An Automatic Design Method and System for Urban Open Spaces Based on Multi-Agent Systems

Abstract

This invention discloses an automatic design method and system for urban open spaces based on multi-agent intelligence, relating to the field of urban planning technology. The method includes the following steps: constructing an urban open space database containing physical space data and an urban open space layout specification library containing design specification data; constructing an urban open space case library using a clustering algorithm; constructing urban open space element parameters for the site to be designed using the entropy weight method; automatically generating urban open spaces using a multi-agent intelligence algorithm; verifying the generated results using a human-computer interaction verification platform; and providing interactive display and printout. This invention effectively saves time and manpower costs in urban open space design by identifying and acquiring urban open space elements through artificial intelligence algorithms, and effectively improves the objectivity of urban open space design by generating and verifying urban open space schemes through a multi-agent intelligence algorithm.

Description

Technical Field

[0001] This invention relates to the field of urban planning technology, and specifically to an automatic design method and system for urban open spaces, and more particularly to an automatic design method and system for urban open spaces based on multi-agent systems. Background Technology

[0002] Currently, there are two main approaches to urban open space design. One relies on manual design by professional designers or teams, who determine design schemes based on personal experience and aesthetic viewpoints through site visits, observations, and analysis. This method requires significant human involvement and time in the early stages of design and is highly dependent on the designer's accumulated knowledge and experience, thus possessing a degree of subjectivity and experientialism. The other approach relies on technology and data, constructing a quantitative evaluation system by collecting relevant indicators. This method often emphasizes large-scale spatial functions and structures while neglecting small-scale characteristics perceived by people, such as the comfort, atmosphere, and appearance of the open space. Therefore, current urban open space design methods fail to comprehensively capture the multi-dimensional characteristics of urban open spaces and struggle to accurately identify their potential advantages and disadvantages, resulting in design schemes that do not align with people's actual needs and trends regarding open spaces. Summary of the Invention

[0003] Purpose of the Invention: To address the shortcomings mentioned in the background art, the purpose of this invention is to provide an automatic design method and system for urban open spaces based on multi-agent intelligence, which addresses the problems of complex and difficult-to-obtain open space elements and strong design subjectivity in the field of urban planning. This invention uses artificial intelligence algorithms to identify and obtain urban open space elements, effectively saving time and manpower costs in urban open space design. By using multi-agent intelligence algorithms to generate and verify urban open space schemes, the objectivity of urban open space design is effectively improved.

[0004] Technical Solution: To achieve the above-mentioned objectives, the present invention adopts the following technical solution: an automatic design method for urban open spaces based on multi-agent systems, comprising the following steps:

[0005] Establish a database of open spaces in target cities and a database of urban open space layout standards;

[0006] Collect data on open space design schemes in case study cities and construct a case study library of urban open spaces. Cluster the case study library according to city type, element type, morphological index and spatial layout parameters, and enter the clustering results into the case study library of urban open spaces.

[0007] The potential parameters of urban open space are set, and the potential of urban open space elements is intelligently determined using the entropy weight method. Based on the potential determination, the initial position of urban open space is set. Combined with the multi-agent algorithm, the layout and movement rules of urban open space are set, and the process is iterated to generate point elements and line elements of urban open space and form an automatic design result.

[0008] As a preferred approach, the clustering method for the case city open space design scheme data is as follows: the case city open spaces are divided according to city type, element type, morphological index, and spatial layout. A set of structural elements of the city open space, including three types of parameters such as element type, morphological index, and spatial layout, is constructed, which includes point elements and line elements. Landscape connectivity and Shannon index are calculated, and K-means clustering is performed on the index calculation results to output the clustering results and obtain open space systems with different morphological structures.

[0009] Preferably, the potential parameters include recreational attractiveness, site development potential, transport accessibility, and development suitability; the formula for calculating recreational attractiveness is as follows:

[0010] Vatre(in) = S(in) * Ve(in)

[0011] Vatra(in) = S(in) * Va(in)

[0012] Where S(in) represents the area of ​​plot in, Ve(in) and Va(in) are the recreational service importance of plot in, and Vatre(in) and Vatra(in) are the ecological recreational attraction and cultural recreational attraction of plot in.

[0013] The formula for calculating the development potential of a land parcel is as follows:

[0014] Vdp(in) = S(in) * Vd(in)

[0015] Wherein, Vd(in) is the standard value of the development potential of plot in, and Vdp(in) is the development potential of plot in;

[0016] The formula for calculating accessibility is as follows:

[0017]

[0018] Where Rl(in) represents the standardized value of the average path distance of plot in, Rd(in) is the standardized value of the road density around plot in, and V ta (in) represents the accessibility of land parcel in by public transport.

[0019] The formula for calculating suitability is as follows:

[0020]

[0021] Among them, V cs (in) represents the suitability of land parcel in for construction, Vp(in) is the publicness score of land parcels surrounding land parcel in, and n is the number of surrounding land parcels.

[0022] As a preferred option, the initial number of locations for urban open spaces is calculated using the following formula:

[0023]

[0024] Where N i P represents the initial number of placement positions for the i-th open space type. 单元人口 S represents the number of people served by the city unit. 平均需求 S represents the minimum per capita area for various types of open spaces as defined in the regulations. i This represents the minimum area for each type of open space.

[0025] As a preferred option, the initial location setting rules for each open space intelligent agent are as follows: call the road network and land use data of the city open space database, and divide the target city land into multiple functional units based on the main roads;

[0026] Comprehensive Park C1: This involves ranking the recreational attractiveness data of various plots within the city's functional units. Recreational attractiveness is the sum of ecological and cultural recreational attractiveness. The top N retrieved data are considered. C1 The high-value points of the position are obtained into a set E. The distance between any two points in this set is greater than the minimum radius of C1. Each point in set E is the initial position of C1.

[0027] City-level square C2: Divide the plots into categories [1.5St, 1), [0.5St, 1.5St), [-0.5St, 0.5St), and (0, -0.5St) according to one standard deviation St of the plot's construction potential. Take the road intersections in [1.5St, 1) as the core potential points of the functional structure, and put them into set F in descending order. The distance between the newly added point and all the previous points must be greater than or equal to the preset distance threshold. If the requirement is not met, continue to the next order for judgment. Finally, the core set of the functional structure is obtained, and a C2 is placed in each core point.

[0028] P1~P4: Based on the population density and land area, calculate the population size and the required number N of community park P1, neighborhood park P2, small park P3, and street square P4 within the functional unit. P1 N P2 N P3 N P4The initial position is randomly assigned to a functional unit road intersection. The probability is determined by the facility density of each intersection, and only one agent of the same type can be generated at each intersection.

[0029] Preferably, the urban open space layout movement rules include agent movement rules and agent iteration rules;

[0030] The agent's movement rules are determined by gravity. Repulsive force The agent's supply saturation rate Rss is a prerequisite for determining the type of force acting on the agent, and its calculation method is as follows:

[0031]

[0032] In the formula, P s P represents the actual population served by the open space. SMAX This represents the maximum service population of the open space. The actual service population is the ratio of the residential land area within the open service space to the per capita residential land area.

[0033] For any agent i, if the agent's supply saturation rate Rss < 100%, and the vector magnitude is... Then we have:

[0034]

[0035] In the formula, q is the gravitational parameter. The location of open space corresponding to agent i, x Dsdmax The intersection with the highest service density within the open space service area. Let i be the vector pointing from agent i to the intersection;

[0036] For any agent i, if And i≠j,

[0037] and

[0038] Then there is

[0039] In the formula, The intelligent agent i, representing community park P1, neighborhood park P2, small park P3, and street square P4 types, experiences a repulsive force. Let be the vector pointing from agent j to agent i. Let the magnitude of the vector be . Let p be the service radius of agents i and j, and p be the repulsive force parameter.

[0040] The agent iteration formula is PT+1 = f(PT, NP, NE), where PT+1 is the state of the detected agent at time T+1; PT is the state of the detected agent at time T; NP is the state of the agent's neighboring agents; and NE is the state of the agent's neighboring environmental elements. The direction of the resultant force on each agent is determined. When the average displacement of three consecutive actions of the agent is less than the set displacement threshold and the service coverage of each type of agent in the functional unit reaches the set threshold, the latest position of all open space points is output.

[0041] Preferably, the urban open space line elements include skeleton network generation and branch element generation. The skeleton network generation method is to input any clustering result based on the urban open space case library into the geographic information platform, and generate the optimal path between the comprehensive park C1, the city-level square C2 and the key urban open space element Q1 according to the pathfinding algorithm to generate the skeleton element H. The branch element generation method is to generate the optimal branch path between the generated skeleton element H, community park P1, neighborhood park P2, small park P3 and street square P4 according to the latest location of the generated skeleton element H, community park P1, neighborhood park P2, small park P3 and street square P4 according to the pathfinding algorithm in the geographic information platform, and output the generation result as branch element J.

[0042] Preferably, the method further includes verifying the automatic design results: for all nodes connected to the network system, it is determined whether the coverage rate of service objects within a preset range is greater than a preset coverage threshold. For non-compliant schemes, the generation and iteration continue until the rule conditions are met. The skeleton elements H and branch elements J generated for schemes that meet the rule conditions are expanded and input into the urban open space basic database for storage.

[0043] Preferably, the method also includes revising the design results and displaying and outputting the open space layout scheme; the revision method is as follows: using the feature merging tool on the geographic information platform, the latest locations of the comprehensive park C1 and the city-level square C2, the latest locations of the community park P1, the neighborhood park P2, the small park P3 and the street square P4, the expanded skeleton element H and the branch element J are merged, and the output file is generated; the approxpolydp function tool is used for morphological correction, and the data is input into the urban open space basic database for storage;

[0044] The display and output include displaying and interacting with the verified automatic design results through a 3D holographic sand table, and integrating and printing the scheme drawings and models through a 3D printer.

[0045] Based on the same inventive concept, this invention provides an automatic design system for urban open spaces based on multiple agents, including: a basic library construction module, used to establish a database of target urban open spaces and a library of urban open space layout specifications;

[0046] The case study clustering module is used to collect data on urban open space design schemes and construct an urban open space case study library. It performs clustering identification on the case study library according to city type, element type, morphological index and spatial layout parameters, and records the clustering results into the urban open space case study library.

[0047] The module for generating urban open spaces is used to set potential parameters for urban open spaces and to intelligently determine the potential of urban open space elements using the entropy weight method. Based on the potential determination, the initial position of the urban open space is set, and combined with the multi-agent algorithm, the layout and movement rules of the urban open space are set, and the process is iterated to generate point elements and line elements of the urban open space and form an automatic design result.

[0048] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0049] 1. Improved Work Efficiency. This invention utilizes artificial intelligence algorithms to achieve more objective and accurate open space design, significantly improving design efficiency. It represents an update and iteration of urban element generation technology under the new technological background, addressing the problems of previous open space design methods that relied on qualitative judgment, were highly subjective, required significant manpower, and were inefficient.

[0050] 2. The method is innovative. Utilizing a multi-agent algorithm, open spaces are treated as agents, and the attractive and repulsive forces of surrounding elements are calculated. A pathfinding algorithm then forms point-like and linear open spaces. This approach demonstrates a certain forward-looking perspective in the field of automatic urban planning generation.

[0051] 3. Improved design accuracy and objectivity. This invention sets the attributes and generation iteration rules of open spaces based on design specifications, and uses design standards to perform secondary verification of the generated open space schemes. Compared with manual design, this effectively improves the accuracy of open space scheme design. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a flowchart of a method according to an embodiment of the present invention.

[0054] Figure 2 This is an example diagram of the open space generation result of point elements in an embodiment of the present invention.

[0055] Figure 3 This is an example diagram of the skeleton element generation result in an embodiment of the present invention.

[0056] Figure 4 This is an example diagram showing the generation result of branch elements in an embodiment of the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] like Figure 1 As shown in the figure, this invention discloses an automatic design method for urban open spaces based on multi-agent systems, which mainly includes the following steps:

[0059] Obtain current status data of open spaces in the target city and establish a database of open spaces in the target city. Collect legal documents for the design of urban open spaces and establish a database of urban open space layout standards.

[0060] Data on open space design schemes in case study cities were collected and a case study database of urban open spaces was constructed. The case study database was clustered according to four parameters: city type, element type, morphological index, and spatial layout. The clustering results were then entered into the case study database of urban open spaces.

[0061] By setting potential parameters for urban open spaces and combining them with the entropy weight algorithm, the potential of urban open space elements is intelligently determined. Based on this potential determination, the initial locations of urban open spaces are set. Then, using a multi-agent algorithm, the layout and movement rules of the urban open spaces are defined, and iterative processes are performed to generate point and line elements of the urban open spaces, resulting in an automated design.

[0062] Construct a digital verification system to check and correct the results of automated design.

[0063] The automated design results after verification are displayed and interacted with through a 3D holographic sand table, and the scheme drawings and models are integrated and printed through a 3D printer.

[0064] The technical solution of this invention will be described in detail below, taking Nanqiao District of Chuzhou City as an example.

[0065] Data on land use, urban road network, and current status of urban open space in Chuzhou City were obtained using remote sensing data from the ZY-3 mapping satellite.

[0066] Obtain the three-dimensional vector data and urban open space design data of the target city from the Chuzhou Municipal Planning Department, and input them into the geospatial system to construct the Chuzhou Municipal Urban Open Space Database and input them into the geographic information platform;

[0067] The "Guidelines for Urban Design of Territorial Spatial Planning" (a legal document for urban open space design) was collected from government websites, and a database of Chuzhou City's open space layout standards was established through a text-to-language digital translation system.

[0068] The open spaces of the case study cities are classified according to city type, element type, morphological index, and spatial layout. The classification rules refer to the "Urban Green Space Classification Standard" (CJJ / T85-2017), "Urban Residential Area Planning and Design Code" (GB50180-2018), and "Urban Road Traffic Planning and Design Code" (GB50220-1995), as shown in Table 1.

[0069] Table 1: Structural Elements of Urban Open Space

[0070]

[0071] In this embodiment, for any case city, the number n of park and plaza elements is calculated. Landscape connectivity and Shannon index are input into the data calculation software platform for calculation. The calculation results of the above indicators are then subjected to K-means clustering in the geographic information platform. In this embodiment, the clustering results are (L1~L4), corresponding to the morphological structures of balanced network, main axis branch, ring branch, and core radial. The above identification results and clustering results are jointly entered into the geographic information platform to construct the urban open space case library module. A spatial link is established between the Ulrich Holographic Sand Table (HT-D) and the urban open space database.

[0072] Potential parameters were set, referring to the "Guidelines for High-Quality Development of Pedestrian Streets", which included four indicators: recreational attractiveness, land development potential, transportation accessibility, and construction suitability.

[0073] The formula for calculating recreational attractiveness is as follows:

[0074] Vatre(in) = S(in) * Ve(in)

[0075] Vatra(in) = S(in) * Va(in)

[0076] Where S(in) represents the area of ​​plot in, Ve(in) and Va(in) are the recreational service importance of the plot, Vatre(in) and Vatra(in) are the ecological recreational attraction and cultural recreational attraction of plot in.

[0077] The formula for calculating the development potential of a land parcel is as follows:

[0078] Vdp(in) = S(in) * Vd(in)

[0079] Where S(in) represents the area of ​​plot in, Vd(in) is the standard value of the development potential of plot in, and Vdp(in) is the development potential of plot in.

[0080] The formula for calculating accessibility V is as follows:

[0081]

[0082] Where Rl(in) represents the standardized value of the average path distance of plot in, Rd(in) is the standardized value of the road density around plot in, and V ta (in) represents the accessibility of land parcel in.

[0083] The formula for calculating the suitability assessment (Vta) is as follows:

[0084]

[0085] Among them, V cs (in) represents the suitability for construction of plot in, Vp(in) is the public accessibility score of plot in surrounding plots, and n is the number of surrounding plots. The standardized calculation formulas for the above indicators are as follows:

[0086]

[0087] Where V(i) is the actual value of the V index for element i, and V_min and V_max are the minimum and maximum values ​​of the V index within the statistical range, respectively. 标准化 (i) is the normalized value of the V index of element i.

[0088] In this embodiment, the parameter setting calculation results are shown in the table below (only the first 5 elements are shown):

[0089] Table 2: Calculation Results of Potential Parameters

[0090] Element number Recreational appeal Land development potential Transportation accessibility Construction suitability 1 0.64 0.72 0.47 0.81 2 0.77 0.84 0.55 0.62 3 0.57 0.78 0.64 0.58 4 0.73 0.28 0.91 0.11 5 0.68 0.27 0.5 0.75

[0091] Using the entropy weighting algorithm, the weights of urban open space elements are intelligently determined. Based on the "Urban Physical Examination and Evaluation Regulations for Territorial Spatial Planning," the weights for recreational attractiveness, land development potential, transportation accessibility, and construction suitability are determined in this embodiment to be 0.24, 0.11, 0.18, and 0.47, respectively. After calculating the weighted indicators, the top 50% of urban open spaces are designated as key elements (Q1), and the bottom 50% as general elements (Q2). These are then imported into the Chuzhou City Urban Open Space Database for storage.

[0092] Set the main parameters of the intelligent agent, including setting urban open space point elements with reference to the Chuzhou City Open Space Layout Specification Library, and the number of initial placement locations N. i .

[0093] The open space types are core elements C1 (comprehensive park) and C2 (city-level square) and general elements P1 (community park), P2 (neighborhood park), P3 (pocket park), and P4 (street square). The initial number of placement locations is N. i The formula is as follows:

[0094]

[0095] Where N i For open space type (i = core elements C1, C2 and general elements P1, P2, P3, P4), P 单元人口 S represents the number of people served by the city unit. 平均需求 The minimum per capita area for various types of open spaces as defined in the "Classification Standard for Urban Green Space" (CJJ / T85-2017) and the "Code for Planning and Design of Urban Residential Areas" (GB50180-2018) is S. i The minimum area (㎡) for each type of open space is defined in this embodiment. C1 (comprehensive park) is 100,000㎡, C2 (city-level square) is 500㎡, P1 (community park) and P2 (neighborhood park) are 10,000㎡, and P3 (small park) and P4 (street square) are 4,000㎡.

[0096] Define the initial position setting rules for each open space agent:

[0097] Using road network and land use data from the urban open space database, and based on main roads, the design site is divided into multiple functional units using a tiling tool in the geographic information platform. In this example, these units are named (I1~I...). 33 ).

[0098] C1 (Comprehensive Park): Located within the city's functional unit I n The recreational attractiveness (the sum of ecological and cultural recreational attractiveness) data of various plots within the area were sorted, and the top N data were retrieved. C1 The high-value points of position are obtained into the set E{e1,e2,e...} 13 ,…e N The distance between any two points in set E is greater than the minimum radius of C1, and each point in set E is the initial position of C1.

[0099] C2 (Urban Plaza): Based on one standard deviation (St) of the plot's development potential, it is divided into categories [1.5St, 1), [0.5St, 1.5St), [-0.5St, 0.5St), and (0, -0.5St). Road intersections within [1.5St, 1) are considered core potential points for the functional structure, and these are placed into set F in descending order of potential. The distance between a newly added point and all previous points must be greater than or equal to 1000m. If this requirement is not met, the process continues to the next position in the sequence. The final set of core potential points for the functional structure is F{f1, f2, f3, ... f... N}, and place a public activity-type core element C2 at each core point.

[0100] P1~P4: Based on the population density and land area, calculate the population size and the required number N of P1 (community park), P2 (neighborhood park), P3 (small park), and P4 (street square) within this functional unit. P1 N P2 N P3 The agent is randomly placed at the road intersection of the functional unit as its initial position. The probability of this is determined by the facility density Dsd of each intersection, and only one agent of the same type can be generated at each intersection.

[0101] Establish rules for the layout and movement of urban open spaces, including rules for agent movement and rules for agent iteration.

[0102] The agent's movement rules are determined by gravity and repulsion, with gravity expressed as... Repulsive force is represented as

[0103] The agent's supply saturation rate Rss is a prerequisite for determining the type of force acting on the agent, and its calculation method is as follows:

[0104]

[0105] In the formula, P s P represents the actual population served by the open space. SMAX This represents the maximum service population of an open space. The actual service population is the ratio of the residential land area within the open service space to the per capita residential land area.

[0106] The numerical calculation method for gravity is as follows: for any agent i, if Rss < 100%, and the vector magnitude is... Then there is

[0107]

[0108] In the formula, The intelligent agent i is subject to gravity, which includes gravity related to transportation accessibility and gravity related to construction suitability. Let q be the vector pointing from agent i to the object exerting the force, and let q be the gravity parameter, which is 1. The location of open space corresponding to agent i, x Dsdmax The intersection with the highest service density within the open space service area. Let i be the vector pointing from agent i to the intersection.

[0109] The repulsive force calculation method is as follows: for any agent i, if And i≠j, and

[0110]

[0111] Then there is

[0112] In the formula, A P-type (P1-P4) agent i experiences a repulsive force. Let be the vector pointing from agent j to agent i. Let the magnitude of the vector be . Let be the service radius of agents i and j, and p be the repulsion parameter, which is set to 1.

[0113] The movement distance is the distance that agent n moves along the direction of the vector by the magnitude of the movement distance, and the movement distance is set to 10m for each iteration.

[0114] The agent iteration rules are implemented in the PyCharm platform.

[0115] The iterative formula is PT+1 = f(PT, NP, NE).

[0116] PT+1 represents the state of the detected agent at time T+1; PT represents the state of the detected agent at time T; NP represents the state of the neighboring agents of this agent; NE represents the state of the neighboring environmental elements of this agent. The direction of the resultant force acting on each agent is determined. When the average displacement of three consecutive actions of an agent is less than 20m and the service coverage rate of each type of agent within the unit reaches 85%, the latest positions of all open space points are output and stored in the Chuzhou City Open Space Basic Database. The generated results are as follows: Figure 2 .

[0117] The generation of urban open space line elements includes skeleton network generation and branch elements:

[0118] To generate the skeleton network, the clustering result L1 is input into the geographic information platform. Based on the pathfinding algorithm, the optimal path is generated between C1 (comprehensive park), C2 (city-level square), and the key element Q1 obtained from the entropy weight method weighted index, thus generating the skeleton element H. The final result is as follows: Figure 3 .

[0119] Branch elements are generated. In the geographic information platform, a pathfinding algorithm is used to generate the optimal branch path by comparing the generated skeleton element H with the latest positions of other generated nodes (P1-P4). The generated result is output as branch element J and stored in the Chuzhou City Open Space Basic Database.

[0120] The automatic design results are verified, ensuring that all nodes are connected to the network system and that the coverage of service objects within 300m is greater than 90%. For non-compliant solutions, the process continues to generate iterations until the rules are met.

[0121] In this embodiment, the skeleton element H and branch element J generated by the scheme are widened. According to the "Urban Green Space Classification Standard", the skeleton element H is widened by 9m and the branch element J is widened by 4m in the geographic information platform, and then stored in the Chuzhou City Open Space Basic Database. The generated result is as follows. Figure 4 .

[0122] The design results are revised and open space layout scheme drawings are output. The scheme revision method involves using the feature merging tool on the geographic information platform to merge the latest positions of the core features (C1-C2) and other generated nodes (P1-P4).

[0123] In a further embodiment, the broadened skeleton element H and branch element J are merged and output as a shapefile. The approxpolydp function is used for shape correction. The result is then stored in the Chuzhou City Open Space Basic Database.

[0124] The method for outputting the open space layout scheme drawing is as follows: In the geographic information platform, the generated key element Q1 is assigned a color with R=0, G=119, B=19; the general element Q2 is assigned a color with R=40, G=145, B=114; C1~C2 is assigned a color with R=116, G=192, B=145; P1~P4 is assigned a color with R=164, G=205, B=111; the skeleton element is assigned a color with R=53, G=106, B=66; and the branch element is assigned a color with R=178, G=222, B=195. The open space layout scheme drawing is then output.

[0125] Furthermore, feedback on the construction and generation results of the scheme modification instruction library is displayed.

[0126] The scheme modification command library includes four types of operations: display, selection, invocation, and modification. It connects the command library with the prediction results and projects them onto a holographic sand table (HT-D). The result feedback display method adjusts geospatial information data using the command library, and displays the adjusted results in real time. The Chuzhou City open space layout scheme is output and printed using an industrial 3D printer.

[0127] Based on the same inventive concept, this invention discloses an automatic urban open space design system based on multi-agent intelligence, mainly comprising: a basic library construction module, used to establish a target city open space database and an urban open space layout specification library; a case library clustering module, used to collect case city open space design scheme data and construct an urban open space case library, clustering the case library according to city type, element type, morphological index, and spatial layout parameters, and recording the clustering results into the urban open space case library; and an urban open space generation module, used to set urban open space potential parameters, intelligently determine the potential of urban open space elements using the entropy weight method, set the initial position of urban open space based on the potential determination, combine the multi-agent algorithm to set the urban open space layout movement rules, iterate, generate urban open space point elements and line elements, and form an automatic design result. Specific implementation details of each module are given in the above method embodiments and will not be repeated here.

Claims

1. A method for automatic design of urban open spaces based on multi-agent systems, characterized in that, Includes the following steps: Establish a database of open spaces in target cities and a database of urban open space layout standards; Collect data on open space design schemes in case study cities and construct a case study library of urban open spaces. Cluster the case study library according to city type, element type, morphological index and spatial layout parameters, and enter the clustering results into the case study library of urban open spaces. The potential parameters of urban open space are set, and the potential of urban open space elements is intelligently determined using the entropy weight method. Based on the potential determination, the initial position of urban open space is set, and the layout movement rules of urban open space are set in combination with a multi-agent algorithm. The process is iterated to generate point elements and line elements of urban open space and form an automatic design result. The layout movement rules of urban open space include agent movement rules and agent iteration rules. The agent's movement rules are determined by gravity. Repulsive force Decision; Agent supply saturation rate The prerequisite for determining the type of force experienced by an intelligent agent is calculated as follows: ； In the formula, This represents the actual population served by the open space, and is the ratio of the residential land area within the open service space to the per capita residential land area. Indicates the maximum population that an open space can serve; For any agent i, if the agent's supply saturation rate And the vector magnitude Then we have: ； In the formula, q is the gravitational parameter. The location of the open space corresponding to agent i. The intersection with the highest service density within the open space service area. Let i be the vector pointing from agent i to the intersection; For any agent i, if ,and , and , Then there is ; In the formula, The intelligent agent i, representing community park P1, neighborhood park P2, small park P3, and street square P4 types, experiences a repulsive force. Let be the vector pointing from agent j to agent i. Let be the magnitude of the vector. Let p be the service radius of agents i and j, and p be the repulsive force parameter. The agent iteration formula is PT+1=f(PT, NP, NE), where PT+1 is the state of the detected agent at time T+1; PT is the state of the detected agent at time T; NP is the state of the agent's neighboring agents; and NE is the state of the agent's neighboring environmental elements. The direction of the resultant force on each agent is determined. When the average displacement of three consecutive actions of the agent is less than the set displacement threshold and the service coverage of each type of agent in the functional unit reaches the set threshold, the latest position of all open space points is output.

2. The automatic design method for urban open spaces based on multi-agent systems according to claim 1, characterized in that, The clustering method for the case city open space design scheme data is as follows: the case city open spaces are divided according to city type, element type, morphological index, and spatial layout. The structural element set of the city open space, including three types of parameters such as element type, morphological index, and spatial layout, is constructed, including point elements and line elements. Landscape connectivity and Shannon index are calculated, and K-means clustering is performed on the index calculation results. The clustering results are output to obtain the open space system with different morphological structures.

3. The automatic design method for urban open spaces based on multi-agent systems according to claim 1, characterized in that, The potential parameters include recreational attractiveness, site development potential, transport accessibility, and suitability for development; The formula for calculating recreational attractiveness is as follows: Vatre(in) = S(in) × Ve(in); Vatra(in) = S(in) × Va(in); Where S(in) represents the area of ​​plot in, Ve(in) and Va(in) are the recreational service importance of plot in, and Vatre(in) and Vatra(in) are the ecological recreational attraction and cultural recreational attraction of plot in. The formula for calculating the development potential of a land parcel is as follows: Vdp(in) = S(in) × Vd(in); Wherein, Vd(in) is the standard value of the development potential of plot in, and Vdp(in) is the development potential of plot in; The formula for calculating accessibility is as follows: ； Among them, Rl Rd(in) represents the standardized value of the average path distance to plot in, and Rd(in) is the standardized value of the road density around plot in. It refers to the accessibility of the land parcel by public transportation; The formula for calculating suitability is as follows: ； in, The suitability of the land parcel for construction. It is the publicness score of the land parcel in the surrounding land parcels, where n is the number of surrounding land parcels.

4. The automatic design method for urban open spaces based on multi-agent systems according to claim 1, characterized in that, The initial number of locations for urban open spaces is calculated using the following formula: ； Where N i P represents the initial number of placement positions for the i-th open space type. 单元人口 S represents the number of people served by the city unit. 平均需求 S represents the minimum per capita area for various types of open spaces as defined in the regulations. i This represents the minimum area for each type of open space.

5. The automatic design method for urban open spaces based on multi-agent technology according to claim 1, characterized in that, The initial location setting rules for each open space intelligent agent are as follows: call the road network and land use data of the city open space database, and divide the target city land into multiple functional units based on the main roads; Comprehensive Park C1: This involves ranking the recreational attractiveness data of various plots within the city's functional units. Recreational attractiveness is the sum of ecological and cultural recreational attractiveness. The top N retrieved data are considered. C1 The high-value points of the position are obtained into a set E. The distance between any two points in this set is greater than the minimum radius of C1. Each point in set E is the initial position of C1. City-level square C2: Divide the plots into categories [1.5St, 1), [0.5St, 1.5St), [-0.5St, 0.5St), and (0, -0.5St) according to one standard deviation St of the plot's construction potential. Take the road intersections in [1.5St, 1) as the core potential points of the functional structure, and put them into set F in descending order. The distance between the newly added point and all the previous points must be greater than or equal to the preset distance threshold. If the requirement is not met, continue to the next order for judgment. Finally, the core set of the functional structure is obtained, and a C2 is placed in each core point. P1~P4: Based on the population density and land area, calculate the population size and the required number N of community park P1, neighborhood park P2, small park P3, and street square P4 within the functional unit. P1 N P2 N P3 N P4 The initial position is randomly assigned to the road intersection of the functional unit. The probability is determined by the facility density of each intersection, and only one agent of the same type can be generated at each intersection. The formula for calculating recreational attractiveness is as follows: Vatre(in) = S(in) × Ve(in); Vatra(in) = S(in) × Va(in); Where S(in) represents the area of ​​plot in, Ve(in) and Va(in) are the recreational service importance of plot in, and Vatre(in) and Vatra(in) are the ecological recreational attraction and cultural recreational attraction of plot in. The formula for calculating the development potential of a land parcel is as follows: Vdp(in) = S(in) × Vd(in); Wherein, Vd(in) is the standard value of the development potential of plot in, and Vdp(in) is the development potential of plot in.

6. The automatic design method for urban open spaces based on multi-agent systems according to claim 1, characterized in that, The urban open space line elements include skeleton network generation and branch element generation. The skeleton network generation method is as follows: input any clustering result based on the urban open space case library into the geographic information platform, and generate the optimal path between the comprehensive park C1, the city-level square C2 and the key urban open space element Q1 according to the pathfinding algorithm to generate the skeleton element H. The branch element generation method is as follows: in the geographic information platform, the latest positions of the generated skeleton element H, community park P1, neighborhood park P2, small park P3 and street square P4 are used to generate the optimal branch path, and the generation result is output as branch element J.

7. The automatic design method for urban open spaces based on multi-agent systems according to claim 1, characterized in that, It also includes verifying the automatic design results: all nodes are connected to the network system, and it is determined whether the coverage of service objects within the preset range is greater than the preset coverage threshold. For non-compliant schemes, iterative generation continues until the rules are met; the skeleton elements H and branch elements J generated for schemes that meet the rules are expanded and input into the urban open space basic database for storage.

8. The automatic design method for urban open spaces based on multi-agent systems according to claim 1, characterized in that, It also includes the revision of the design results and the display and output of the open space layout scheme; the revision method is to use the feature merging tool on the geographic information platform to merge the latest locations of the comprehensive park C1 and the city-level square C2, the latest locations of the community park P1, the neighborhood park P2, the small park P3 and the street square P4, the expanded skeleton element H and the branch element J, and output the file; use the approxpolydp function tool to perform morphological correction, and input it into the urban open space basic database for storage; The display and output include displaying and interacting with the verified automatic design results through a 3D holographic sand table, and integrating and printing the scheme drawings and models through a 3D printer.

9. A multi-agent-based automatic design system for urban open spaces, used to implement the multi-agent-based automatic design method for urban open spaces according to any one of claims 1-8, characterized in that, include: The basic library construction module is used to establish a database of open spaces in the target city and a library of open space layout specifications for the city. The case study clustering module is used to collect data on urban open space design schemes and construct an urban open space case study library. It performs clustering identification on the case study library according to city type, element type, morphological index and spatial layout parameters, and records the clustering results into the urban open space case study library. The module for generating urban open spaces is used to set potential parameters for urban open spaces and to intelligently determine the potential of urban open space elements using the entropy weight method. Based on the potential determination, the initial position of the urban open space is set, and combined with the multi-agent algorithm, the layout and movement rules of the urban open space are set, and the process is iterated to generate point elements and line elements of the urban open space and form an automatic design result.

Images