can be used to perform value calculation on the constructed decision state vector to obtain the value of each candidate optimization action. The preset reinforcement learning decision model adopts a deep Q-network (DQN) architecture. The input layer dimension is consistent with the decision state vector dimension (i.e., 3), the hidden layer contains two fully connected layers, and the output layer dimension is consistent with the number of candidate optimization actions (3 in this embodiment, corresponding to three types of actions). The core formula for value calculation can be:

[0103]

[0104] in, Let be the decision state vector. Optimize actions for candidates, = For model parameters, , It is a fully connected layer. This is the activation function. In practice, the decision state vector can be input into the DQN model, and the model can output the Q-value (i.e., action value) corresponding to each candidate action, which is stored in an action value list, with the list index corresponding one-to-one with the candidate action.

[0105] Subsequently, by traversing the action value list and calling the maximum value lookup function, the index corresponding to the element with the largest Q value in the list can be determined. Then, through a preset index-action mapping table (e.g., index 0 corresponds to adjusting the planning boundary, index 1 corresponds to initiating asset replacement), the index is mapped to the corresponding candidate optimization action, which is then determined as the optimal optimization action for the current conflict unit and stored in the attribute dictionary of the conflict unit. Based on the optimal optimization action for each conflict unit, a corresponding instruction can be generated according to the type of the optimal optimization action. For example, if the optimal optimization action is adjusting the planning boundary, an SQL update instruction is generated; if the action is initiating asset replacement, an SQL insert instruction is generated (e.g., "INSERT INTO Asset Replacement Table (Spatiotemporal Binding Code, Replacement Type, Target Asset Code) VALUES ([Conflict Unit Code], 'Farmland Replacement', [Target Asset Code])"). By classifying the instructions corresponding to all conflict units according to their operation types, a business execution instruction set can be formed and stored as an instruction file that can be directly executed by the planning management system or asset registration system, thus realizing the direct conversion of decision results into business operations.

[0106] In one embodiment, the method further includes:

[0107] The system executes a set of business execution instructions to update the planning database and asset status table, resulting in an updated planning database and an updated asset status table. The planning database is used to store vector data for land and space planning, and the asset status table is used to store data on natural resource asset registration.

[0108] Based on the planning database and the updated planning database, the asset status table and the updated asset status table, the asset value difference calculation and ecological compliance assessment are performed respectively to obtain the asset value change and ecological compliance index.

[0109] Based on changes in asset value and ecological compliance indicators, the actual reward value is obtained through calculation using a preset reward function.

[0110] Based on the actual reward value and the execution time of the business execution instruction set, the timestamp attribute of the spatiotemporal binding code corresponding to the conflict unit is updated to obtain the updated spatiotemporal binding code.

[0111] Specifically, the business execution instruction set can adopt a combination of structured SQL instructions and spatial data operation instructions. To avoid data inconsistency during execution, a transaction mechanism can be used to encapsulate the instruction execution process. All instructions are treated as an atomic transaction, committed only when all instructions are successfully executed. If any instruction fails, a rollback is triggered, ensuring consistency between the planning database and the asset status table. For example, the instruction set can be loaded by calling the database execution interface. The planning database, using a PostGIS database that supports spatial data storage, executes spatial data update instructions. For instance, when adjusting the vector coordinates of planning boundaries, the ST_Update function provided by PostGIS can accurately replace the boundary vectors. The asset status table, using a MySQL database, executes asset attribute update instructions. For example, when asset ownership changes or value levels are adjusted, the corresponding field values ​​can be modified using UPDATE statements. After execution, the system automatically verifies the update results. The planning database can compare the hash values ​​of the vector data before and after the update using the ST_Equals function, and the asset status table can compare field value snapshots. Once verification is successful, the updated planning database and asset status table are output, both associated with the corresponding execution batch number, providing a basis for the correlation between old and new data in subsequent indicator calculations.

[0112] Based on the planning database and the updated planning database, the asset status table and the updated asset status table, the asset value difference calculation and ecological compliance assessment are performed respectively to obtain the asset value change and ecological compliance index.

[0113] Specifically, you can first associate the asset status table before the update with the batch number (denoted as...). ) and the updated asset status table (denoted as The system matches new and old asset data within the same spatial-asset unit by spatiotemporal binding coding. Furthermore, it can invoke preset valuation models for different asset types; for example, land assets use the benchmark land price adjustment method, with the calculation formula as follows:

[0114]

[0115] Where V is the assessed value, S is the area, and P is the benchmark land price. This is the floor area ratio correction factor. This is the term adjustment factor. Mineral assets can be calculated using the income approach, with the following formula: ,in Let r be the expected return in year t, r be the discount rate, and n be the number of years the return is received. Calculate the value before the update. With the updated value Subsequently, the change in asset value = - , A positive value indicates increased value, while a negative value indicates decreased value. The assessment of ecological compliance indicators can be based on the ecological constraint rules of the planning database, that is, by linking the batch number to the planning database before the update (denoted as...). ) and the updated planning database (denoted as The ecological attributes of the updated units are extracted (such as whether they are located within the ecological red line, whether the plot ratio exceeds the standard, and whether the pollution index meets the standard). Each attribute is scored from 0 to 1 (1 for meeting the standard and 0 for not meeting the standard) according to the threshold range in the regional ecological protection planning standards. Then, the ecological compliance index E is obtained by weighting the data according to preset weights, such as 0.4 for ecological red line, 0.3 for plot ratio, and 0.3 for pollution index. The closer the index is to 1, the stronger the compliance.

[0116] Specifically, based on changes in asset value and ecosystem compliance indicators, the actual reward value can be calculated using a pre-defined reward function. This pre-defined reward function can employ a multi-objective optimization logic design to balance asset value enhancement and ecosystem compliance, avoiding decision-making biases caused by a single-objective approach. The core formula of the reward function can be...

[0117]

[0118] Where R is the actual reward value, The asset value weight can be preset according to the regional development positioning, such as economic development zones. =0.6, ecological protection zone =0.4, For ecological compliance weight (and) The sum is 1). The ecological violation penalty coefficient (default value is 1.5). The normalized result of the change in asset value (using min-max normalization, formula is...) , The maximum historical value change within the preset area. (the smallest change in history) For ecological violations, penalties apply when E ≥ 0.8. =0, when E < 0.8 =0.8-E.

[0119] As an example, you can first call the data interface to obtain... The calculation results of E are substituted into the above reward function formula, and the actual reward value R is calculated by the floating-point operation module. R∈[-0.5,1]. This value can directly reflect the comprehensive effect of the business execution instruction set and provide a basis for validity judgment for subsequent timestamp updates.

[0120] Subsequently, based on the actual reward value and the execution time of the business execution instruction set, the timestamp attribute of the spatiotemporal binding code corresponding to the conflict unit can be updated. This injects the latest execution node's time information into the spatiotemporal binding code, ensuring the timeliness of data traceability and allowing the filtering of valid execution results through the actual reward value, avoiding redundant updates caused by invalid execution. For example, the execution completion time of the business execution instruction set can be extracted first and converted into the UTC standard time format as the base time data for the timestamp. Then, it can be determined whether the actual reward value R meets the preset valid threshold, which can be determined by statistical analysis of the reward values ​​of historical valid execution cases. If R does not reach the threshold, it means that the execution effect has not met expectations, and the timestamp update is not performed. If R reaches the threshold, the code update interface can be called, inputting the original spatiotemporal binding code of the conflict unit. The interface splits the original timestamp field according to the code structure and replaces it with the new UTC execution time, while other fields (longitude grid chain, asset attribute code, etc.) remain unchanged. After the update is completed, the uniqueness of the new code can be verified. For example, the SHA-256 hash algorithm is used to calculate the code hash value and compare it with the code library. If the comparison and verification pass, the updated spatiotemporal binding code can be obtained, stored in the code management library, and associated with the original code to form a version traceability relationship. This can provide timely data support for subsequent feature extraction and conflict analysis.

[0121] Based on the same inventive concept, this application also provides a system for integrating natural resource asset management and territorial spatial planning systems to implement the aforementioned method for integrating natural resource asset management and territorial spatial planning systems. The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the system for integrating natural resource asset management and territorial spatial planning systems provided below can be found in the limitations of the method for integrating natural resource asset management and territorial spatial planning systems described above, and will not be repeated here.

[0122] In one exemplary embodiment, such as Figure 3 As shown, a system 300 integrating natural resource asset management and territorial spatial planning is provided, including:

[0123] The data encoding and spatiotemporal binding module 301 is used to perform unified spatiotemporal encoding processing on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes;

[0124] The feature fusion and coupling scoring module 302 is used to construct a spatial topology map and a time series map based on spatiotemporal binding coding, extract planning constraint feature vectors and asset dynamic feature vectors from the spatial topology map and the time series map respectively, and fuse the planning constraint features and asset dynamic features to generate a planning asset coupling scoring matrix.

[0125] The conflict identification and instruction generation module 303 is used to identify a set of conflict units based on the planning asset coupling scoring matrix; classify each conflict unit in the set of conflict units based on the set of conflict units and the preset business rule base to obtain the conflict type; and generate a set of business execution instructions based on the conflict type through a preset reinforcement learning decision model.

[0126] In one exemplary embodiment, the present invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method for integrating natural resource asset management and territorial spatial planning systems as described in this application. A multi-core processor is preferred to improve the system's parallel processing capability. The memory provides sufficient temporary storage space to support program execution and data processing. The memory capacity should be large enough to accommodate large amounts of data and computational tasks.

[0127] In one exemplary embodiment, the present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method for integrating natural resource asset management and territorial spatial planning systems as described in this application. The computer-readable storage medium may include: a read-only memory, a random access memory (RAM), a solid-state drive (SSD), or an optical disc, etc.

[0128] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for integrating natural resource asset management with land spatial planning, characterized in that, The method includes: Spatiotemporal unified coding processing is performed on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes; Based on the spatiotemporal binding encoding, a spatial topology map and a temporal sequence map are constructed. Planning constraint feature vectors and asset dynamic feature vectors are extracted from the spatial topology map and the temporal sequence map, respectively. The planning constraint feature vectors and the asset dynamic feature vectors are then fused to generate a planning-asset coupling scoring matrix. Based on the planned asset coupling scoring matrix, a set of conflict units is identified; based on the set of conflict units and a preset business rule base, each conflict unit in the set of conflict units is classified to obtain a conflict type; based on the conflict type, a set of business execution instructions is generated through a preset reinforcement learning decision model.

2. The method according to claim 1, characterized in that, The process of performing spatiotemporal unified coding on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes includes: The geospatial planning vector data is subjected to three-dimensional meshing using the GeoSOT-3D meshing algorithm to obtain the meshed grid. Based on the subdivided grid and the coordinates of the center points of the land parcels in the land spatial planning vector data, a spatial reference code is generated. The spatial reference code includes longitude grid chains, latitude grid chains, elevation identifiers, and timestamp information. The natural resource asset registration data is parsed and processed to obtain resource type, ownership identifier, and value level; The resource type, ownership identifier, and value level are encoded to generate an asset attribute code, which includes administrative division code, resource classification code, ownership identifier, and value level information. Spatial position relationship analysis is performed on the spatial entity corresponding to the spatial reference code and the asset entity corresponding to the asset attribute code. A unique spatial relationship identifier is assigned to each spatial position relationship to determine the correspondence between the spatial reference code and the asset attribute code. The spatial position relationship includes inclusion, intersection, or adjacency. Based on the correspondence, the spatial reference code, the asset attribute code, and the spatial relationship identifier are bound together to generate the spatiotemporal binding code.

3. The method according to claim 1, characterized in that, The process involves constructing a spatial topology map and a time-series map based on the spatiotemporal binding encoding, extracting planning constraint feature vectors and asset dynamic feature vectors from the spatial topology map and the time-series map respectively, and fusing the planning constraint feature vectors and the asset dynamic feature vectors to generate a planning-asset coupling scoring matrix, including: Using the spatiotemporal binding code as the first node, the spatial topology map is constructed based on the adjacency relationship of land parcels in the land spatial planning vector data. The edges of the spatial topology map indicate that the land parcels corresponding to the first node have an adjacency relationship. The spatial topology graph is processed by a graph convolutional network to extract features, and the planning constraint feature vector is obtained. The planning constraint feature vector includes land use type, plot ratio and ecological identification information. Based on the timestamp information carried by the spatiotemporal binding code, asset data at different time points corresponding to the spatiotemporal binding code are obtained, and the asset data at different time points are sorted in chronological order to form a historical state sequence. The historical state sequence is used as the second node to construct the time sequence diagram in chronological order. The time series graph is processed by a gated graph neural network to extract features, and the asset dynamic feature vector is obtained. The asset dynamic feature vector includes land price, utilization rate and pollution index information. The planning constraint feature vector and the asset dynamic feature vector are weighted using a cross-channel attention mechanism to obtain a fusion weight. Based on the fusion weights, the planning constraint feature vector and the asset dynamic feature vector are weighted and summed to obtain a coupling score value. The coupling score values ​​are then arranged in the order of the corresponding spatiotemporal binding codes to obtain the planning asset coupling score matrix.

4. The method according to claim 1, characterized in that, The conflict unit set is identified based on the planning asset coupling scoring matrix; based on the conflict unit set and the preset business rule base, each conflict unit in the conflict unit set is classified to obtain the conflict type; Based on the aforementioned conflict type, a set of business execution instructions is generated using a pre-defined reinforcement learning decision model, including: The coupling score value of each element in the planned asset coupling score matrix is ​​compared with a preset score threshold, and the spatial-asset unit corresponding to the element whose coupling score value is lower than the preset score threshold is determined as a conflict unit. The conflicting units are combined to form the conflicting unit set; Obtain and parse the spatiotemporal binding code corresponding to each conflict unit in the conflict unit set. Based on the association between the spatiotemporal binding code and the planning attribute and asset attribute, obtain the planning attribute and asset attribute corresponding to each conflict unit. Based on the preset business rule base and the planning attributes and asset attributes, each conflict unit is classified to obtain the conflict type. The preset business rule base contains the correspondence between the planning attributes, asset attributes and the conflict type. Based on the conflict type, the asset value level in the asset attribute, and the planning priority in the planning attribute, a decision state vector is constructed for each conflict unit. The value of each candidate optimization action is obtained by performing value calculation on the decision state vector through the preset reinforcement learning decision model. The action with the highest value is selected from the candidate optimization actions as the optimal optimization action. The candidate optimization actions include adjusting the planning boundary, initiating asset swap, and implementing hybrid development. The service execution instruction set is generated based on the optimal optimization action corresponding to each conflict unit.

5. The method according to claim 1, characterized in that, The method further includes: The business execution instruction set is executed to update the planning database and asset status table, resulting in an updated planning database and an updated asset status table. The planning database is used to store the land spatial planning vector data, and the asset status table is used to store the natural resource asset registration data. Based on the planning database and the updated planning database, the asset status table and the updated asset status table, the asset value difference calculation and ecological compliance assessment are performed respectively to obtain the asset value change and the ecological compliance index. Based on the change in asset value and the ecological compliance index, the actual reward value is obtained by calculation using a preset reward function. Based on the actual reward value and the execution time of the business execution instruction set, the timestamp attribute of the spatiotemporal binding code corresponding to the conflict unit is updated to obtain the updated spatiotemporal binding code.

6. The method according to claim 3, characterized in that, The coupling score is calculated using the following formula: in, For the first The coupling score value of each space-asset unit ranges from [0,1]. The contribution weight of the planning constraint features; The number of planning constraint features; The quantity of dynamic characteristics of assets; For the first The first space-asset unit Original values ​​of each planning constraint characteristic; For each of the space-asset units within the preset area, the first The maximum value of each planning constraint characteristic; For each of the space-asset units within the preset area, the first Minimum value of each planning constraint characteristic; For the first The first space-asset unit Original values ​​of dynamic characteristics of each asset; For each of the space-asset units within the preset area, the first The maximum value of the dynamic characteristics of an asset; For each of the space-asset units within the preset area, the first Minimum value of dynamic characteristics of an asset; This is the penalty coefficient for conflict risk; For the first Spatial conflict risk index for each space-asset unit.

7. A system integrating natural resource asset management and territorial spatial planning, characterized in that, The system includes: The data encoding and spatiotemporal binding module is used to perform unified spatiotemporal encoding processing on land spatial planning vector data and natural resource asset registration data to generate unified spatiotemporal binding codes; The feature fusion and coupling scoring module is used to construct a spatial topology map and a temporal sequence map based on the spatiotemporal binding encoding, extract planning constraint feature vectors and asset dynamic feature vectors from the spatial topology map and the temporal sequence map respectively, and perform fusion processing on the planning constraint features and the asset dynamic features to generate a planning asset coupling scoring matrix. The conflict identification and instruction generation module is used to identify a set of conflict units based on the planning asset coupling scoring matrix; classify each conflict unit in the set of conflict units based on the set of conflict units and a preset business rule base to obtain the conflict type; and generate a set of business execution instructions based on the conflict type through a preset reinforcement learning decision model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images