A land spatial planning management system and method based on big data

Abstract

The present invention discloses a land spatial planning management system and method based on big data, which relates to the technical field of spatial planning management. The spatio-temporal coding mechanism adopted by the present invention deeply integrates the historical evolution of ecological elements with spatial topology, realizes the real-time deduction of the multi-dimensional impacts of red line adjustment on the groundwater level, species migration, etc., and breaks through the lag of the traditional static rule base; quantifies the propagation path of cascade effects based on the graph diffusion model, and automatically identifies cross-media contradictions such as habitat fragmentation caused by the increase in land development intensity, solving the problem that it is difficult for artificial experience to handle complex causal chains.

Description

Technical Field

[0001] The present invention relates to the technical field of spatial planning management, and particularly to a national land spatial planning management system and method based on big data. Background Art

[0002] The national land spatial planning management system relies on a static spatial database constructed by the Geographic Information System (GIS), and mainly determines the scope of the ecological protection red line in an artificial delineation manner; this system updates the land use nature conflict through a rule engine based on periodic census data, and relies on expert experience meetings to decide on adjustment plans; it works okay during the relatively stable period of the ecological pattern; however, it performs poorly in the face of dynamic interference factors such as climate change and increasing human activities.

[0003] The ecological element correlation analysis of the existing system is limited to the two-dimensional plane, and a three-dimensional impact conduction model for the above-ground and underground, water area and land area is not constructed, resulting in the difficulty of predicting the cross-media chain reaction caused by the adjustment of the red line; at the same time, conflict detection depends on a preset rule library, and it is impossible to identify hidden causal chains across spatio-temporal scales such as the shrinkage of the ecological red line to the increase in the development intensity of surrounding land and the fragmentation of the habitats of migratory species; in addition, the response mechanism is fragmented, and the data standards and update cycles of departments such as land, environmental protection, and water conservancy are not unified, resulting in the lag of the domino effect warning behind the occurrence of actual damage; therefore, there is an urgent need for a national land spatial planning management solution based on big data to solve such problems. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] The present invention provides a national land spatial planning management system and method based on big data to solve the problems that traditional national land planning depends on static GIS data and artificial rule libraries, lacks dynamic association modeling capabilities, is difficult to warn of cross-media chain reactions caused by ecological adjustments, and leads to missed detection of hidden conflicts and lagged responses.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a national land spatial planning management system based on big data, which includes,

[0008] A data acquisition layer, configured to obtain remote sensing image data, Internet of Things sensor monitoring data, and multi-department planning document data in real time;

[0009] A dynamic knowledge graph construction module, connected to the data acquisition layer, which converts land attribute, ecological element, and infrastructure data into an entity-relationship network including spatio-temporal attributes;

[0010] A conflict detection engine, based on the dynamic knowledge graph, uses a graph neural network to mine implicit conflicts between cross-level planning elements;

[0011] A visualization decision-making platform for three-dimensional space deduction and visual feedback of dynamic adjustment plans.

[0012] As a preferred solution of the land space planning management system based on big data according to the present invention, wherein: the dynamic knowledge graph construction module includes:

[0013] An entity recognition unit, which uses a pre-trained BERT model to extract ecological reserve boundaries and land use property change record entities from unstructured planning documents; the BERT model is trained by transfer learning using texts in the field of land space planning, and the training corpus includes historical approval documents, ecological assessment reports, and policy and regulation texts;

[0014] A relationship extraction unit, which analyzes the spatial topological relationship and policy constraint relationship between entities through a graph convolutional network GCN;

[0015] A spatio-temporal encoding unit, which attaches a time stamp attribute and a spatial coordinate attribute to each entity node.

[0016] As a preferred solution of the land space planning management system based on big data according to the present invention, wherein: in the dynamic knowledge graph construction module, the stage of converting land attributes, ecological elements, and infrastructure data into an entity-relationship network containing spatio-temporal attributes includes:

[0017] A multi-source feature fusion stage, defining the fusion formula as:

[0018] , where represents the fusion feature vector at time , represents the time index, represents the remote sensing image feature fusion weight, represents the remote sensing image spectral-texture matrix at time , represents the sensor data feature fusion weight, represents the Internet of Things sensor observation vector at time , represents the planning document semantic feature fusion weight, represents the planning document semantic embedding vector at time ;

[0019] An entity generation mapping stage, defining the node set formula as: , where represents the entity node set at time , Denote the mapped entity node as the feature - type mapping function, denote the land category label of the node denote the spatial coordinate vector of the node as the entity index;

[0020] In the edge weight calculation stage, define the comprehensive weight and spatial distance formula as:

[0021] , ,

[0022] where denote the and comprehensive weight between nodes, denote the spatial distance attenuation coefficient, denote the and Euclidean distance between nodes, respectively denote the spatial coordinate components of the node respectively denote the spatial coordinate components of the node denote the semantic difference attenuation coefficient, denote the and semantic Hamming distance;

[0023] In the relationship set screening stage, define the edge set formula as: , where denote the edge set at time denote the weight threshold;

[0024] In the spatio - temporal embedding generation stage, define the node embedding and graph structure:

[0025] , , ,

[0026] where denote the spatio - temporal embedding vector of the node denote the node attribute vector, denote the node timestamp, denote the node spatial coordinate vector, denote the dynamic knowledge graph at time denote the embedding matrix composed of all node embedding vectors arranged diagonally, Represents the number of nodes.

[0027] As a preferred solution of the land spatial planning and management system based on big data according to the present invention, wherein: the conflict detection engine includes:

[0028] The implicit conflict identification sub-module calculates the influence weight of the ecological red line adjustment on the associated plots based on the graph attention mechanism GAT;

[0029] The domino effect prediction sub-module simulates the cascade propagation path of ecological element changes through the temporal graph diffusion model;

[0030] The conflict quantification and evaluation sub-module outputs the ecological risk index of the affected area and the predicted value of economic loss.

[0031] As a preferred solution of the land spatial planning and management system based on big data according to the present invention, wherein: in the domino effect prediction sub-module, the cascade influence propagation rule includes:

[0032] The iterative model based on node influence intensity combines spatio-temporal attenuation and relationship weight, defined as:

[0033] ,

[0034] Wherein, Represents the node At the th iteration, the influence intensity, Represents the node self-attenuation coefficient, Represents the set of nodes adjacent to , Represents the time decay rate, Represents the node To The time difference, Represents the edge The relationship weight, Represents the node index located in the neighbor set , and iterates until Or , Represents the maximum number of iteration rounds, Indicates that when the norm of the difference in influence intensity between two iterations is less than this threshold, the iteration stops;

[0035] This rule reflects the suppression effect of propagation speed and time interval through the exponential kernel at each moment, and combines the edge weight to reflect the strength of cross-element influence, and can capture the implicit impact in the form of dominoes.

[0036] As a preferred solution of the land spatial planning and management system based on big data according to the present invention, wherein: the visualization decision-making platform includes:

[0037] A three-dimensional sand table modeling unit that maps the dynamic knowledge graph into an interactive three-dimensional space model;

[0038] A real-time deduction unit that responds to the planning parameter adjustment instruction and synchronously updates the ecological carrying capacity heat map of the associated area;

[0039] A pre-plan generation unit that automatically outputs a set of adjustment plans that meet multi-objective constraints based on the conflict detection results.

[0040] As a preferred solution of the land spatial planning and management system based on big data according to the present invention, wherein: in the real-time deduction unit, the synchronous update method of the ecological carrying capacity heat map is:

[0041] Map the propagated node influence field to a continuous space grid and update the carrying capacity field at the th moment according to the following formula:

[0042] ,

[0043] wherein, represents the ecological carrying capacity of the th grid cell at the moment , respectively represent the row index and column index of the two-dimensional grid, represents the influence conversion coefficient, represents the grid 's neighborhood set, represents the spatial convolution kernel weight, which is related to distance or geographical elements, represents the value of the calculated node influence intensity projected onto the grid . After the update is completed, is normalized and mapped to a color scale to generate a heat map.

[0044] In a second aspect, the present invention provides a land spatial planning and management method based on big data, including,

[0045] Step S1, obtaining multi-source heterogeneous data in real time, including satellite remote sensing image streams, groundwater level sensor data, and planning approval electronic files;

[0046] Step S2, constructing a spatio-temporal fusion dynamic knowledge graph, and encoding the ecological red line range, land use approval records, and infrastructure layout data into a weighted graph structure;

[0047] Step S3, analyzing the cascading effects of ecological element changes through a graph neural network to identify hidden conflicts across media and spatio-temporal scales;

[0048] Step S4: Dynamically display the multi-dimensional impacts of the planning adjustment plan in the 3D visualization interface and generate adaptive optimization suggestions.

[0049] As a preferred solution of the method for managing territorial spatial planning based on big data according to the present invention, wherein: Step S3 specifically includes:

[0050] Use a temporal graph convolutional network to model the impact propagation of the changes in the ecological red line boundary on the surrounding plots;

[0051] Quantify the attenuation ability of different land types to ecological impacts through node embedding technology;

[0052] Generate a warning level map of the affected area and a recommended buffer zone setting plan.

[0053] As a preferred solution of the method for managing territorial spatial planning based on big data according to the present invention, wherein: The generation of the adaptive optimization suggestions in Step S4 includes:

[0054] Establish a multi-objective optimization model, and the constraint conditions include ecological protection thresholds, land use efficiency, and infrastructure carrying capacity;

[0055] Use the non-dominated sorting genetic algorithm NSGA-II to solve the Pareto optimal solution set;

[0056] Based on topological data analysis, screen the adjustment plan with the optimal spatial continuity.

[0057] The beneficial effects of the present invention are as follows: The spatio-temporal coding mechanism adopted by the present invention deeply integrates the historical evolution of ecological elements with spatial topology, realizes real-time deduction of multi-dimensional impacts such as red line adjustment on groundwater level, species migration, etc., and breaks through the lag of traditional static rule bases; based on the graph diffusion model, quantify the propagation path of cascade effects, automatically identify cross-media contradictions such as habitat fragmentation caused by the increase in land development intensity, and solve the pain points of complex causal chains that are difficult to handle by manual experience; the 3D deduction platform couples multi-objective optimization algorithms, optimizes spatial continuity while ensuring ecological thresholds, and avoids the problem of spatial layout fragmentation caused by traditional fragmented approvals. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings;

[0059] Figure 1Schematic diagram of the framework of the big data-based national territorial space planning management system in Embodiment 1;

[0060] Figure 2 Schematic diagram of the process of the big data-based national territorial space planning management method in Embodiment 1. Specific implementation manners

[0061] To make the above objects, features and advantages of the present invention more obvious and understandable, the specific implementation manners of the present invention will be described in detail below with reference to the accompanying drawings of the specification.

[0062] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0063] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other from other embodiments.

[0064] Embodiment 1, referring to Figure 1 and Figure 2 , this embodiment provides a big data-based national territorial space planning management system, including:

[0065] A data acquisition layer, configured to acquire remote sensing image data, Internet of Things sensor monitoring data, and multi-department planning document data in real time;

[0066] A dynamic knowledge graph construction module, connected to the data acquisition layer, and converting land attributes, ecological elements, and infrastructure data into an entity-relationship network including spatio-temporal attributes;

[0067] The dynamic knowledge graph construction module includes:

[0068] An entity recognition unit, which extracts ecological protection area boundaries and land use property change record entities from unstructured planning documents by using a pre-trained BERT model; the BERT model is trained by transfer learning using texts in the field of national territorial space planning, and the training corpus includes historical approval documents, ecological assessment reports, and policy and regulation texts;

[0069] A relationship extraction unit, which analyzes the spatial topological relationship and policy constraint relationship between entities through a graph convolutional network GCN;

[0070] A spatio-temporal encoding unit, which attaches a time stamp attribute and a spatial coordinate attribute to each entity node;

[0071] In the dynamic knowledge graph construction module, the stage of transforming land attributes, ecological elements, and infrastructure data into an entity-relationship network with spatio-temporal attributes includes:

[0072] The multi-source feature fusion stage, where the fusion formula is defined as:

[0073] , where represents the fusion feature vector at time , represents the time index, represents the remote sensing image feature fusion weight, represents the remote sensing image spectral-texture matrix at time , represents the sensor data feature fusion weight, represents the Internet of Things sensor observation vector at time , represents the document semantic feature fusion weight, represents the semantic embedding vector of the planning document at time ;

[0074] The entity generation mapping stage, where the node set formula is defined as: , where represents the entity node set at time , represents the th mapped entity node, represents the feature-type mapping function, represents the land category label of node , represents the spatial coordinate vector of node , represents the entity index;

[0075] The edge weight calculation stage, where the comprehensive weight and spatial distance formula are defined as:

[0076] , ,

[0077] where represents the comprehensive weight between node and , represents the spatial distance attenuation coefficient, represents the Euclidean distance between node and node , respectively represent the spatial coordinate components of node , respectively represent the spatial coordinate components of node Spatial coordinate components Represents the semantic difference attenuation coefficient Represents a node and Semantic Hamming distance

[0078] In the relationship set screening stage, the edge set formula is defined as: where Represents the moment Edge set Represents the weight threshold

[0079] In the spatio-temporal embedding generation stage, the node embedding and graph structure are defined:

[0080] , , ,

[0081] where Represents the spatio-temporal embedding vector of node Represents the node attribute vector Represents the node timestamp Represents the node spatial coordinate vector Represents the moment Dynamic knowledge graph Represents the embedding matrix composed of all node embedding vectors arranged diagonally Represents the number of nodes

[0082] Conflict detection engine, based on the dynamic knowledge graph, uses graph neural networks to mine implicit conflicts between cross-level planning elements

[0083] The conflict detection engine includes:

[0084] Implicit conflict recognition sub-module, based on the graph attention mechanism GAT, calculates the influence weight of the ecological red line adjustment on related plots

[0085] Domino effect prediction sub-module, simulates the cascading propagation path of ecological element changes through the temporal graph diffusion model

[0086] Conflict quantification and evaluation sub-module, outputs the ecological risk index of the affected area and the predicted value of economic loss

[0087] In the domino effect prediction sub-module, the cascading influence propagation rules include:

[0088] Iterative model based on node influence strength, combines spatio-temporal attenuation and relationship weight, and is defined as:

[0089] ,

[0090] ​Among them, represents the influence intensity of the node at the th iteration, represents the self-decay coefficient of the node, represents the set of adjacent nodes, represents the time decay rate, represents the node to time difference, represents the edge relationship weight, represents being located in the neighbor set node index, iterate until or , represents the maximum number of iteration rounds, represents stopping iteration when the norm of the difference in influence intensity between two iterations is less than this threshold;

[0091] This rule reflects the inhibitory effect of propagation speed and time interval through the exponential kernel at each moment, combined with the edge weight to reflect the strength of cross-factor influence and can capture domino-like implicit impacts;

[0092] Visual decision-making platform for three-dimensional space deduction and visual feedback of dynamic adjustment schemes;

[0093] The visual decision-making platform includes:

[0094] Three-dimensional sand table modeling unit that maps the dynamic knowledge graph into an interactive three-dimensional space model;

[0095] Real-time deduction unit that responds to the planning parameter adjustment instruction and synchronously updates the ecological carrying capacity heat map of the associated area;

[0096] Plan generation unit that automatically outputs a set of adjustment schemes that meet multi-objective constraints based on the conflict detection results;

[0097] In the real-time deduction unit, the synchronous update method of the ecological carrying capacity heat map is:

[0098] Map the propagated node influence field to the continuous space grid and update the carrying capacity field at the th moment according to the following formula:

[0099] ,

[0100] Among them, represents the ecological carrying capacity of the th grid cell at the moment , respectively represent the row index and column index of the two-dimensional grid, represent the influence conversion coefficient, represent the grid of the neighborhood set, represent the spatial convolution kernel weights, related to distance or geographical features, represent the projection of the calculated node influence intensity onto the grid value. After the update is completed, will be normalized and mapped to a color scale to generate a heat map.

[0101] Example 2. This example provides a method for managing territorial spatial planning based on big data, including:

[0102] Step S1, obtaining multi-source heterogeneous data in real time, including satellite remote sensing image streams, groundwater level sensor data, and planning approval electronic files;

[0103] Step S2, constructing a spatio-temporal fusion dynamic knowledge graph, encoding the ecological red line scope, land use approval records, and infrastructure layout data into a weighted graph structure;

[0104] Step S3, analyzing the cascading effects of ecological element changes through a graph neural network, and identifying implicit conflicts across media and spatio-temporal scales;

[0105] Step S3 specifically includes:

[0106] Using a temporal graph convolutional network to model the impact propagation of ecological red line boundary changes on surrounding plots;

[0107] Quantifying the attenuation ability of different land types to ecological impacts through node embedding technology;

[0108] Generating a warning level map of the affected area and a recommended buffer zone setting plan;

[0109] Step S4, dynamically displaying the multi-dimensional impacts of the planning adjustment plan in a three-dimensional visualization interface, and generating adaptive optimization suggestions;

[0110] The generation of adaptive optimization suggestions in Step S4 includes:

[0111] Establishing a multi-objective optimization model, with constraint conditions including ecological protection thresholds, land use efficiency, and infrastructure carrying capacity;

[0112] Using the non-dominated sorting genetic algorithm NSGA-II to solve the Pareto optimal solution set;

[0113] Based on topological data analysis, screening the adjustment plan with the optimal spatial continuity;

[0114] In Step S4, the steps for solving the Pareto optimal set include:

[0115] Construct a decision vector , and objective functions:

[0116] ,

[0117] Among them, represents the vector of planning adjustment variables, represents the decision space, represents the variable dimension, represents the th objective function, represents the set of real numbers;

[0118] Define the dominance relationship: dominates if and only if

[0119] ,

[0120] Among them, means for all, means there exists;

[0121] Calculate the forward dominance set and the dominated count for each individual in the population and the dominated count :

[0122] , ,

[0123] Among them, represents the set cardinality, are all individual indices in the population, where is the currently examined individual, is the index of other individuals compared with it;

[0124] For all assign the rank front 1, and iteratively eliminate the sorted individuals to generate the rank set ;

[0125] Sort the individuals within each rank according to the crowding distance, and the crowding distance is calculated as:

[0126] ,

[0127] Among them, represents the function value of adjacent individuals on the th objective, the objective The maximum and minimum values; select the elite population according to the priority of levels and the order of crowding degree, generate a new generation through crossover and mutation, and repeat the iteration until the maximum number of generations is reached or convergence occurs;

[0128] In step S4, the steps of screening the adjustment plan with the optimal spatial continuity include:

[0129] For the Pareto solution set obtained by NSGA-II , for each solution Construct its spatial graph

[0130] , where is the plot node, is the adjacency relationship;

[0131] Define the spatial continuity index:

[0132] ,

[0133] where, represents the set of connected subgraphs of graph , represents the number of subgraph nodes, represents the total number of nodes; select the solution set corresponding to the maximum in as the set of optimal plans for spatial continuity;

[0134] Specifically, in this step, NSGA-II is used to simultaneously optimize multiple objectives such as ecological carrying capacity, land use efficiency, and infrastructure carrying capacity. Non-dominated sorting and crowding distance ensure that the results are diverse and evenly distributed on the Pareto front. From the perspective of graph theory, the plan is mapped to a spatial adjacency graph for the frontier solution set, and the proportion of the largest connected subgraph is used to measure the spatial integrity of the planning plan, so as to ensure that the plot adjustment is not only optimal in terms of the objective function, but also coherent and easy to implement in terms of spatial layout.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A national land space planning management system based on big data, characterized in that: include, The data collection layer is used to obtain remote sensing image data, IoT sensor monitoring data, and multi-department planning document data in real time; A dynamic knowledge graph construction module, connected to the data collection layer, converts land attributes, ecological elements and infrastructure data into an entity-relationship network including spatiotemporal attributes; A conflict detection engine, based on the dynamic knowledge graph, uses a graph neural network to mine implicit conflicts between cross-level planning elements; Visual decision-making platform, used for visual feedback of three-dimensional space deduction and dynamic adjustment plans; In the dynamic knowledge graph construction module, the stage of converting land attributes, ecological elements and infrastructure data into entity-relationship networks with spatiotemporal attributes include: In the multi-source feature fusion stage, the fusion formula is defined as: ,in, Indicates time The fusion feature vector of Represents the time index, represents the remote sensing image feature fusion weight, Indicates time The remote sensing image spectrum-texture matrix, represents the sensor data feature fusion weight, Indicates time The IoT sensor observation vector, represents the document semantic feature fusion weight, Indicates time The planning document semantic embedding vector; In the entity generation mapping phase, the node set formula is defined as: ,in, Indicates time The entity node set, Indicates The mapped entity nodes, represents a feature-type mapping function, Representation Node The land category label, Representation Node The space coordinate vector of Represents an entity index; In the edge weight calculation stage, the comprehensive weight and spatial distance formula is defined as: ， ， in, Representation Node and The comprehensive weight between represents the spatial distance attenuation coefficient, Representation Node With Node The Euclidean distance of Respectively represent nodes The spatial coordinate components of Respectively represent nodes The spatial coordinate components of represents the semantic difference attenuation coefficient, Representation Node and Semantic Hamming distance; In the relationship set screening phase, the edge set formula is defined as: ,in, Indicates time The edge set of represents the weight threshold; In the spatiotemporal embedding generation phase, node embedding and graph structure are defined: ， ， ， in, Representation Node The spatiotemporal embedding vector of represents the node attribute vector, Represents the node timestamp, represents the node space coordinate vector, Indicates time Dynamic knowledge graph of represents the embedding matrix consisting of all node embedding vectors arranged diagonally, Indicates the number of nodes; The conflict detection engine comprises: The implicit conflict identification submodule calculates the impact weight of ecological red line adjustment on related plots based on the graph attention mechanism GAT; The domino effect prediction submodule simulates the cascading propagation path of ecological element changes through a time series diffusion model; The conflict quantitative assessment submodule outputs the ecological risk index and economic loss prediction value of the affected area.

2. A land space planning management system based on big data as claimed in claim 1, characterized in that: The dynamic knowledge graph construction module includes: The entity recognition unit uses a pre-trained BERT model to extract ecological protection zone boundaries and land use nature change record entities from unstructured planning documents; the BERT model uses texts in the field of national land space planning for migration training, and the training corpus includes historical approval documents, ecological assessment reports, and policy and regulatory texts; The relationship extraction unit analyzes the spatial topological relationship and policy constraint relationship between entities through the graph convolutional network GCN; The spatiotemporal coding unit adds timestamp attributes and spatial coordinate attributes to each entity node.

3. The national land space planning management system based on big data as claimed in claim 1, characterized in that: In the domino effect prediction submodule, the cascade impact propagation rules include: The iterative model based on node influence strength combines spatiotemporal decay with relationship weights and is defined as: ， in, Representation Node In the The impact strength at the iteration, represents the node self-attenuation coefficient, Representation and The set of adjacent nodes, represents the time decay rate, Representation Node arrive The time difference, Represents edge The relationship weight, Indicates that it is located in the neighbor set Node index, iterate until or , represents the maximum number of iterations, It means that the iteration will be stopped when the norm of the difference between the impact strengths of two iterations is less than the threshold.

4. The national land space planning management system based on big data as claimed in claim 1, characterized in that: The visual decision-making platform includes: 3D sandbox modeling unit, which maps dynamic knowledge graphs into interactive 3D space models; The real-time simulation unit responds to the planning parameter adjustment instructions and synchronously updates the ecological carrying capacity heat map of the associated area; The plan generation unit automatically outputs a set of adjustment plans that meet multi-objective constraints based on the conflict detection results.

5. A national land space planning management system based on big data as claimed in claim 4, characterized in that: In the real-time deduction unit, the ecological carrying capacity heat map is updated synchronously in the following manner: Map the node influence field obtained by propagation to a continuous space grid and update the first The bearing field at the moment: ， in, Indicates The grid cell at time The ecological carrying capacity of Represent the row index and column index of the two-dimensional grid, respectively. represents the impact conversion coefficient, Represents a grid The neighborhood set of represents the spatial convolution kernel weight, Represents the calculated node influence intensity projected onto the grid After the update is completed, The heatmap is generated after normalization and mapping to a color scale.

6. A land space planning management method based on big data, based on a land space planning management system based on big data according to any one of claims 1 to 5, characterized in that: include: Step S1, real-time acquisition of multi-source heterogeneous data, including satellite remote sensing image streams, groundwater level sensor data, and planning approval electronic files; Step S2, constructing a dynamic knowledge graph that integrates time and space, encoding the ecological red line range, land use approval records, and infrastructure layout data into a weighted graph structure; Step S3, analyzing the cascading impacts of ecological element changes through graph neural networks to identify hidden conflicts across media and time and space scales; Step S4, dynamically displaying the multi-dimensional impact of the planning adjustment scheme in a three-dimensional visualization interface and generating adaptive optimization suggestions.

7. A land space planning management method based on big data as claimed in claim 6, characterized in that: Step S3 includes: A temporal graph convolutional network is used to model the impact of changes in ecological red line boundaries on surrounding plots. Quantify the attenuation capacity of different land types to ecological impacts through node embedding technology; Generate warning level maps for affected areas and recommended buffer zone setting plans.

8. The method for land space planning management based on big data as claimed in claim 6, characterized in that: The generation of the adaptive optimization suggestion in step S4 includes: Establish a multi-objective optimization model with constraints including ecological protection threshold, land use efficiency and infrastructure carrying capacity; The non-dominated sorting genetic algorithm NSGA-II is used to solve the Pareto optimal solution set; Screen the adjustment plan with optimal spatial continuity based on topological data analysis.

Images