Intelligent optimization method for 3d form of old block based on rigid-flexible rule driving

Abstract

The application discloses an old street block three-dimensional form intelligent updating optimization method based on rigid-flexible rule driving, which comprises the following steps: collecting target old street block vector data, design specification and standard text data of a city; establishing a rule graph of rigid industry specification requirements of city design; automatically labeling spatial structure, function layout, road system and blue-green system vector data of the design by using Mask R-CNN coding technology, forming a knowledge graph of classic city design achievement rules, and establishing a rule graph of flexible professional knowledge guidance of city design; establishing a multi-layer network association of old street block three-dimensional form graph network, rigid rule graph and flexible knowledge graph; outputting an optimized three-dimensional form scheme of the target old street block to a spatial holographic sand table, and finally outputting an optimized three-dimensional form graph network and a visual scheme of the old street block. The application can realize rigid rule and flexible knowledge hybrid driving to optimize the three-dimensional form of the old street block.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban planning, in particular to a three-dimensional form intelligent updating optimization method for old street blocks based on rigid-flexible rule driving. BACKGROUND

[0002] Old street block form updating optimization is an important part of urban design work, and has important significance in urban sustainable development and urban image shaping. Reasonable updating optimization of old street block form will take into account rigid rules and flexible knowledge, not only to ensure that the urban renewal plan meets the requirements of specifications and policies, but also to meet the design experience, local culture and aesthetic standards, and help to balance the interests of government, developers, designers and residents. The traditional old street block form updating optimization still has a lot of room for improvement in terms of rigid-flexible rule consideration, and lacks rule atlas support and cannot be iteratively optimized based on intelligent algorithms and network association. SUMMARY

[0003] (I) Technical problems solved

[0004] In view of the deficiencies of the prior art, the present application provides a three-dimensional form intelligent updating optimization method for old street blocks based on rigid-flexible rule driving, which solves the problem of lack of rigid rule control and flexible design knowledge in traditional old street block form updating optimization, lack of scientific rationality, and inability to dynamically interact with the three-dimensional form of old street blocks to iteratively optimize the scheme.

[0005] (II) Technical solutions

[0006] To achieve the above purpose, the present application is realized by the following technical solutions:

[0007] A three-dimensional form intelligent updating optimization method for old street blocks based on rigid-flexible rule driving, comprising:

[0008] (1) Collecting target old street block vector data, design specification and standard text data of the city where the old street block is located, and design scheme picture data of other blocks in the city.

[0009] (2) Extracting control terms and control specific content from the design specification and standard text data of the city where the old street block is located, and using the extensible basic element model method to perform basic structural representation of "matter element-event element-relationship element"; from the association relationship of the sentence, according to the conjugate relationship-complementary relationship-constraint relationship-suggestion relationship, the basic element model is classified and stored, and a rule atlas of rigid industry specification requirements of urban design is established.

[0010] (3) Using topological graph structure to associate with three-dimensional space calculation, extracting professional knowledge rules of urban design space aesthetics; using Mask R-CNN coding technology to automatically label the spatial structure, functional layout, road system, blue-green system vector data of classic urban design results, forming the knowledge graph of classic urban design results rules, and establishing the rule graph of flexible professional knowledge guidance of urban design.

[0011] (4) Establishing the multi-layer network association of the three-dimensional form graph network, rigid rule graph and flexible knowledge graph of the old street block, gradually iterating the three-dimensional form graph network of the old street block until meeting the constraint conditions of the rigid rule graph and the flexible knowledge graph, and outputting the scheme.

[0012] (5) Outputting the optimized three-dimensional form scheme of the target old street block to the spatial holographic sand table, performing street block three-dimensional multi-scheme display and interactive optimization, and finally outputting the optimized three-dimensional form graph network and visual scheme of the old street block.

[0013] Preferably, the vector data of the target old street block, the design specification and standard text data of the city where the target old street block is located, and the design scheme picture data of other blocks in the city are collected in step (1), specifically including:

[0014] According to the determined range of the target old street block, the vector data of the block in the city where the target old street block is located is obtained through the OpenStreetMap open data platform, and the design specification and standard text data of the city where the target old street block is located and the design scheme picture data of other blocks in the city are obtained through the Baidu open data platform;

[0015] The vector data of the block in the city where the target old street block is located includes plot data and building data, and the data format is shapefile format; the plot data shp file should include land use type information, geographic coordinate information, plot area information and plot shape information; the building data shp file should include building function information, building layer information, building height information, building shape information and building position information;

[0016] The design specification and standard text data of the city where the target old street block is located include city land use classification and planning construction land standard, city public facility planning specification, city road traffic planning and design specification, city environmental sanitation facility planning specification, city residential area planning and design specification, city engineering pipeline comprehensive planning specification, city environment planning standard, city water supply engineering design specification, city drainage engineering design specification, city road greening planning and design specification, target city planning technical standard and criterion;

[0017] Image data of design plans for other blocks in the city, including design drawings for block stock renewal, environmental improvement of old blocks, optimization of public facilities and green spaces, and protection and renewal of urban landscape.

[0018] Preferably, in step (2), control terms and specific control contents are extracted from the design specifications and standard text data of the city, and a basic structured representation of "matter element-event element-relation element" is performed using the extension primitive model method, specifically including:

[0019] Use the collected design specifications and standard text data from the cities where the old blocks are located to extract control terms and specific control content, including road network control rules, open space control rules, height form control rules, and environmental and facility control rules;

[0020] The extension element model method refers to the use of formal language to extract control terms and specific control contents from the design specifications and standard text data of the city. Its logical unit is the basic element, including matter element, event element and relationship element.

[0021] The matter-element M that describes things is expressed as:

[0022] M=(O m ,c m ,v m )

[0023] Among them, O m Represents an object, c m Indicates the name of the feature, v m Indicates O m About c m the value of the quantity taken;

[0024] The expression of event element A describing the interaction between objects is:

[0025] A=(O a ,c a ,v a )

[0026] Among them, O a Represents an object, c a Indicates the name of the feature, v a Indicates O a About c a the value of the quantity taken;

[0027] The relational element R that describes the relationship between features is expressed as:

[0028]

[0029] Among them, O rIndicates relationship, c r1 , c r2 ,...c rn represents n features, v r1 , v r2 ,...v rn Indicates the corresponding value.

[0030] Preferably, in step (2), the primitive models are classified and stored according to the conjugate relationship, complementary relationship, constraint relationship and suggestion relationship from the association relationship of the statements, and a rule map of the rigid industry specification requirements of urban design is established, which specifically includes:

[0031] Analyze the relationships among the matter-elements, primitives, and relational elements that have been represented in a basic structured manner, and enter them into the NEO4J platform according to four categories: conjugate relationships, complementary relationships, constraint relationships, and recommended relationships. Use the NEO4J platform to digitally encode the rigid industry specifications for urban design that have been represented in a structured manner, and use algorithms to visualize the primitive relationships and generate a rule map.

[0032] We use large language models (LLama-7B, chat-GLM, and Bloom) to fine-tune language constraint models.

[0033] Preferably, in step (3), the topological graph structure is used to perform correlation calculation with the three-dimensional space to extract professional knowledge rules in the urban design space aesthetics, specifically including:

[0034] Aerial images of urban spaces in iconic urban areas around the world were collected as a sample library. These images were segmented and identified using semantic segmentation and recognition technology, with image elements classified into five categories: buildings, roads, squares, green spaces, and water systems. These elements were then abstracted into a topological graph structure composed of nodes and edges. A multi-layered graph neural network deep learning model was used to process this topological graph structure, capturing patterns and information within the graph by learning the relationships between nodes. This capture was enhanced through a hierarchical learning structure. Spatial self-learning technology was used to calculate quantitative indicators such as spatial agglomeration, openness, and staggeredness in the case study blocks. These indicators were then correlated with the topological graph structure to extract professional knowledge and rules regarding urban design spatial aesthetics, forming a knowledge graph of spatial aesthetic rules.

[0035] Preferably, in step (3), Mask R-CNN encoding technology is used to automatically annotate the spatial structure, functional layout, road system, and blue-green system vector data of the design to form a knowledge graph of classic urban design results rules and establish a rule graph required by flexible industry specifications for urban design, specifically including:

[0036] Classic urban design achievements are collected, and image data of their design proposals is identified using a convolutional neural network. This data is extracted as two-dimensional vector graphics representing urban design floor plans. Mask R-CNN encoding technology is then used to annotate the spatial structure, functional layout, road system, and blue-green system vector data of the design proposals. The spatial structure vector data includes the geometric center of each functional core, such as the primary and secondary cores; the functional layout vector data includes the functional layout data for commercial, office, residential, industrial, green space, public services, and transportation; the road system vector data includes three-level vector information for main roads, secondary roads, and branches; and the blue-green system vector data includes vector data for green corridors, secondary green corridors, and water areas. Furthermore, knowledge graph technology is used to reverse-analyze the masterful styles of urban design in these classic works, forming a knowledge graph of classic design rules. Furthermore, this knowledge graph, combined with the knowledge graph of spatial aesthetic rules, forms a rule graph that meets the requirements of flexible industry specifications for urban design.

[0037] Preferably, in step (4), a multi-layer network association of the old block three-dimensional morphological graph network, the rigid rule graph and the flexible knowledge graph is established, specifically including:

[0038] Based on the 3D morphology network of old neighborhoods, the rule graphs required by rigid industry specifications for urban design, and the rule graphs required by flexible industry specifications for urban design, the 3D morphology network of old neighborhoods is gradually iterated according to the site location of the design plan until the constraints of the rigid rule graph and the flexible knowledge graph are satisfied, outputting the rules and knowledge parameter data of potential urban design solutions. On this basis, vector data associated with potential solutions from the sample library is screened out, thereby establishing a process for automatically associating knowledge rules with physical spatial elements. This screening process utilizes a gradient descent method, observing the fluctuations in the loss function of the convolutional pooling layer of the corresponding model during training for each parameter. The optimal learning rate and number of iterations are determined by comparing training time and generated results until the optimal value is achieved. Ultimately, a multi-layer network association is established, connecting the 3D morphology network of old neighborhoods, the rigid rule graph, and the flexible knowledge graph.

[0039] Preferably, the step (5) of outputting the optimized three-dimensional morphological network and visualization scheme of the old blocks specifically includes:

[0040] The three-dimensional morphological schemes of the old blocks generated by the layer-by-layer iteration of the simulation, which meet both the rigid rule graph and the flexible knowledge graph, are output to the spatial holographic sandbox for display; with the assistance of VR\AR equipment and combined with gesture instructions, the three-dimensional morphological schemes of the old blocks are displayed, compared, selected, and modified, and the local spatial structure, functional layout, road system, and blue-green system are adjusted and interactively optimized before being output in the form of a three-dimensional model file of SketchUp; the VR\AR equipment includes VRTRIX data gloves, VR glasses, and VR 3D display stands.

[0041] (3) Beneficial effects

[0042] (1) The present invention is based on a rigid-flexible rule-driven intelligent renewal and optimization method for the three-dimensional form of old blocks. By performing basic structural representation on the collected old block vector data, design specifications and standard text data, and design scheme image data, the method ensures the standardized extraction of target urban area information, clearly expresses the logical relationship between design elements, and assists design decisions based on complex urban systems.

[0043] (2) The present invention is based on a method for intelligently updating and optimizing the three-dimensional morphology of old blocks driven by rigid and flexible rules. This method overcomes the problem that the traditional optimization method for the morphology update of old blocks still has a large room for improvement in terms of balancing rigid and flexible rules, lacks support from rule graphs, and cannot perform iterative optimization of solutions based on intelligent algorithms and network associations. Instead, it constructs a rule graph based on rigid rules and flexible knowledge.

[0044] (3) The present invention is based on a method for intelligently updating and optimizing the three-dimensional form of old blocks driven by rigid-flexible rules. It uses Mask R-CNN coding technology to automatically annotate design vector data to form a flexible knowledge graph. By automatically associating knowledge rules with physical spatial elements, it establishes design standards in terms of spatial structure, functional layout, road system, and blue-green system, thereby reducing the arbitrariness and uncertainty based on flexible knowledge in the design process.

[0045] (4) The present invention is based on a three-dimensional intelligent renewal optimization method for old blocks driven by rigid and flexible rules. It uses a gradient descent method to screen out the rules and knowledge parameter data of the optimal potential urban design solutions, reducing the time and energy of manual parameter adjustment, automatically executing the iterative process, avoiding the problems of low design efficiency and the inability to ensure the scientificity and rationality of design decisions, and ensuring that the design solution can better adapt to the renewal optimization process of old urban areas under different design scenarios and needs.

[0046] (5) The present invention is based on a method for intelligently updating and optimizing the three-dimensional form of old blocks driven by rigid-flexible rules. The present invention allows users to output the three-dimensional model of the block through the Hololens 2 holographic computer and display the corresponding image on VR glasses. With the help of VRTRIX data gloves and VR glasses, users can wear full-angle equipment and freely select and operate any local spatial structure, functional layout, road system, and blue-green system to make push-pull adjustments. By clicking on the indicator calculation function, the impact of the block adjustment on the overall three-dimensional form characteristic parameters of the block can be displayed in real time, thereby enhancing the interactivity, display, and convenience of the decision-making process. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Flow chart of the method of the present invention; DETAILED DESCRIPTION

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0049] Example

[0050] like Figure 1 As shown, the technical solution of the present invention will be described in detail below by taking the urban design of an old block in a certain city as an example.

[0051] (1) Collect vector data of the target old block, text data of the design specifications and standards of the city where it is located, and image data of design plans for other blocks in the city, including:

[0052] (1.1) Based on the scope of the identified target old neighborhoods, obtain the neighborhood vector data of the city where the target old neighborhoods are located through the OpenStreetMap open data platform, including land parcel data and building data, in shapefile format. The land parcel data (shp file) should include land use type information, geographic coordinate information, land parcel area information, and land parcel shape information; the building data (shp file) should include building function information, building number of floors information, building height information, building shape information, and building location information;

[0053] (1.2) Utilize Baidu Open Data Platform to obtain text data on design specifications and standards for the city where the applicant is located, as well as image data on design plans for other blocks in the city where the applicant is located; the text data on design specifications and standards for the city where the applicant is located shall include urban land classification and planning and construction land standards, urban public facilities planning specifications, urban road traffic planning and design specifications, urban environmental sanitation facilities planning specifications, urban residential area planning and design specifications, urban engineering pipeline comprehensive planning specifications, urban environmental planning standards, urban water supply engineering design specifications, urban drainage engineering design specifications, urban road greening planning and design specifications, and target city planning technical standards and criteria; image data on design plans for other blocks in the city where the applicant is located shall include design plan drawings for block stock renewal, environmental improvement of old blocks, optimization of public facilities and green space, and protection and renewal of urban landscape.

[0054] (2) Extract control terms and specific control contents from the design specifications and standard text data of the city, and use the extension primitive model method to perform the basic structured representation of "matter element-event element-relation element"; from the association relationship of the sentences, classify and store the primitive models according to the conjugate relationship-complementary relationship-constraint relationship-suggestion relationship, and establish a rule map of the rigid industry specifications for urban design, including:

[0055] (2.1) Using the design specifications and standard text data of the city where the old blocks are located collected in step (1), extract the control terms and specific control content, including road network control rules, open space control rules, height form control rules, and environment and facility control rules;

[0056] The extension element model method refers to the use of formal language to extract control terms and specific control contents from the design specifications and standard text data of the city. Its logical unit is the basic element, including matter element, event element and relationship element.

[0057] The matter-element M that describes things is expressed as:

[0058] M=(O m ,c m ,v m )

[0059] Among them, O m Represents an object, c m Indicates the name of the feature, v m Indicates O m About c m the value of the quantity taken;

[0060] The expression of event element A describing the interaction between objects is:

[0061] A=(O a ,ca ,v a )

[0062] Among them, O a Represents an object, c a Indicates the name of the feature, v a Indicates O a About c a the value of the quantity taken;

[0063] The relational element R that describes the relationship between features is expressed as:

[0064]

[0065] Among them, O r Indicates relationship, c r1 , c r2 ,...c rn represents n features, v r1 , v r2 ,...v rn Indicates the corresponding value.

[0066] (2.2) Analyze the relationships among the matter-elements, primitives, and relational elements that have been represented in a basic structured manner in step (2.1), and enter them into the NEO4J platform according to four categories: conjugate relationships, complementary relationships, constraint relationships, and recommended relationships. Use the NEO4J platform to digitally encode the rigid industry specifications for urban design that have been represented in a structured manner, and use algorithms to visualize the primitive relationships to generate a rule map.

[0067] (2.3) Using large language models to fine-tune language constraint models; the large language models include LLama-7B, chat-GLM, and Bloom.

[0068] (3) Use topological graph structure and three-dimensional space to perform correlation calculations and extract professional knowledge rules in urban design space aesthetics; use Mask R-CNN coding technology to automatically annotate the spatial structure, functional layout, road system, and blue-green system vector data of the design to form a knowledge map of classic urban design results rules and establish a rule map guided by flexible professional knowledge of urban design.

[0069] Specifically include:

[0070] (3.1) Collect aerial photos of urban space in classic urban areas around the world as a sample library. Segment and identify the aerial photos using semantic segmentation and recognition technology, and identify image elements into five categories: buildings, roads, squares, green spaces, and water systems. These elements are abstracted into a topological graph structure composed of nodes and edges. Use a multi-level graph neural network deep learning model to process the topological graph structure, capture patterns and information in the graph by learning the relationships between nodes, and improve the capture effect through a hierarchical learning structure. Use spatial self-learning technology to calculate quantitative indicators such as spatial agglomeration, openness, and staggeredness of case blocks, and perform correlation calculations with the topological graph structure to extract professional knowledge rules in urban design space aesthetics and form a knowledge graph of spatial aesthetic rules.

[0071] (3.2) Collect classic urban design achievements, use convolutional neural networks to identify design scheme image data of these classic urban design achievements, extract them into two-dimensional vector graphics data of urban design floor plans, and use Mask R-CNN encoding technology to annotate the spatial structure, functional layout, road system, and blue-green system vector data of the design scheme. The spatial structure vector data includes the geometric center of each functional core, such as the primary core and secondary core; the functional layout vector data includes the functional layout data of commercial, office, residential, industrial, green space, public services, and transportation; the road system vector data includes three-level vector information of main roads, secondary roads, and branches; and the blue-green system vector data includes vector data information of green corridors, secondary green corridors, and water areas. Furthermore, using knowledge graph technology, reverse-analyze the master styles of urban design in each classic work to form a knowledge graph of classic design rules. Furthermore, combined with the knowledge graph of spatial aesthetic rules described in step (3.1), a rule graph that meets the requirements of flexible industry specifications for urban design is formed.

[0072] (4) Establish a multi-layer network association between the old block three-dimensional morphological graph network, the rigid rule graph, and the flexible knowledge graph, so that the old block three-dimensional morphological graph network is gradually iterated until the constraints of the rigid rule graph and the flexible knowledge graph are met, and the output solution is output, which specifically includes:

[0073] (4.1) Based on the target old block vector data described in step (1), a three-dimensional morphological map network of the old blocks is established. Combined with the rule map required by the rigid industry specification for urban design described in step (2) and the rule map required by the flexible industry specification for urban design described in step (3), the three-dimensional morphological map network of the old blocks is gradually iterated according to the location of the design scheme until the constraints of the rigid rule map and the flexible knowledge map are met, and the rule and knowledge parameter data of the potential urban design scheme are output.

[0074] (4.2) Based on this, potential solutions are screened for vector data associated with the sample library, thereby establishing an automatic association process between knowledge rules and physical spatial elements. This screening process uses a gradient descent method to observe the fluctuations in the loss function of the convolutional pooling layer during training of each parameter-based model. The optimal learning rate and number of iterations are determined by comparing training time and generated results until the optimal value is achieved.

[0075] (4.3) Finally, a multi-layer network association of the three-dimensional morphological graph network of old blocks, the rigid rule graph and the flexible knowledge graph is established.

[0076] (5) Output the optimized three-dimensional morphological scheme of the target old block to the spatial holographic sand table, conduct three-dimensional multi-scheme display and interactive optimization of the block, and finally output the optimized three-dimensional morphological network and visualization scheme of the old block, specifically including:

[0077] (5.1) Output the multiple three-dimensional morphological schemes of old blocks generated by the layer-by-layer iteration of the simulation, which satisfy both the rigid rule graph and the flexible knowledge graph, to the spatial holographic sand table for display;

[0078] (5.2) Utilizing VR / AR equipment and combined with gesture commands, the three-dimensional morphological schemes of old neighborhoods are displayed, compared, selected, and modified. The local spatial structure, functional layout, road system, and blue-green system are adjusted and interactively optimized, and then output in the form of a SketchUp three-dimensional model file. The VR / AR equipment includes VRTRIX data gloves, VR glasses, and VR 3D display stands.

[0079] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention, and such changes and modifications fall within the scope of the invention as claimed.

Claims

1. An intelligent optimization method for updating the three-dimensional form of old blocks based on rigid-flexible rule driving is characterized by: The following steps are involved: (1) Collect vector data of the target old block, text data of the design specifications and standards of the city where it is located, and image data of design plans for other blocks in the city; (2) Extract control terms and specific control contents from the design specifications and standard text data of the city, and use the extension element model method to perform the basic structured representation of "matter element-event element-relation element"; From the association relationship of the statements, the primitive models are classified and stored according to the conjugation relationship, complementary relationship, constraint relationship and suggestion relationship, and a rule map of the rigid industry specifications of urban design is established; (3) Use topological graph structure and three-dimensional space to perform correlation calculations and extract professional knowledge rules in urban design space aesthetics; use Mask R-CNN coding technology to automatically annotate the spatial structure, functional layout, road system, and blue-green system vector data of classic urban design results, form a knowledge map of classic urban design results rules, and establish a rule map guided by flexible professional knowledge of urban design; (4) Establish a multi-layer network association between the old block three-dimensional morphological graph network, the rigid rule graph, and the flexible knowledge graph, so that the old block three-dimensional morphological graph network is gradually iterated until the constraints of the rigid rule graph and the flexible knowledge graph are met, and the solution is output; (5) Output the optimized three-dimensional morphological scheme of the target old block to the spatial holographic sand table, conduct three-dimensional multi-scheme display and interactive optimization of the block, and finally output the optimized three-dimensional morphological network and visualization scheme of the old block; In step (3), the topological graph structure is used to perform correlation calculation with the three-dimensional space to extract professional knowledge rules in the urban design space aesthetics, specifically including: Aerial photos of urban space in classic urban areas around the world were collected as a sample library. These photos were segmented and identified using semantic segmentation and recognition technology, with image elements identified as five categories: buildings, roads, squares, green spaces, and water systems. These elements were then abstracted into a topological graph structure consisting of nodes and edges. A multi-level graph neural network deep learning model was used to process the topological graph structure, capturing patterns and information in the graph by learning the relationships between nodes. This capture effect was improved through a hierarchical learning structure. Spatial self-learning technology was used to calculate quantitative indicators of the spatial agglomeration, openness, and staggeredness of the case blocks, and these indicators were correlated with the topological graph structure to extract professional knowledge and rules in urban design space aesthetics, forming a knowledge graph of spatial aesthetic rules. In step (3), the Mask R-CNN coding technology is used to automatically annotate the spatial structure, functional layout, road system, and blue-green system vector data of the design, forming a knowledge graph of classic urban design results rules and establishing a rule graph required by flexible industry specifications for urban design, specifically including: Collect classic urban design achievements, use convolutional neural networks to identify design scheme image data of the classic urban design achievements, extract them as two-dimensional vector graphic data of urban design plan drawings, and use Mask R-CNN encoding technology to annotate the spatial structure, functional layout, road system, and blue-green system vector data of the design scheme; wherein, the spatial structure vector data includes the geometric center of each functional core; the functional layout vector data includes the functional layout data of commercial, office, residential, industrial, green space, public service and transportation; the road system vector data includes three-level vector information of main roads, secondary roads and branches; the blue-green system vector data includes vector data information of green corridors, secondary green corridors and water areas; reversely analyze the master styles of urban design in various classic works through knowledge graph technology to form a knowledge graph of classic design rules; combine the knowledge graph of spatial aesthetic rules to form a rule graph required by flexible industry specifications for urban design; The step (4) establishes a multi-layer network association of the old block three-dimensional morphological graph network, the rigid rule graph and the flexible knowledge graph, specifically including: Based on the vector data of the target old blocks, a three-dimensional morphological map network of the old blocks is established. Combined with the rule map required by the rigid industry specifications for urban design and the rule map required by the flexible industry specifications for urban design, the three-dimensional morphological map network of the old blocks is gradually iterated according to the site location of the design scheme until the constraints of the rigid rule map and the flexible knowledge map are met, and the rule and knowledge parameter data of the potential urban design scheme are output; on this basis, the vector data associated with the potential scheme in the sample library are screened out, thereby establishing an automatic association process between knowledge rules and physical space elements; wherein, the screening process adopts the gradient descent method to observe the fluctuation of the loss function of the convolution pooling layer of the corresponding model of each parameter during training, and compare the training time and generation results to determine the optimal learning rate and number of iterations until the optimal is achieved, and finally a multi-layer network association of the three-dimensional morphological map network of the old blocks, the rigid rule map and the flexible knowledge map is established.

2. The method for intelligently updating and optimizing the three-dimensional form of old blocks based on rigid-flexible rule driving according to claim 1 is characterized in that: The step (1) involves collecting vector data of the target old block, text data of the design specifications and standards of the city, and image data of design plans for other blocks in the city, specifically including: Based on the scope of the identified target old blocks, obtain block vector data of the city where the target old blocks are located through the OpenStreetMap open data platform, and use the Baidu open data platform to obtain the city's design specifications and standard text data and image data of other block design plans in the city; The block vector data of the city where the target old block is located includes plot data and building data in shapefile format; the plot data (shp file) should include land use type information, geographic coordinate information, plot area information, and plot shape information; the building data (shp file) should include building function information, building number of floors information, building height information, building shape information, and building location information; Design specifications and standard text data for the city in question, including urban land classification and planning and construction land standards, urban public facilities planning specifications, urban road traffic planning and design specifications, urban environmental sanitation facilities planning specifications, urban residential area planning and design specifications, urban engineering pipeline comprehensive planning specifications, urban environmental planning standards, urban water supply engineering design specifications, urban drainage engineering design specifications, urban road greening planning and design specifications, and target city planning technical standards and guidelines; Image data of design plans for other blocks in the city, including design drawings for block stock renewal, environmental improvement of old blocks, optimization of public facilities and green spaces, and protection and renewal of urban landscape.

3. The method for intelligently updating and optimizing the three-dimensional form of old blocks based on rigid-flexible rule driving according to claim 2 is characterized in that: In step (2), control terms and specific control contents are extracted from the design specifications and standard text data of the city, and a basic structured representation of "matter element-event element-relation element" is performed using the extension primitive model method, specifically including: Use the collected design specifications and standard text data from the cities where the old blocks are located to extract control terms and specific control content, including road network control rules, open space control rules, height form control rules, and environmental and facility control rules; The extension element model method refers to the use of formal language to extract control terms and specific control contents from the design specifications and standard text data of the city. Its logical unit is the basic element, including matter element, event element and relationship element. The matter-element M that describes things is expressed as: M=(O m ,c m ,in m ) Among them, O m Represents an object, c m Indicates the name of the feature, v m Indicates O m About c m the value of the quantity taken; The expression of event element A describing the interaction between objects is: A=(O a ,c a ,in a ) Among them, O a Represents an object, c a Indicates the name of the feature, v a Indicates O a About c a the value of the quantity taken; The relational element R that describes the relationship between features is expressed as: Among them, O r Indicates relationship, c r1 , c r2 ,...c rn represents n features, v r1 , v r2 ,...v rn Indicates the corresponding value.

4. The method for intelligently updating and optimizing the three-dimensional form of old blocks based on rigid-flexible rule driving according to claim 3 is characterized in that: In step (2), the primitive models are classified and stored based on the conjugate relationship, complementary relationship, constraint relationship and suggestion relationship from the association relationship of the statements, and a rule map of the rigid industry specification requirements of urban design is established, which specifically includes: Analyze the relationships among the matter-elements, primitives, and relational elements that have been represented in a basic structured manner, and enter them into the NEO4J platform according to four categories: conjugate relationships, complementary relationships, constraint relationships, and recommended relationships. Use the NEO4J platform to digitally encode the rigid industry specifications for urban design that have been represented in a structured manner, and use algorithms to visualize the primitive relationships and generate a rule map. We use large language models (LLama-7B, chat-GLM, and Bloom) to fine-tune language constraint models.

5. The method for intelligently updating and optimizing the three-dimensional form of old blocks based on rigid-flexible rule driving according to claim 4 is characterized in that: The step (5) of outputting the optimized three-dimensional morphological network and visualization scheme of the old blocks specifically includes: The three-dimensional morphological schemes of the old blocks generated by the layer-by-layer iteration of the simulation, which meet both the rigid rule graph and the flexible knowledge graph, are output to the spatial holographic sandbox for display; with the assistance of VR\AR equipment and combined with gesture instructions, the three-dimensional morphological schemes of the old blocks are displayed, compared, selected, and modified, and the local spatial structure, functional layout, road system, and blue-green system are adjusted and interactively optimized before being output in the form of a three-dimensional model file of SketchUp; the VR\AR equipment includes VRTRIX data gloves, VR glasses, and VR 3D display stands.

Images