Three-dimensional city three-dimensional space intelligent planning management and control method and system and storage medium

Abstract

The application discloses a three-dimensional city space intelligent planning management and control method and system and a storage medium, and the method comprises the following steps: based on a pre-constructed space rule knowledge base, RAG query enhancement and rule digital translation are performed on to-be-queried geographic rule information, and a standardized JSON Schema is generated; a machine executable task chain is generated through a finite state machine model according to the standardized JSON Schema; based on the machine executable task chain, a three-dimensional space scene is constructed, space region analysis and space conflict identification are performed through a multi-modal Agent intelligent agent; and a space planning analysis report is generated according to the constructed three-dimensional space scene graph, the space region analysis result and the space conflict identification result, and geographic information is planned and controlled. The application realizes city intelligent planning management and control by intelligently translating unstructured planning management and control rules into machine executable instructions and autonomously driving a three-dimensional space analysis tool by using an intelligent agent.

Description

Technical Field

[0001] This invention relates to the field of urban intelligent planning and control technology, and in particular to a three-dimensional urban spatial intelligent planning and control method, system and storage medium. Background Technology

[0002] With the accelerating pace of urbanization, urban space utilization is gradually shifting from two-dimensional to three-dimensional, presenting urban planning and management with unprecedented complexity and demands for refined management. Traditional planning control methods primarily rely on manual comparison and judgment of two-dimensional drawings and textual specifications, resulting in low efficiency, strong subjectivity, and difficulty in identifying potential conflicts in three-dimensional space (such as interference between buildings and elevated bridges). As urban buildings develop towards high density and three-dimensionality, the spatial relationship between new buildings and the existing urban environment is becoming increasingly complex, with a significant increase in the risk of conflicts such as sunlight obstruction, visual corridors, building setbacks, aviation height restrictions, and fire safety distances.

[0003] Although existing technologies such as 3D GIS and BIM have emerged to assist spatial analysis, enabling collision detection and viewability analysis, significant limitations still exist:

[0004] First, planning rules are mostly in the form of natural text language or images, which are difficult for computers to understand and execute directly, and lack the ability to automatically analyze the semantics of complex planning and control rules and make real-time decisions.

[0005] Second, the analysis process relies on manually pre-set rules and procedures, which cannot intelligently adapt to changing and complex management needs. After the rules are updated, scripts or plugins need to be redeveloped, resulting in poor flexibility.

[0006] Third, while large language models (LMMs) possess powerful language understanding and reasoning capabilities, they suffer from "illusions" when directly used in planning scenarios. They also struggle to handle multimodal data (text, images, point clouds, Building Information Modeling (BIM), etc.) and cannot intelligently understand planning conditions described in natural language and automatically convert them into executable spatial verification instructions.

[0007] Therefore, how to achieve intelligent planning and management of three-dimensional urban space has become an urgent problem to be solved.

[0008] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0009] The main objective of this invention is to provide a method, system, and storage medium for intelligent planning and control of three-dimensional urban space, aiming to address the technical problem of how to achieve intelligent planning and control of three-dimensional urban space.

[0010] To achieve the above objectives, the present invention provides a three-dimensional urban spatial intelligent planning and control method, the three-dimensional urban spatial intelligent planning and control method comprising:

[0011] Based on a pre-built spatial rule knowledge base, RAG query enhancement and rule digital translation are performed on the geographic rule information to be queried, generating a standardized JSON Schema related to geographic information;

[0012] Based on the standardized JSON Schema, a machine-executable task chain is generated using a finite state machine model;

[0013] Based on the machine-executable task chain, a multimodal agent is used to construct a three-dimensional spatial scene, analyze spatial regions, and identify spatial conflicts. The multimodal agent is built based on the LangChain framework.

[0014] A spatial planning analysis report is generated based on the constructed 3D spatial scene map, spatial region analysis results, and spatial conflict identification results.

[0015] The geographic information is planned and managed according to the spatial planning analysis report.

[0016] Optionally, before performing RAG query enhancement and rule digitization translation on the geographic rule information to be queried based on the pre-built spatial rule knowledge base, the process includes:

[0017] Obtain the text of the Geographical Planning Regulations, attached diagrams of the articles, ecological red line maps and texts, solar radiation coefficient tables, and real-world images;

[0018] The text of the geographical planning regulations, the attached diagrams of the articles, the ecological red line map and text, the solar radiation coefficient table, and the real-scene images are vectorized and embedded to construct a spatial rule knowledge base.

[0019] Optionally, the step of performing RAG query enhancement and rule digitization on the geographic rule information to be queried, generating a standardized JSON Schema related to geographic information, includes:

[0020] Semantic retrieval of the geographic rule information to be queried is performed using dual encoders;

[0021] The geographic rule information to be queried and the semantic search results are concatenated into a text;

[0022] A preliminary structured fragment is generated based on the spliced ​​text, and geographical information is extracted from the preliminary structured fragment;

[0023] The preliminary structured fragments are then digitally translated according to rules to generate a standardized JSON Schema related to the geographic information.

[0024] Optionally, generating a machine-executable task chain using a finite state machine model based on the standardized JSON Schema includes:

[0025] The standardized JSON Schema is parsed using a finite state machine model to extract rule elements;

[0026] Generate a machine-executable task chain based on the parsed rule elements.

[0027] Optionally, the step of constructing a three-dimensional spatial scene, analyzing spatial regions, and identifying spatial conflicts through a multimodal agent includes:

[0028] A multimodal agent is used to perform coordinate transformation and data loading on the geographic information to generate a three-dimensional spatial scene map;

[0029] Buffer analysis and view area analysis are performed on the three-dimensional spatial scene map to obtain spatial region analysis results;

[0030] Based on the spatial region analysis results, three-dimensional urban spatial conflicts are identified using the triangular facet-level collision intersection detection method.

[0031] Optionally, the step of performing coordinate transformation and data loading on the geographic information through a multimodal agent to generate a three-dimensional spatial scene map includes:

[0032] Based on a multimodal agent, the geographic information is converted into a three-dimensional geographic coordinate range through a geographic encoder;

[0033] Based on the stated 3D geographic coordinate range, the tilted feature model, terrain TIN, and vector boundary data are matched from the 3D spatiotemporal database.

[0034] A three-dimensional spatial scene map is constructed using a three-dimensional engine based on the tilted terrain model, the terrain TIN, and the vector boundary data.

[0035] Optionally, the step of performing buffer analysis and view area analysis on the three-dimensional spatial scene map to obtain spatial region analysis results includes:

[0036] The buffer analysis and view area analysis were performed on the three-dimensional spatial scene map using the Constant-Slope Offset method and the binary voxel mesh method, respectively, to obtain the buffer analysis results and view area analysis results.

[0037] By integrating the buffer analysis results and the visible area analysis results, spatial region analysis results are obtained.

[0038] Optionally, the step of identifying three-dimensional urban spatial conflicts using a triangular-level collision intersection detection method based on the spatial region analysis results includes:

[0039] Based on the spatial region analysis results, the three-dimensional spatial scene map is spatially partitioned using the octree method;

[0040] The collision and intersection detection method at the triangular level is used to identify three-dimensional urban spatial conflicts in each zone.

[0041] Furthermore, to achieve the above objectives, the present invention also proposes a three-dimensional urban spatial intelligent planning and control system, which includes:

[0042] The data processing module is used to perform RAG query enhancement and rule digital translation on the geographic rule information to be queried based on a pre-built spatial rule knowledge base, and generate a standardized JSON Schema related to geographic information.

[0043] The intelligent planning and control module is used to generate a machine-executable task chain based on the standardized JSON Schema using a finite state machine model;

[0044] The intelligent planning and control module is also used to construct three-dimensional spatial scenes, analyze spatial regions, and identify spatial conflicts based on the machine-executable task chain through a multimodal agent intelligent body. The multimodal agent intelligent body is built based on the LangChain framework.

[0045] The intelligent planning and control module is also used to generate a spatial planning analysis report based on the constructed three-dimensional spatial scene map, spatial region analysis results, and spatial conflict identification results.

[0046] The intelligent planning and control module is also used to plan and control the geographic information based on the spatial planning analysis report.

[0047] Furthermore, to achieve the above objectives, the present invention also proposes a three-dimensional urban spatial intelligent planning and control device, the device comprising: a memory, a processor, and a three-dimensional urban spatial intelligent planning and control program stored in the memory and executable on the processor, the three-dimensional urban spatial intelligent planning and control program being configured to implement the steps of the three-dimensional urban spatial intelligent planning and control method described above.

[0048] Furthermore, to achieve the above objectives, the present invention also proposes a storage medium storing a three-dimensional urban spatial intelligent planning and control program, wherein when the three-dimensional urban spatial intelligent planning and control program is executed by a processor, it implements the steps of the three-dimensional urban spatial intelligent planning and control method described above.

[0049] This invention first enhances the query results of geographic rules by performing RAG query enhancement and digital translation based on a pre-built spatial rule knowledge base, generating a standardized JSON schema related to geographic information. Then, based on the standardized JSON schema, a machine-executable task chain is generated using a finite state machine model. Based on this machine-executable task chain, a multimodal agent is used to construct a 3D spatial scene, analyze spatial regions, and identify spatial conflicts. The multimodal agent is built based on the LangChain framework. Subsequently, a spatial planning analysis report is generated based on the constructed 3D spatial scene map, spatial region analysis results, and spatial conflict identification results. Finally, geographic information is planned and controlled based on the spatial planning analysis report. This invention intelligently translates unstructured planning and control rules into machine-executable instructions and utilizes an AI agent to autonomously drive 3D spatial analysis tools, achieving digitalization of planning and control rules, automatic identification of 3D spatial conflicts, and intelligent decision-making, thus meeting the needs of intelligent and automated planning management. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the structure of a three-dimensional urban spatial intelligent planning and control device for the hardware operating environment involved in the embodiments of the present invention;

[0051] Figure 2 This is a flowchart illustrating the first embodiment of the three-dimensional urban spatial intelligent planning and control method of the present invention;

[0052] Figure 3 This is a structural block diagram of the first embodiment of the three-dimensional urban spatial intelligent planning and control system of the present invention.

[0053] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0054] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0055] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a three-dimensional urban spatial intelligent planning and control device for the hardware operating environment involved in the embodiments of the present invention.

[0056] like Figure 1 As shown, the three-dimensional urban spatial intelligent planning and control device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk storage device. Optionally, the memory 1005 may also be a storage system independent of the aforementioned processor 1001.

[0057] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the intelligent planning and control equipment for three-dimensional urban space. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0058] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating system, a network communication module, a user interface module, and a three-dimensional urban spatial intelligent planning and control program.

[0059] exist Figure 1In the illustrated three-dimensional urban spatial intelligent planning and control device, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the three-dimensional urban spatial intelligent planning and control device of the present invention can be set in the three-dimensional urban spatial intelligent planning and control device, and the three-dimensional urban spatial intelligent planning and control device calls the three-dimensional urban spatial intelligent planning and control program stored in the memory 1005 through the processor 1001 and executes the three-dimensional urban spatial intelligent planning and control method provided in the embodiment of the present invention.

[0060] This invention provides a three-dimensional urban spatial intelligent planning and control method, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the intelligent planning and control method for three-dimensional urban spatial structure according to the present invention.

[0061] In this embodiment, the three-dimensional urban spatial intelligent planning and control method includes the following steps:

[0062] S1, based on a pre-built spatial rule knowledge base, performs RAG query enhancement and rule digitization translation on the geographic rule information to be queried, generating a standardized JSON Schema related to geographic information.

[0063] It is easy to understand that the executing entity of this embodiment can be a three-dimensional urban spatial intelligent planning and control system with functions such as data processing, network communication and program operation, or other computer equipment with similar functions. This embodiment does not limit it.

[0064] It should be noted that the process involves acquiring the text of the geographical planning regulations, attached diagrams of the articles, ecological red line maps and texts, sunshine coefficient tables, and real-scene images; and then performing vectorization embedding processing on the text of the geographical planning regulations, attached diagrams of the articles, ecological red line maps and texts, sunshine coefficient tables, and the real-scene images to construct a spatial rule knowledge base.

[0065] In practical implementation, the process of collecting rule samples (i.e., texts of geographical planning regulations, attached diagrams of articles, ecological red line maps and texts, solar radiation coefficient tables, and real-world images) and performing vectorized encoding to form a local rule knowledge base (i.e., a spatial rule knowledge base) is as follows:

[0066] (1) Collect multimodal data such as national / local planning regulations text (i.e., geographical planning regulations text), articles and illustrations, ecological red line maps and texts, solar radiation coefficient tables, and real-scene images;

[0067] (2) The text is segmented into four levels of granularity: “clause-figure-table-example”, and then vectorized and embedded to form a local rule knowledge base. The text is embedded using the Baidu General Embedding-large-Chinesebge-large-zh model, and the images are embedded using the contrastive language-image pre-trained visual transformer CLIP-ViT, for a total of 1024-dimensional vectors.

[0068] Taking the "Urban Residential Area Planning and Design Standards" as an example, Article 4.0.9 of the original content states: The sunshine standard for newly built residential buildings in old urban areas should not be less than 1 hour of sunshine on the day of the Great Cold.

[0069] Text data processing involves segmentation into a standardization process:

[0070] Clause Level: Extracting Clause Numbers and Core Rules

[0071] →{Clause ID: 4.0.9, Type: Spacing Control, Content: Sunlight Standards for New Residential Buildings in Old District Urban Development Projects...}

[0072] Image level: Attached diagrams to related clauses (e.g., schematic diagram of sunshine standards).

[0073] →{Figure ID: 4.0.9-fig1, Type: Schematic Diagram, Related Terms: 4.0.9, Image Path: . / figs / 4.0.9.png}

[0074] Table level: Extraction of residential building daylighting standards table

[0075] → {Table ID: 4.0.9-tab1, Type: Residential Building Sunlight Standard, Related Clause: 4.0.9, Data: [{Urban Permanent Population (10,000s): >50}, {Sunlight Standard Day: Great Cold Day}{Sunlight Hours: >2h}...]}

[0076] Example level: Provide real-world case studies (such as building spacing design for a residential community at 30° latitude).

[0077] →{Example ID: 4.0.9-exp1, Type: Case, Related Clause: 4.0.9, Description: A new construction project in an old urban area with a permanent population of >500,000 needs to meet the requirement of more than 2 hours of sunshine per day}

[0078] The structured data described above is input into the corresponding embedding model to generate a 1024-dimensional vector, which is then stored in the local rule knowledge base. See Table 1 for an example of the storage content of the local rule knowledge base.

[0079] Table 1

[0080]

[0081] It should also be noted that, in order to achieve rapid adaptation and continuous learning after changes to the planning rules, conflict results can be manually confirmed and then fed back into the rule knowledge base for incremental fine-tuning, thus enabling continuous learning. The specific process is as follows:

[0082] Manual confirmation of results → positive / negative samples → LoRA incremental fine-tuning → online update of local rule knowledge base → adaptive rule evolution.

[0083] Furthermore, the processing method for enhancing the RAG query and digitizing the rules to generate a standardized JSON Schema related to geographic information is as follows: semantic retrieval of the geographic rule information to be queried is performed using a dual encoder; the geographic rule information to be queried and the semantic retrieval results are concatenated; a preliminary structured fragment is generated based on the concatenated text, and geographic information is extracted from the preliminary structured fragment; the preliminary structured fragment is then digitized to generate a standardized JSON Schema related to geographic information.

[0084] It should be noted that the geographic rule information to be queried is the query rule information for a specific geographic location entered by the user. This rule information can be text, tables, images, etc., such as "query the rules and regulations of the Shenzhen Mangrove Reserve".

[0085] The specific implementation method is as follows:

[0086] a) Multi-stage Retrieval Enhancement Generation (RAG)

[0087] Search phase:

[0088] Based on a pre-built rule knowledge base, a dual encoder is used to perform semantic retrieval on the input rule text. Let the query text be Q, and the embedding vector of the i-th entry in the knowledge base be... The retrieval scoring function is as shown in Formula 1:

[0089] Formula 1

[0090] in and The encoders for the query and the document are respectively used to calculate the relevance using cosine similarity and return the Top-K most relevant entries.

[0091] Generation phase:

[0092] The search results are concatenated with the original rule text and input into a fine-tuned language model to generate preliminary structured fragments. The generation process is based on conditional probability formula 2:

[0093] Formula 2

[0094] Where X is the input rule, R is the retrieved relevant knowledge entries, Y is the generated sequence, and t represents the position in the sequence, i.e., the sequential index.

[0095] b) Digital translation of rules

[0096] Geographic semantic analysis:

[0097] Geographic information (such as "Shenzhen Mangrove Nature Reserve") is extracted from rules by using a pre-trained geographic entity recognition model.

[0098] Spatial operator mapping:

[0099] Natural language descriptions (such as "intersection" and "overlay") are mapped to standard operators (such as "intersect" and "cross") based on semantic matching algorithms (such as edit distance and word vector cosine similarity weighting).

[0100] Numerical threshold parsing: This method combines regularization rules with neural networks to extract numerical values ​​and units. For example:

[0101] Threshold = {Value: v, Unit: u, Comparison operator: op}

[0102] v is extracted from the text using a CRF model, and op (e.g., >, <) is determined by a classifier.

[0103] Object type classification:

[0104] Entity linking is performed based on a type dictionary in the knowledge base (e.g., "bird migration corridor" → bird_migration_corridor), and context relevance is calculated using an attention mechanism, as shown in Formula 3:

[0105] Formula 3

[0106] Where t is the context vector, e i This is the embedding of the i-th entry in the type dictionary.

[0107] Control intensity level: Based on the source of the rules, the severity of the wording, and the policy level, it is assigned a level of "high", "medium" or "low" for subsequent conflict adjudication and priority determination.

[0108] The above elements together constitute a rule-based logical unit, which is essentially a formal expression of applying conditional judgments to spatial objects.

[0109] c) Standardized JSON Schema Generation

[0110] Decoding strategies based on template constraints (such as JSON Schema-guided decoding) ensure that the output conforms to a predetermined structure, including position (e.g., "Shenzhen Mangrove Nature Reserve"), operator (e.g., "intersect"), target (e.g., "bird_migration_corridor"), threshold (e.g., {"height":">100m"}), and severity (e.g., "high").

[0111] Standardized JSON Schema (e.g., {"position":"Shenzhen Mangrove Nature Reserve", "operator": "intersect", "target": "bird_migration_corridor", ...}).

[0112] S2, Generate a machine-executable task chain using a finite state machine model based on the standardized JSON Schema.

[0113] S3, based on the machine-executable task chain, a multimodal agent is used to construct a three-dimensional spatial scene, analyze spatial regions, and identify spatial conflicts. The multimodal agent is built based on the LangChain framework.

[0114] It should be noted that the multimodal large model agent intelligent system based on the LangChain framework performs the following process for intelligent planning and control: rule matching (Planner) → data coordinate transformation and loading (Geocoder) → spatial analysis (GIS-Analysis) → conflict determination (Conflict-Detector) → report output (Report-Generator).

[0115] Furthermore, the process of generating a machine-executable task chain based on the standardized JSON Schema using a finite state machine model is as follows: the standardized JSON Schema is parsed using a finite state machine model to extract rule elements; and a machine-executable task chain is generated based on the parsed rule elements.

[0116] The specific implementation method is as follows:

[0117] The Finite State Machine (FSM) model is used to parse the rule elements of the standardized JSON Schema and generate a machine-executable task chain.

[0118] Task chain is defined as:

[0119] TaskChain=[Geocoder,LoadData,GIS-Analysis,Conflict-Detector,Report-Generator]

[0120] Geocoder parses the input location information into geographic coordinates and loads them. GIS-Analysis maps the data to specific spatial analysis operations based on the operator type (such as intersect, buffer).

[0121] Assuming a standardized JSON Schema, such as {"position":"Shenzhen Mangrove Reserve", "operator": "intersect", "target": "bird_migration_corridor", ...}, then a finite state machine model is used to parse the rule elements and obtain a machine-executable task sequence (such as [Position(Shenzhen Mangrove Reserve), Buffer(100m), Intersect(bird_migration_corridor)]).

[0122] Furthermore, the processing method for constructing three-dimensional spatial scenes, analyzing spatial regions, and identifying spatial conflicts through multimodal agent intelligence is as follows: the multimodal agent intelligence performs coordinate transformation and data loading on geographic information to generate a three-dimensional spatial scene map; buffer analysis and visibility analysis are performed on the three-dimensional spatial scene map to obtain spatial region analysis results; and based on the spatial region analysis results, the three-dimensional urban spatial conflicts are identified using the triangular facet-level collision intersection detection method.

[0123] Furthermore, the process of generating a 3D spatial scene map by performing coordinate transformation and data loading on the geographic information using a multimodal agent is as follows: Based on the multimodal agent, the geographic information is converted into a 3D geographic coordinate range through a geographic encoder; according to the 3D geographic coordinate range, tilted feature models, terrain TINs, and vector boundary data are matched from a 3D spatiotemporal database; and a 3D spatial scene map is constructed using a 3D engine based on the tilted feature models, terrain TINs, and vector boundary data.

[0124] In this embodiment, the geographic information appearing in the rules is converted into a three-dimensional geographic coordinate range (latitude, longitude, and elevation) using a geocoding service. The coordinates are then uniformly converted to the local 2000 National Geodetic Coordinate System + 1985 Elevation Datum. Subsequently, based on the coordinate range, the corresponding oblique photogrammetry model, terrain TIN, vector boundary, and other data are queried from a three-dimensional spatiotemporal database, and combined with the three-dimensional engine to construct a three-dimensional spatial scene graph.

[0125] Furthermore, the spatial region analysis results are obtained by performing buffer analysis and view area analysis on the 3D spatial scene map. The processing method is as follows: the buffer analysis and view area analysis are performed on the 3D spatial scene map by the Constant-Slope Offset method and the binary voxel mesh method respectively to obtain the buffer analysis results and view area analysis results; the buffer analysis results and view area analysis results are integrated to obtain the spatial region analysis results.

[0126] In this embodiment, the 3D GIS kernel is invoked to perform buffer and view area analysis. The Agent parallelizes the GPU spatial analysis operators on the server side, and all calculations are completed in the GPU memory, ensuring ≥10 6 Piece throughput per second.

[0127] Buffer analysis

[0128] For non-flat surfaces, the Constant-Slope Offset method is used: for any vertex p, its expansion vector is calculated as shown in Formula 4:

[0129] Formula 4

[0130] Where d is the horizontal buffer distance, and θ is the slope threshold (default 45°). It is the surface normal vector. It is a slope vector.

[0131] After generating a 3D dilated mesh, perform a Boolean difference with the original object to obtain a buffer shell.

[0132] Viewshed analysis

[0133] GPU-based ray casting: A ray is emitted from the viewpoint O to a 512×512 spherical pixel, and voxels are traversed using Bresenham 3D. Early termination is achieved using a Z-Buffer: if the maximum elevation of the current ray is greater than or equal to the elevation of the target voxel, it is marked as "invisible".

[0134] The calculation of binary voxel meshes is shown in Formula 5:

[0135] Formula 5

[0136] Where Nx = number of voxels in the X-axis direction, Ny = number of voxels in the Y-axis direction, Nz = number of voxels in the Z-axis direction, and 1 indicates that it is visible.

[0137] The results of the above spatial operators are encapsulated into OGC 3DIM standard GeoJSON and written back to shared memory with a topological signature (MD5 hash) for use in subsequent collision phases.

[0138] Furthermore, based on the spatial region analysis results, the method for identifying three-dimensional urban spatial conflicts using the triangular facet-level collision intersection detection method is as follows: based on the spatial region analysis results, the three-dimensional spatial scene map is divided into spatial partitions using the octree method; and the three-dimensional urban spatial conflicts within each partition are identified using the triangular facet-level collision intersection detection method.

[0139] In the implementation, an octree is used to spatially partition the 3D scene graph, improving analysis efficiency. A dynamic BVH (Bounding Volume Hierarchy) is constructed, allowing only overlapping nodes to undergo detailed inspection.

[0140] Triangle-level collision intersection detection (intersect):

[0141] The SAP+GPU parallel scanning algorithm is adopted: First, a spatially sorted array A is constructed along the Z-axis. Then, a Segment Tree is used to store the maximum and minimum elevations of the interval. Möller intersection tests are performed on each pair of triangular faces to obtain the intersection line set Lines={l i The proportion of the intersecting area to the object's surface is calculated using Formula 6:

[0142] Formula 6: η = Area(Lines) / Area(obj)

[0143] Where Area(Lines) represents the projected area of ​​these intersecting lines on the common plane, and Area(obj) is the surface area of ​​the feature.

[0144] If Lines ≠ If the area ratio η > 0.01, then it is determined that "intersection is established".

[0145] Rule violation determination:

[0146] Compare the analysis results with the threshold (e.g., "height > 100m"). If the analysis results meet the threshold (e.g., v ≤ 100m), then mark it as a conflict.

[0147] S4 generates a spatial planning analysis report based on the constructed 3D spatial scene map, spatial region analysis results, and spatial conflict identification results.

[0148] The spatial planning analysis report is a report with 3D scene screenshots, conflict list, legal basis, rectification suggestions or solutions. The screenshots, tables and clauses are populated into a predefined HTML template using a template engine (such as Jinja2) and then exported as a PDF.

[0149] 3D scene screenshots:

[0150] The system navigates to a 3D view based on the conflict area, highlights the conflict area, and then calls the scene screenshot interface to generate a scene screenshot with highlighted annotations.

[0151] Conflict list generation:

[0152] Extract the attributes (ID, location, rule violation) of the conflicting object, sort them by severity, and format them as a Markdown table.

[0153] Legal basis:

[0154] Based on the RAG knowledge base, the corresponding legal provisions for the conflict are matched (such as "Article X of the Measures for the Management of Ecological Protection Red Lines").

[0155] In the specific implementation, the constructed 3D spatial scene map, spatial region analysis results, and spatial conflict identification results are analyzed, and corresponding rectification suggestions or solutions are given; then, a spatial planning analysis report is generated based on the constructed 3D spatial scene map, spatial region analysis results, spatial conflict identification results, and corresponding rectification suggestions or solutions.

[0156] S5. Based on the spatial planning analysis report, the geographic information is planned and managed.

[0157] In practice, users can refer to spatial planning analysis reports to assist in planning and management.

[0158] In this embodiment, firstly, based on a pre-built spatial rule knowledge base, the geographic rule information to be queried is enhanced using RAG query and digitally translated into rules, generating a standardized JSON schema related to geographic information. Then, based on the standardized JSON schema, a machine-executable task chain is generated using a finite state machine model. Based on this machine-executable task chain, a multimodal agent is used to construct a 3D spatial scene, analyze spatial regions, and identify spatial conflicts. The multimodal agent is built based on the LangChain framework. Subsequently, a spatial planning analysis report is generated based on the constructed 3D spatial scene map, spatial region analysis results, and spatial conflict identification results. Finally, geographic information is planned and controlled based on the spatial planning analysis report. This embodiment intelligently translates unstructured planning and control rules into machine-executable instructions and utilizes an AI agent to autonomously drive 3D spatial analysis tools, achieving digitalization of planning and control rules, automatic identification of 3D spatial conflicts, and intelligent decision-making, thus meeting the needs of intelligent and automated planning management.

[0159] Reference Figure 3 , Figure 3 This is a structural block diagram of the first embodiment of the three-dimensional urban spatial intelligent planning and control system of the present invention.

[0160] like Figure 3 As shown, the three-dimensional urban spatial intelligent planning and control system proposed in this embodiment of the invention includes:

[0161] Data processing module 3001 is used to perform RAG query enhancement and rule digital translation on the geographic rule information to be queried based on a pre-built spatial rule knowledge base, and generate a standardized JSON Schema related to geographic information.

[0162] The intelligent planning and control module 3002 is used to generate a machine-executable task chain based on the standardized JSON Schema using a finite state machine model;

[0163] The intelligent planning and control module 3002 is also used to construct three-dimensional spatial scenes, analyze spatial regions, and identify spatial conflicts through a multimodal agent based on the machine-executable task chain. The multimodal agent is built based on the LangChain framework.

[0164] The intelligent planning and control module 3002 is also used to generate a spatial planning analysis report based on the constructed three-dimensional spatial scene map, spatial region analysis results and spatial conflict identification results;

[0165] The intelligent planning and control module 3002 is also used to plan and control the geographic information based on the spatial planning analysis report.

[0166] Other embodiments or specific implementations of the three-dimensional urban spatial intelligent planning and control system of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0167] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0168] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0169] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0170] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A three-dimensional urban space intelligent planning and management method, characterized in that, The method comprises the following steps: Based on the pre-constructed spatial rule knowledge base, the RAG query enhancement and rule digitalization translation of the to-be-queried geographic rule information are performed, and a standardized JSON Schema related to geographic information is generated; According to the standardized JSON Schema, a machine executable task chain is generated through a finite state machine model; Based on the machine executable task chain, a three-dimensional space scene is constructed, spatial region analysis and spatial conflict identification are performed through a multi-modal Agent intelligent agent, and the multi-modal Agent intelligent agent is constructed based on a LangChain framework; A spatial planning analysis report is generated according to the constructed three-dimensional space scene graph, spatial region analysis result and spatial conflict identification result; The geographic information is planned and controlled according to the spatial planning analysis report; Before the RAG query enhancement and rule digitalization translation of the to-be-queried geographic rule information based on the pre-constructed spatial rule knowledge base, the following steps are included: Obtain the geographic planning regulation text, article appendix, ecological red line graph and text, sunshine coefficient table and real scene image; The geographic planning regulation text, article appendix, ecological red line graph and text, sunshine coefficient table and real scene image are subjected to vector embedding processing to construct a spatial rule knowledge base; The RAG query enhancement and rule digitalization translation of the to-be-queried geographic rule information, and the generation of the standardized JSON Schema related to geographic information, comprise the following steps: The to-be-queried geographic rule information is subjected to semantic retrieval through a double encoder; The to-be-queried geographic rule information and the semantic retrieval result are subjected to text splicing; Preliminary structured segments are generated according to the spliced text, and geographic information is extracted from the preliminary structured segments; The preliminary structured segments are subjected to rule digitalization translation to generate the standardized JSON Schema related to geographic information; The three-dimensional space scene construction, spatial region analysis and spatial conflict identification through the multi-modal Agent intelligent agent comprise the following steps: The geographic information is subjected to coordinate conversion and data loading through the multi-modal Agent intelligent agent to generate a three-dimensional space scene graph; The three-dimensional space scene graph is subjected to buffer zone analysis and visual range analysis to obtain a spatial region analysis result; According to the spatial region analysis result, a three-dimensional urban space conflict is identified through a triangular face level collision intersection detection method; The geographic information is converted into a three-dimensional geographic coordinate range through a geographic encoder based on the multi-modal Agent intelligent agent; According to the three-dimensional geographic coordinate range, a tilted ground object model, terrain TIN and vector boundary data are matched from a three-dimensional space-time database; According to the tilted ground object model, the terrain TIN and the vector boundary data, a three-dimensional space scene graph is constructed through a three-dimensional engine; According to the spatial region analysis result, a three-dimensional urban space conflict is identified through a triangular face level collision intersection detection method, which comprises the following steps: ​ Based on the spatial region analysis result, the three-dimensional space scene graph is spatially partitioned by an octree method; Each partitioned space is identified for three-dimensional city space conflict by a triangular face level collision intersection detection method.

2. The method of claim 1, wherein, The machine executable task chain is generated by a finite state machine model according to the standardized JSON Schema, including: The rule element of the standardized JSON Schema is parsed by a finite state machine model; The machine executable task chain is generated according to the parsed rule element.

3. The method of claim 1, wherein, The spatial region analysis result is obtained by buffer analysis and viewshed analysis of the three-dimensional space scene graph, including: The buffer analysis and viewshed analysis of the three-dimensional space scene graph are performed by a Constant-Slope Offset method and a binary voxel grid method respectively to obtain buffer analysis results and viewshed analysis results; The spatial region analysis result is obtained by integrating the buffer analysis results and the viewshed analysis results.

4. A three-dimensional urban space intelligent planning and management system, characterized in that, The system includes: A data processing module configured to perform RAG query enhancement and rule digital translation on the to-be-queried geographic rule information based on a pre-constructed spatial rule knowledge base, and generate a standardized JSON Schema related to geographic information; An intelligent planning and control module configured to generate a machine executable task chain by a finite state machine model according to the standardized JSON Schema; The intelligent planning and control module is further configured to perform three-dimensional space scene construction, spatial region analysis and spatial conflict identification by a multi-modal Agent intelligent agent based on the machine executable task chain, wherein the multi-modal Agent intelligent agent is constructed based on a LangChain framework; The intelligent planning and control module is further configured to generate a spatial planning analysis report according to the constructed three-dimensional space scene graph, spatial region analysis result and spatial conflict identification result; The intelligent planning and control module is further configured to perform planning and control on the geographic information according to the spatial planning analysis report; The three-dimensional city space intelligent planning and control system is executed to implement the steps of the three-dimensional city space intelligent planning and control method of claim 1.

5. A storage medium, characterized by The storage medium stores a three-dimensional city space intelligent planning and control program, and the three-dimensional city space intelligent planning and control program is executed by the processor to implement the steps of the three-dimensional city space intelligent planning and control method of any one of claims 1 to 3.

Images