An artificial intelligence-based automatic layout method for urban green space

Abstract

The application discloses an urban green space automatic layout method based on artificial intelligence, which comprises data acquisition and database construction, intelligent identification of urban green space elements, intelligent grading of urban green space elements, urban green space space attraction and repulsion iteration, online and offline intelligent verification of urban green space layout, automatic generation of urban green space layout scheme, and scheme auxiliary decision and holographic interactive display. The application can solve the problems in the field of urban planning and design, such as slow green space data collection, strong design subjective experience and difficult display interaction, and realizes automatic layout of urban green space, multi-scheme generation and interactive display through hardware devices by using artificial intelligence algorithm, thereby providing an efficient and convenient way for urban green space system planning, providing an auxiliary decision for managers, providing a design reference for planners, and providing a data sand table and three-dimensional display for public participation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of urban planning, in particular to a city green space automatic layout method and decision-making interactive system based on artificial intelligence. BACKGROUND

[0002] Green space system planning is an important part of urban planning, and reasonable green space system planning can effectively improve the urban ecological environment, optimize the urban microclimate and improve the livability of the city. Under the background of the national promotion of ecological civilization construction, the role of green space system planning has also been further valued to promote the construction of beautiful China and healthy China.

[0003] Green space system planning needs to combine green space landscape resource distribution and urban resident demand, and under the condition of green space development index, it needs to consider multiple possibilities of green space layout scheme. In the traditional design process, this work from green space development index to green space layout scheme is mostly completed manually by planning and design personnel, which needs to manually sort out a large amount of green space landscape resource data and consider multiple possibilities of green space layout scheme. The traditional design process has the following limitations:

[0004] Firstly, because the urban green space elements are complex and diverse in scale, a lot of effort is needed to complete the data collection, which has high labor cost, economic cost and time cost;

[0005] Secondly, the design scheme is highly subjective and has low design precision, which cannot accurately meet the relevant specification requirements of the "Urban Green Space Layout Specification".

[0006] Thirdly, the interactive display of previous special research often relies on the overall scheme of urban design, and the urban green space design scheme often lacks a separate interactive display and auxiliary decision-making platform.

[0007] With the gradual application of artificial intelligence technology in the field of urban planning, the automatic generation of green space layout scheme has been initially realized, but the standardization and landing of the generated scheme are limited, which is difficult to pass the examination of the "Urban Green Space Layout Specification" and cannot meet the actual needs of the current green space system planning. SUMMARY

[0008] To solve the problems mentioned in the background art, the purpose of the present application is to provide a city green space automatic layout method based on artificial intelligence, which can intelligently layout the city green space and make interactive decisions.

[0009] The purpose of the present application can be achieved by the following technical scheme: a city green space automatic layout method based on artificial intelligence, comprising the following steps:

[0010] Step S1: Collect satellite remote sensing data of the target city using the High Resolution 2 satellite with a resolution of 0.8m, and perform on-site verification and supplement. Obtain three-dimensional vector data of the target city from the local planning department, input into the geographic spatial system to build a database, and collect green space design specification files to build a specification library.

[0011] Step S2: Intelligent identification of urban green space elements is performed using the nearest neighbor rule classification algorithm (KNN). The intelligent identification of green space element features includes land use function characteristics, area characteristics, spatial distribution characteristics, and element type characteristics, including block element type and linear element type. The intelligent identification results are input into the geographic spatial database.

[0012] Step S3: Intelligent grading indicators are constructed for block element green space and linear element green space, respectively. The database and specification library are called to perform quantitative calculation and assignment of the grading indicators for green space blocks within the design range, and the indicator values are stored in the geographic information system.

[0013] Step S4: According to the "Urban Green Space Planning Standard GB / T 51346-2019" and landscape ecology, the grading standard table for block element green space and linear element green space is constructed. The block element green space within the design range is classified into three levels, and the linear element green space within the design range is classified into five levels.

[0014] Step S5: The classification results of linear element green space in step S4 are imported, and a three-level green corridor is generated using a pathfinding algorithm. The classification results of block element green space in step S4 are imported, and a multi-agent algorithm is used to generate and initially randomly arrange block green space, and if it meets the specifications, a preliminary green space layout scheme is generated and output.

[0015] Step S6: Green space element spatial repulsion calculation. The green space elements are classified by area size using the MATLAB shape recognition system. The green space element spatial repulsion calculation is performed using a multi-agent algorithm, and if it is compliant, it enters step S8, otherwise it enters step S7.

[0016] Step S7: Adjust the spatial repulsion of green space elements that violate the rules using a multi-agent algorithm. According to the size of the spatial repulsion, a spatial movement vector is assigned, and it enters step S6.

[0017] Step S8: Green space element spatial attraction calculation. The green space element spatial attraction calculation is performed using a multi-agent algorithm, and if it meets the threshold conditions, it enters step S10, otherwise it enters step S9.

[0018] Step S9: Adjust the green space elements that violate the spatial attraction by the multi-agent algorithm. Assign a spatial movement vector according to the size of the spatial attraction, and enter step S8.

[0019] Step S10: Check according to the urban green space layout specification. Establish a digital checking system for the "Urban Green Space Layout Specification" through a text language digitization translation system. The digital checking system checks the green space layout scheme and deletes the illegal scheme. Then, manually check the green space layout scheme and delete the illegal scheme.

[0020] Step S11: Select the green space layout scheme by expert scoring through an online judge scoring system. The full score is 10 points, and the integer is taken. Remove the two highest scores and the two lowest scores, and select the green space layout scheme with the highest average score as the final scheme.

[0021] Step S12: Realize the three-dimensional display of the green space layout scheme. Display the green space layout scheme generated in S11 in a three-dimensional platform with holographic display function.

[0022] Step S13: Construct a green space layout scheme auxiliary decision-making instruction library containing four types of operations: display, selection, call, and modification. And map it in the display device with interactive and display functions.

[0023] Step S14: Assist in decision-making for the green space layout scheme through 3D holographic projection. Use the decision-making instruction library constructed in step S13 to display, select, call, and modify the green space layout scheme.

[0024] Step S15: Realize the three-dimensional model of the green space layout scheme and the design manual printing. Integrate the data using data integration and translation equipment, print the three-dimensional model of the scheme using an industrial 3D printer, and print the green space layout scheme drawing into a design manual using a printing device.

[0025] Further, the step S1 comprises the following steps:

[0026] Step S1-1: Obtain geographic spatial information data and establish a database

[0027] Collect 4-band multispectral remote sensing images of the city area using a 0.8m resolution Gaofen-2 satellite. Take green space real scene pictures of the area by a WiFi version portable computer tablet with Beidou navigation system, check and supplement the satellite remote sensing images. Obtain three-dimensional vector data and urban green space planning scheme data of the city where the new city area is located from the local planning department.

[0028] Step S1-2: Obtain specification file data and establish a specification library

[0029] Collect the control detailed planning text and relevant legal specification files related to the design area, and establish the digital verification system of "Urban Green Space Planning Standard GB / T 51346-2019" through the text language digitization translation system.

[0030] The geospatial information data is generated after the unified city three-dimensional vector data is converted into WGS84 geographic coordinates, and includes four types of surface data, i.e. road blocks, water system blocks, green land blocks and other functional blocks, and two types of linear data, i.e. road center lines and block boundary lines, wherein the road center line is a continuous and non-closed line segment, and the block boundary line is a closed line segment.

[0031] Further, the step S2 comprises the following steps:

[0032] Step S2-1: calling a geospatial information database to build a geospatial digital sand table.

[0033] Step S2-2: matching the land use function characteristics of the green land blocks, wherein the land use function characteristics include water system block land use and green land block land use.

[0034] Step S2-3: using a computational geometry tool in the geographic information system to numerically calculate the area of the green land block, denoted as an area characteristic.

[0035] Step S2-4: using a Minimum Bounding Geometr tool in the geographic information system to build a minimum circumscribed rectangle of the green land block as a layout characteristic spatial unit, and recording all road data contained in the spatial unit, wherein the road data includes the orientation of the road to the green land block, the road name, the road length and the road grade.

[0036] Step S2-5: importing the minimum circumscribed rectangle of the green land block built in step S2-4, and if the length-width ratio of the block minimum circumscribed rectangle is less than 2, marking it as a point element type, and if the length-width ratio of the minimum circumscribed rectangle is greater than 2, marking it as a linear element type.

[0037] Further, the step S3 comprises the following steps:

[0038] Step S3-1: constructing an intelligent grading index a1 of the block element green land, and inputting the area of the green land block as the intelligent grading index a1.

[0039] Step S3-2: constructing an intelligent grading index a2 of the linear element green land, marking the green land block with an area greater than 50 hectares as a large ecological source point, calculating the distance between the centroid of the green land block and the centroid of the large ecological source point in the geographic information system after unifying the coordinate system as the intelligent grading index a2.

[0040] Step S3-3: Constructing the intelligent grading index a3 of linear element green space, inputting the road data in the space unit in step S2-4, and taking the highest level in the road grade as the intelligent grading index a3.

[0041] Step S3-4: Calculating the intelligent grading index value of the green land plot and embedding it in the green land plot in the form of an attribute table.

[0042] Further, the step S4 comprises the following steps:

[0043] Step S4-1: Intelligent grading of block element green space, extracting the intelligent grading index a1 data in step S3-1, and comparing with the park green space grading setting requirements in the "Urban Green Space Planning Standard GB / T 51346-2019" to divide the block green space into three levels: greater than 5.0 hectares, 1.0-5.0 hectares, and 0.2-1.0 hectares.

[0044] Step S4-2: Intelligent grading of linear element green space, extracting the distance from large ecological source point a2 data and road grade a3 data in steps S3-2 and S3-3, constructing a linear element green space grading standard table, and dividing the linear green space into five levels.

[0045] Further, the step S5 comprises the following steps:

[0046] Step S5-1: Calling the road data in the database, using the pathfinding algorithm (Dijkstra), and taking the linear green space graded in step S4 as the starting block S to automatically connect and generate a first-level green corridor, a second-level green corridor, and a third-level green corridor.

[0047] Step S5-2: Using the multi-agent algorithm, combining the block green space graded in step S4 and its service radius, and automatically generating and randomly arranging the block green space. If all green land plots meet the requirements of the "Urban Green Space Planning Standard GB / T 51346-2019", a preliminary green space layout scheme is generated and stored in the NAS network storage server.

[0048] Further, the step S6 comprises the following steps:

[0049] Step S6-1: Classification of green space elements by area size

[0050] The green space elements in the green space layout scheme output by step S5 are identified by the MATLAB shape recognition system. The green space elements are classified by geometric area size, including less than 2 hectares, 2-5 hectares, 5-20 hectares, and more than 20 hectares. The classification standard by area is from the area interval of different service type green spaces divided in the "Urban Green Space Layout Specification".

[0051] Step S6-2: Green space element same element space repulsion calculation

[0052] The spatial repulsion between the classified green space elements of the same type is calculated by the multi-agent algorithm. The calculation formula is Frep=H(S1+S2) / D, wherein Frep is the spatial repulsion between the same type elements, H is a constant, S1 and S2 are the geometric areas of the same type elements, and D is the geometric center distance between the same type elements. The direction of the spatial repulsion between the same type elements is on a straight line with the geometric center, and the size is proportional to the sum of the geometric areas of the same type elements and inversely proportional to the geometric center distance between the same type elements.

[0053] Step S6-3: Determine whether the spatial repulsion of the green space element of the same type is in compliance

[0054] The spatial repulsion of the green space element is compared with the threshold value, if both are less than the threshold value, step S8 is entered, otherwise, the green space element with the spatial repulsion greater than the threshold value is marked, and step S7 is entered. The threshold value is the spatial repulsion of the green space in the interval of the maximum service radius of 2 times specified in the "Urban Green Space Layout Specification".

[0055] Further, the step S7 specifically includes:

[0056] Step S7: Adjust the green space element with spatial repulsion violation

[0057] According to the spatial repulsion of the green space element marked in S6-3, a spatial movement vector is given by the multi-agent algorithm, and the optimized green space element is obtained, the green space layout scheme is updated, and S6 is entered.

[0058] Further, the step S8 includes the following steps:

[0059] Step S8-1: Green space element different element space attraction calculation

[0060] The spatial attraction between the classified green space elements of different types is calculated by the multi-agent algorithm. The calculation formula is Fattr=R(S1+S2) / D, wherein Fattr is the spatial attraction between different type elements, R is a constant, S1 and S2 are the geometric areas of different type elements, and D is the geometric center distance between different type elements. The direction of the spatial attraction between different type elements is on a straight line with the geometric center, and the size is proportional to the sum of the geometric areas of the same type elements and inversely proportional to the geometric center distance between the same type elements.

[0061] Step S8-2: Determine whether the spatial attraction of the green space element of different types is in compliance

[0062] Comparing the spatial attraction of green space elements with the size of the threshold value, if both are less than the threshold value, enter step S10; otherwise, mark the green space elements with a spatial attraction greater than the threshold value, and enter step S9. The threshold value is the sum of the spatial attraction of the two levels of green space when the maximum service radius specified in the Urban Green Space Layout Specification is reached.

[0063] Further, the step S9 specifically includes:

[0064] Step S9: Adjusting green space elements with spatial attraction violations

[0065] Through the multi-agent algorithm, the spatial movement vector is given according to the spatial attraction of the green space elements marked in S8-2, and the optimized green space elements are obtained, the green space layout scheme is updated, and S8 is entered.

[0066] Further, the step S10 specifically includes:

[0067] Step S10-1: Establishing a digital verification system for Urban Green Space Layout Specification

[0068] Establishing a text language digitization translation system to translate Urban Green Space Layout Specification into a rule language, and establishing a digital verification system for Urban Green Space Layout Specification;

[0069] Step S10-2: Digital verification system verifies green space layout scheme

[0070] Verify the green space layout scheme through the Urban Green Space Layout Specification digital verification system, and delete the green space layout scheme that violates the specification;

[0071] Step S10-3: Artificially checking green space layout scheme

[0072] Offline artificial verification of green space layout scheme according to Urban Green Space Layout Specification through geographic information platform database management system, and delete the green space layout scheme that violates the specification.

[0073] Further, the step S11 specifically includes:

[0074] Step S11: Selecting green space layout scheme through expert scoring

[0075] Select 20 experts who have worked in urban planning or landscape industry for more than 10 years, score the green space layout scheme through online judge scoring system, take the integer of the full score of 10, remove the two highest scores and two lowest scores, and select the green space layout scheme with the highest average score as the final scheme.

[0076] Further, the step S12 specifically includes:

[0077] Step S12: Three-dimensional display of green space layout scheme

[0078] The generated green space layout scheme is projected in a three-dimensional platform with holographic display function at a scale of 1:1000, and the green space layout scheme is displayed in all directions using 3D holographic projection. The equipment includes a VR panorama display platform and 3D tracking glasses. The display content includes a plan, important node plan, and three-dimensional holographic sand table.

[0079] Further, the step S13 specifically includes:

[0080] Step S13: Construction of green space layout scheme auxiliary decision-making instruction library

[0081] The auxiliary decision-making instruction library is connected with the green space layout scheme generated in S11 through four types of operations including display, selection, calling, and modification, and is mapped in a display device with interactive and display functions through a projection device.

[0082] Further, the step S14 includes the following steps:

[0083] The green space layout scheme generated in step S11 is imported into a city three-dimensional space digital model using a 3D holographic projection device, VR glasses, and virtual reality data gloves. The decision-making instruction library constructed in step S13 is used to display, select, call, and modify the green space layout scheme.

[0084] The city three-dimensional space digital model is generated after the unified city three-dimensional vector data is converted to the 2000 national geodetic coordinate system, and contains city geographic elevation, road network, city water system, and city mountain information.

[0085] The modification in the instruction library is the adjustment of the parameter values of the green space level and spatial distribution position. The green space level adjustment refers to the calculation of the green space levels L1, L2, and L3 by the computer through the grading index, and the calculation of the average error using a weight formula, and the computer automatically corrects the error. The green space layout adjustment refers to the calculation of the spatial distribution of each green space by the computer through the force iteration of attraction and repulsion using a multi-agent algorithm, and the adjustment of the force iteration interval by the user combined with the decision-making demand, and the computer automatically corrects the error.

[0086] Further, the step S15 includes the following steps:

[0087] The three types of data of green space layout plan, hierarchical index and weight coefficient are integrated by using data integration and translation equipment, and are displayed in a holographic sand table, a three-dimensional model of the scheme is printed by an industrial 3D printer, the scheme plan drawing of 1:1000 scale, the scheme bird's eye view of 1:1000 scale, the scheme node effect drawing of 1:500 scale, the scheme hierarchical index file and the scheme weight coefficient file are output by the drawing data integration equipment, and the above contents are printed into a design manual by the printing equipment.

[0088] Beneficial effects

[0089] 1. Process efficiency: the present application classifies and quickly generates the potential of urban green space through the nearest neighbor rule classification algorithm and pathfinding algorithm, iterates the force of attraction and repulsion of urban green space based on the multi-agent algorithm to optimize the layout, can generate multiple schemes in a short time and further filter effective schemes, can reduce the design time from at least three weeks to one day, and only one designer is needed to generate and filter multiple schemes instead of at least ten designers, effectively reducing time and labor costs and improving design efficiency.

[0090] 2. Scheme objectivity: the present application establishes a digital verification system of Urban Green Space Layout Specification through a text language digitization translation system, checks and verifies the green space layout scheme online and offline, can ensure the effectiveness of scheme generation, breaks through the expert judgment of traditional urban green space layout, avoids the uncontrollability of previous artificial intelligence urban design scheme generation, and directly selects effective multiple schemes instead of filtering effective schemes from several million schemes, which promotes the reliability and efficiency of scheme selection.

[0091] 3. Can provide new decision-making scenarios: the present application provides four types of operation instructions including display, selection, call and modification by constructing an auxiliary decision-making instruction library and applying 3D holographic projection, realizes interactive decision-making of urban green space layout scheme, and perfects the business scenario. The present application displays the plan, hierarchical index, weight coefficient, important node plan and three-dimensional holographic sand table of green space layout scheme through a three-dimensional platform with holographic display function, makes the urban green space display visual and perceptual, and improves the display effect. BRIEF DESCRIPTION OF DRAWINGS

[0092] Figure 1 is a method flowchart of the present application;

[0093] Figure 2 is a schematic diagram of the data acquisition equipment of step S1 of the present application;

[0094] Figure 3is a schematic diagram of the intelligent division and index process of step S3 of the present application;

[0095] Figure 4 is a schematic diagram of the stress iteration process of green space elements by multi-agent algorithm in steps S6-S9 of the present application;

[0096] Figure 5 is a design range map of the urban green space layout scheme of the present application;

[0097] Figure 6 is an automatic layout scheme map of the urban green space of the present application.

[0098] Figure 7 is a scheme interaction and printing schematic diagram of the present application. DETAILED DESCRIPTION

[0099] The technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0100] An urban green space automatic layout method based on artificial intelligence, as shown in Figures 1-7 includes the following steps:

[0101] Step S1: First, collect 4-band multispectral remote sensing images of Chuzhou City area with 0.8m resolution of Gaofen-2 satellite, and shoot green space real scene pictures in the design range through a portable computer tablet with Beidou navigation system, to check the satellite remote sensing images on site, and supplement if there is a lack. At the same time, obtain urban three-dimensional vector data and urban green space planning scheme data from the local planning department, and build a geographic spatial information database. The geographic spatial information data is generated after unifying the urban three-dimensional vector data to WGS84 geographic coordinates, containing four types of surface data of road blocks, water system blocks, green blocks and other functional blocks, and two types of linear data of road center lines and block boundary lines, wherein the road center line is a continuous and non-closed line segment, and the block boundary line is a closed line segment. Collect the control detailed planning text and related legal specification files related to the design area, and establish the digital verification system of "Urban Green Space Planning Standard GB / T51346-2019" through the text language digitization translation system.

[0102] Step S2: Call the geospatial information database and build a geospatial digital sand table. The urban green space elements are intelligently identified by the nearest neighbor rule classification algorithm (KNN). There are a total of 102 green space elements in the design area. Match the land use function characteristics of each green space element, including water system land and green land. Use the computational geometry tool in the geographic information system to calculate the area value of the green space block, denoted as the area characteristic. Use the Minimum Bounding Geometr tool in the geographic information system to build the minimum bounding rectangle of the green space block as a layout characteristic space unit. Record all road data contained in the space unit, including the orientation of the road to the green space block, road name, road length, and road grade. If the aspect ratio of the minimum bounding rectangle of the green space block is less than 2, it is marked as a point-like element type, a total of 18 point-like green space elements. If the aspect ratio of the minimum bounding rectangle is greater than 2, it is marked as a linear element type, a total of 84 linear green space elements.

[0103] Step S3: Build intelligent classification indicators for block-like green space elements. Use the area characteristic as the intelligent classification indicator a1. In the geographic information platform, mark green space blocks with an area greater than 50 hectares as large ecological source points. Build intelligent classification indicators for linear green space elements. Calculate the distance between the centroid of the green space block and the centroid of the large ecological source point in the geographic information system after unifying the coordinate system as an intelligent classification indicator a2. Use the highest road grade in the adjacent road of the green space element as an intelligent classification indicator a3. Calculate the intelligent classification indicator values for block-like green space and linear green space, and embed them in the green space block in the form of an attribute table.

[0104] Step S4: First, intelligently classify the 18 block-like green space elements. Extract the area a1 data. According to the "Urban Green Space Planning Standard GB / T 51346-2019", block-like green space with an area of 5.0-10.0 hectares is classified as level one, belonging to the community park type, with a service radius of 1000m. Block-like green space with an area of 1.0-5.0 hectares is classified as level two, belonging to the community park type, with a service radius of 500m. Block-like green space with an area of 0.2-1.0 hectares is classified as level three, belonging to the garden type, with a service radius of 300m. Intelligently classify the 84 linear green space elements. Extract the distance a2 data and road grade a3 data from S3-2 and S3-3. Build a linear green space classification standard table. Road grade is the main road and distance from the large ecological source point is less than or equal to 10km, marked as A-level linear green space. Road grade is the main road and distance from the large ecological source point is greater than 10km, marked as B-level linear green space. Road grade is the secondary road and distance from the large ecological source point is less than or equal to 10km, marked as C-level linear green space. Road grade is the secondary road and distance from the large ecological source point is greater than 10km, marked as D-level linear green space. Road grade is the branch road, marked as E-level linear green space.

[0105] Step S5: First, the connecting linear green space elements generate green corridors. Call the road data in the database, use the pathfinding algorithm (Dijkstra), and use the hierarchical linear green space in step S4 as the starting block S. The first case is to automatically connect the A-level linear green space and the B-level linear green space to generate a first-level green corridor, connect the C-level linear green space and the D-level linear green space to generate a second-level green corridor, and connect the E-level green space to generate a third-level green corridor. The second case is to automatically connect the A-level linear green space and the B-level linear green space to generate a first-level green corridor, connect the C-level linear green space to generate a second-level green corridor, and connect the D-level and E-level green space to generate a third-level green corridor. The third case is to automatically connect the A-level linear green space to generate a first-level green corridor, connect the B-level linear green space and the C-level linear green space to generate a second-level green corridor, and connect the D-level and E-level linear green space to generate a third-level green corridor. The fourth case is to automatically connect the A-level linear green space to generate a first-level green corridor, connect the B-level linear green space to generate a second-level green corridor, and connect the C-level, D-level, and E-level linear green space to generate a third-level green corridor. A total of 4 green corridor layout schemes are generated. Using the multi-agent algorithm, in combination with the hierarchical block green space and its service radius in step S4, the block green space is automatically generated and randomly arranged. If all green space blocks meet the requirements of “Urban Green Space Planning Standard GB / T 51346-2019”, the four preliminary green space layout schemes are generated and stored in the NAS network storage server.

[0106] Step S6: First, classify green space elements by area size. Through the MATLAB shape recognition system, 30458 green space elements in the 128 green space layout schemes output in step S5 are planar shape recognized. According to the geometric area size, the green space elements are classified, and the categories include below 2 hectares, 2-5 hectares, 5-20 hectares, and above 20 hectares. Second, calculate the spatial repulsion of green space elements of the same type. Through the multi-agent algorithm, the spatial repulsion between the classified green space elements of the same type is calculated. The formula is Frepulsion = H(S1+S2) / D, where Frepulsion is the spatial repulsion of the same type, H is a constant, S1 and S2 are the geometric areas of the same type, and D is the geometric center distance between the same type. Then, determine whether the spatial repulsion of green space elements of the same type is compliant. Compare the spatial repulsion of green space elements with the threshold value, if both are less than the threshold value, enter step S8; otherwise, mark the green space elements with a spatial repulsion greater than the threshold value through dynamic calculation engine technology, and enter step S7. After the spatial repulsion calculation of the same type, 17450 green space elements have a spatial repulsion greater than the threshold value of 18.32N, which need to enter step S7 to assign a spatial movement vector to correct the repulsion.

[0107] Step S7: The 17450 green space elements that do not meet the threshold value 18.32N in step S6 are assigned a spatial movement vector according to the repulsive force size by the multi-agent algorithm. S = Frepulsion / a, S is the distance size of the spatial movement vector, and its direction is the same as the repulsive force direction. Then step S6 is performed again, and there are still 1238 green space elements whose repulsive force size exceeds the threshold value, and the spatial movement vector is re-assigned. Step S6 is performed again, and there are 243 green space elements whose repulsive force size exceeds the threshold value……After 15 iterations, the repulsive force size of all green space elements is less than the threshold value, and step S8 is entered.

[0108] Step S8: First, the spatial attraction calculation of green space elements of different types is performed. The spatial attraction between the classified green space elements of different types is calculated by the multi-agent algorithm. The calculation formula is Fattraction = R(S1+S2) / D, where Fattraction is the spatial attraction of different types of elements, R is a constant, S1 and S2 are the geometric areas of different types of elements, and D is the geometric center distance between different types of elements. Second, it is determined whether the spatial attraction of green space elements of different types is in compliance. In this embodiment, the 128 green layout schemes obtained after iteration, a total of 30458 green space elements, are calculated by the spatial attraction of different types of elements. There are 10340 green space elements with a spatial attraction greater than the threshold value of 35.23N, which need to enter step S9 to assign a spatial movement vector to correct the repulsive force.

[0109] Step S9: First, the 10340 green space elements that do not meet the threshold value 35.23N in step S8 are assigned a spatial movement vector according to the attractive force size. S = Fattraction / a, S is the distance size of the spatial movement vector, and its direction is the same as the repulsive force direction. Then step S6 is performed again, and there are still 489 green space elements whose repulsive force size exceeds the threshold value, and the spatial movement vector is re-assigned. Step S8 is performed again, and there are 212 green space elements whose repulsive force size exceeds the threshold value……After 9 iterations, the attractive force size of all green space elements is less than the threshold value, and step S10 is entered.

[0110] Step S10: According to the urban green space layout specification, the digital verification system of “Urban Green Space Layout Specification” is established by the text language digitization translation system. The digital verification system verifies the green layout scheme, and deletes the illegal scheme. Then the green layout scheme is manually verified, and the illegal scheme is deleted.

[0111] Step S10-1: First, the text language digital translation system is established to translate the "Urban Green Space Layout Specification" into a rule language, and the "Urban Green Space Layout Specification" digital verification system is established. Second, the green space layout scheme is verified through the digital verification system, and the green space layout scheme that violates the specification is deleted. After 128 green spaces, the green space layout scheme is verified offline by the geographic information platform database management system according to the "Urban Green Space Layout Specification", and the green space layout scheme that violates the specification is deleted. The remaining 15 schemes are checked by artificial, 6 of which do not meet the specification requirements and are deleted, and 9 green space layout schemes are obtained.

[0112] Step S11: Select the green space layout scheme by expert scoring through the online judge scoring system. Select 20 experts who have worked in the city planning or landscape industry for more than 10 years, score the green space layout scheme through the online judge scoring system, take the integer of the full score of 10, remove the 2 highest scores and the 2 lowest scores, and select the green space layout scheme with the highest average score as the final scheme. After expert scoring, the 9 green space layout schemes have an average score of 7.64, 7.32, 6.32, 6.30, 6.12, 6.03, 5.98, 5.32, and 5.21, respectively. Finally, the green space layout scheme with a score of 7.64 is selected as the final scheme.

[0113] Step S12: Based on the final scheme generated in step S11, further three-dimensional display of the green space layout scheme is performed, and the scheme is projected in a three-dimensional platform with holographic display function at a scale of 1:1000. The display content includes plan, important node plan, and three-dimensional holographic sand table.

[0114] Step S13: Construct a green space layout scheme auxiliary decision-making instruction library. It contains four types of operations: display, selection, call, and modification. Connect the auxiliary decision-making instruction library with the green space layout scheme generated in step S11, and map it in the display device with interactive and display functions.

[0115] Step S14: Assist decision-making for green space layout scheme through 3D holographic projection. Use 3D holographic projection equipment, VR glasses, and virtual reality data gloves to interact with the green space layout scheme generated in step S11. Use the decision-making instruction library constructed in step S13 to display, select, call, and modify the Chuzhou green space layout scheme. The display is mainly in two perspectives of plan and three-dimensional; the selection is mainly in three types of model, drawing, and perspective rotation; the call is mainly for the call of green space grading indicators and weights; and the modification is mainly for the modification of linear green space and block green space in spatial layout.

[0116] Step S15: realize green space layout scheme three-dimensional model and design manual printing, use data integration and translation equipment to integrate data of Chuzhou green space layout scheme, including four types of plan, hierarchical index, weight coefficient, three-dimensional model, then, in the holographic sand table, the scheme is displayed, through the selection, rotation tool in the instruction library to clear the three-dimensional model angle, through the industrial 3d printer to print the scheme three-dimensional model with 1:1000 scale, through the drawing data integration equipment to output 1:1000 scale scheme plan, 1:1000 scale scheme bird's eye view, 1:500 scale scheme node effect diagram, scheme hierarchical index file, scheme weight coefficient file, through the printing equipment to print the above contents into the design manual.

Claims

1. An artificial intelligence-based automatic layout method for urban green space, characterized in that, The method comprises the following steps: Step S1: collecting satellite remote sensing data of the target city using the Gaofen-2 satellite with a resolution of 0.8 m, and performing field verification and supplement, obtaining three-dimensional vector data of the target city from the local planning department, inputting the data into a geographic spatial system to build a database, and collecting green space design specification files to build a specification library; Step S2: intelligently identifying the urban green space elements by using a nearest neighbor rule classification algorithm, wherein the intelligent identification features of the green space elements include land use function features, area features, spatial distribution features, and element type features, the element type features include block element type and linear element type, and the intelligent identification results are input into a geographic spatial database; Step S3: building intelligent grading indexes for block element green space and linear element green space respectively, calling the database and the specification library to quantitatively calculate and assign the grading indexes of the green space blocks in the design range, and storing the index values in the geographic information system; Step S4: building grading standard tables for block element green space and linear element green space according to the Urban Green Space Planning Standard GB / T 51346-2019 and landscape ecology, and classifying the block element green space in the design range into three levels and the linear element green space in the design range into five levels according to the grading indexes; Step S5: importing the classification results of the linear element green space in S4, using a pathfinding algorithm to generate a three-level green corridor, importing the classification results of the block element green space in S4, combining the green space service radius, and using a multi-agent algorithm to generate and initially randomly arrange the block green space, and if the arrangement is in compliance with the specification, a preliminary green space layout scheme is generated and output; Step S6: green space element spatial repulsion calculation; classifying green space elements according to area size through a MATLAB shape recognition system; calculating green space element spatial repulsion of the same type of elements through a multi-agent algorithm, and if the calculation is in compliance with the threshold condition, entering step S8, or if the calculation is not in compliance with the threshold condition, entering step S7; Step S7: adjusting the green space elements with spatial repulsion violations through a multi-agent algorithm, assigning a spatial movement vector according to the size of the spatial repulsion, and entering step S6; Step S8: green space element spatial attraction calculation of different types of elements; calculating green space element spatial attraction of different types of elements through a multi-agent algorithm, and if the calculation is in compliance with the threshold condition, entering step S10, or if the calculation is not in compliance with the threshold condition, entering step S9; Step S9: adjusting the green space elements with spatial attraction violations through a multi-agent algorithm, assigning a spatial movement vector according to the size of the spatial attraction, and entering step S8; Step S10: verifying according to the urban green space layout specification; establishing a digital verification system of the Urban Green Space Layout Specification through a text language digitization translation system; verifying the green space layout scheme through the digital verification system, deleting the non-compliant scheme; and performing manual verification of the green space layout scheme again, and deleting the non-compliant scheme; Step S11: selecting a green space layout scheme through an online judge scoring system; taking the integer part of the full score of 10, removing the two highest scores and the two lowest scores, and selecting the green space layout scheme with the highest average score as the final scheme. Step S12: realize the three-dimensional display of the green space layout scheme, display the green space layout scheme generated in S11 in a three-dimensional platform with holographic display function; Step S13: construct a green space layout scheme auxiliary decision-making instruction library containing four types of operations: display, selection, call, and modification; and map it in a display device with interactive and display functions through a projection device; Step S14: use the decision-making instruction library constructed in step S13 to perform auxiliary decision-making on the green space layout scheme by 3D holographic projection, and use the display, selection, call, and modification instructions to display the green space layout scheme; Step S15: realize the three-dimensional model of the green space layout scheme and the printing of the design manual; integrate the data using data integration and translation equipment, print the three-dimensional model of the scheme using an industrial 3D printer, and print the green space layout scheme drawings into a design manual using a printing device.

2. The method according to claim 1, wherein, The step S1 specifically comprises: Step S1-1: Obtain geographic space information data and establish a database 0.8m resolution high-resolution satellite 2 satellite is used to collect 4-band multispectral remote sensing images of urban area, and green field real scene pictures of the area are taken by WiFi version portable computer tablet with Beidou navigation system, satellite remote sensing images are checked and supplemented; obtain three-dimensional vector data and urban green space planning scheme data of the city where the design range is located from the local planning department; Step S1-2: Obtain specification file data and establish a specification library Collect the control detailed planning text and related legal specification files related to the design area, and establish the digital verification system of "Urban Green Space Planning Standard GB / T 51346-2019" through the text language digitization translation system.

3. The method according to claim 2, wherein, The geographic space information data in step S1 is generated by converting the unified city three-dimensional vector data to WGS84 geographic coordinates, including four types of surface data: road blocks, water system blocks, green space blocks, and other functional blocks, and two types of linear data: road center lines and block boundary lines. Among them, the road center line is a continuous and non-closed line segment, and the block boundary line is a closed line segment.

4. The method according to claim 3, wherein, The step S2 specifically comprises: Step S2-1: Call the geographic space information database to build a geographic space digital sand table; Step S2-2: Match the land use function characteristics of the green space block, including water system block land use and green space block land use; Step S2-3: Use the computational geometry tool in the geographic information system to numerically calculate the area of the green space block, denoted as the area characteristic; Step S2-4: Use the Minimum Bounding Geometr tool in the geographic information system to construct the minimum circumscribed rectangle of the green space block as a layout characteristic space unit, and record all road data contained in the space unit, including the orientation of the road to the green space block, road name, road length, and road grade; Step S2-5: Import the minimum circumscribed rectangle of the green land plot constructed in step S2-4, if the aspect ratio of the minimum circumscribed rectangle is less than 2, mark it as a block element type, if the aspect ratio of the minimum circumscribed rectangle is greater than 2, mark it as a linear element type.

5. The method of claim 4, wherein the method is based on artificial intelligence. The step S3, specifically includes: Step S3-1: Construct an intelligent grading index α1 of block element green land, input the area of green land plot as the intelligent grading index α1; Step S3-2: Construct an intelligent grading index α2 of linear element green land, mark the green land plot with an area greater than 50 hectares as a large ecological source point, calculate the distance between the centroid of the green land plot and the centroid of the large ecological source point in the geographic information system after unifying the coordinate system as the intelligent grading index α2; Step S3-3: Construct an intelligent grading index α3 of linear element green land, input the road data in the spatial unit in step S2-4, and take the highest level in the road grade as the intelligent grading index α3; Step S3-4: Calculate the intelligent grading index value of the green land plot, and embed it in the green land plot in the form of an attribute table.

6. The method of claim 5, wherein the method is based on artificial intelligence. The step S4, specifically includes: Step S4-1: Intelligent grading of block element green land, extract the intelligent grading index α1 data in step S3-1, and compare it with the park green land grading setting requirements in the "Urban Green Land Planning Standard GB / T 51346-2019" to divide the block green land into three levels: greater than 5.0 hectares, 1.0-5.0 hectares, and 0.2-1.0 hectares; Step S4-2: Intelligent grading of linear element green land, extract the distance α2 data from the large ecological source point and the road grade α3 data in steps S3-2 and S3-3, construct a linear element green land grading standard table, and divide the linear green land into five levels.

7. The method according to claim 6, wherein, The step S5, specifically includes: Step S5-1: Call the road data in the database, use the pathfinding algorithm, and take the linear green land graded in step S4 as the starting block S, automatically connect and generate a first-level green corridor, a second-level green corridor, and a third-level green corridor; Step S5-2: Use the multi-agent algorithm, combine the block green land graded in step S4 and its service radius, and automatically generate and randomly arrange the block green land, if all green land plots meet the requirements of the "Urban Green Land Planning Standard GB / T 51346-2019", generate and store the preliminary green land space layout scheme in the NAS network storage server.

8. The method according to claim 7, wherein, The step S6, specifically includes: Step S6-1: Classification of green land space elements by area Perform planar shape recognition on the green land space elements in the green land layout scheme output in step S5 through the MATLAB shape recognition system; classify the green land space elements by geometric area size, including less than 2 hectares, 2-5 hectares, 5-20 hectares, and more than 20 hectares; Step S6-2: Calculation of spatial repulsive force of green land space elements of the same type The repulsion force between the classified green space elements of the same kind is calculated by the multi-agent algorithm; the calculation formula is Frepulsion=H(S1+S2) / D, wherein Frepulsion is the repulsion force between the elements of the same kind, H is a constant, S1 and S2 are the geometric areas of the elements of the same kind, and D is the geometric center distance between the elements of the same kind; Step S6-3: determining whether the repulsion force between the green space elements of the same kind is in compliance The repulsion force and the threshold value of the green space elements are compared, if both are less than the threshold value, step S8 is entered; otherwise, the green space elements with the repulsion force greater than the threshold value are marked, and step S7 is entered; the threshold value is the repulsion force of the green space elements of the same kind when the interval is twice the maximum service radius specified in the Urban Green Space Layout Specification.

9. The method according to claim 8, wherein, The step S7 specifically comprises: Step S7: adjusting the green space elements with the repulsion force in violation of the rules The green space elements with the repulsion force in violation of the rules are adjusted by the multi-agent algorithm, the space movement vector is given according to the repulsion force of the green space elements marked in S6-3, the optimized green space elements are obtained, the green space layout scheme is updated, and S6 is entered.

10. The method of claim 9, wherein the method is based on artificial intelligence. The step S8 specifically comprises: Step S8-1: calculating the attraction force between the green space elements of different kinds The attraction force between the classified green space elements of different kinds is calculated by the multi-agent algorithm; the calculation formula is Fattraction=R(S1+S2) / D, wherein Fattraction is the attraction force between the elements of different kinds, R is a constant, S1 and S2 are the geometric areas of the elements of different kinds, and D is the geometric center distance between the elements of different kinds; Step S8-2: determining whether the attraction force between the green space elements of different kinds is in compliance The attraction force and the threshold value of the green space elements are compared, if both are less than the threshold value, step S10 is entered; otherwise, the green space elements with the attraction force greater than the threshold value are marked, and step S9 is entered; the threshold value is the attraction force of the green space elements of different kinds when the interval is the sum of the maximum service radii specified in the Urban Green Space Layout Specification.

11. The method of claim 10, wherein the method is based on artificial intelligence. The step S9 specifically comprises: Step S9: adjusting the green space elements with the attraction force in violation of the rules The green space elements with the attraction force in violation of the rules are adjusted by the multi-agent algorithm, the space movement vector is given according to the attraction force of the green space elements marked in S8-2, the optimized green space elements are obtained, the green space layout scheme is updated, and S8 is entered.

12. The method of claim 11, wherein, The step S10 specifically comprises: Step S10-1: establishing the digital verification system of the Urban Green Space Layout Specification The Urban Green Space Layout Specification is translated into a rule language by establishing a text language digitization translation system, and the digital verification system of the Urban Green Space Layout Specification is established; Step S10-2: verifying the green space layout scheme by the digital verification system The green space layout scheme is verified by the digital verification system of the Urban Green Space Layout Specification, and the green space layout scheme in violation of the specification is deleted; Step S10-3: manually checking the green space layout scheme The green space layout scheme is checked according to the Urban Green Space Layout Specification by offline manual through the geographic information platform database management system, and the green space layout scheme in violation of the specification is deleted.

13. The method of claim 12, wherein the method is based on artificial intelligence. The step S11 specifically comprises: Step S11: selecting the green space layout scheme by expert scoring Select 20 experts in the city planning or landscape industry work more than 10 years, through the online scoring system to score green space layout scheme, full score 10 points, take the integer, remove the two highest score and two lowest score, select the highest average score of green space layout scheme as the final scheme.

14. The method according to claim 13, wherein, The step S12 specifically comprises: Step S12: green space layout scheme three-dimensional display The generated green space layout scheme is projected in a three-dimensional platform with holographic display function at a scale of 1:1000, and the green space layout scheme is displayed in all directions using 3D holographic projection. The equipment includes VR panorama display table, 3D tracking glasses; the display content includes plan, important node plan, three-dimensional holographic sand table.

15. The method of claim 14, wherein the method is based on artificial intelligence. The step S13 specifically comprises: Step S13: construction of green space layout scheme auxiliary decision-making instruction library The auxiliary decision-making instruction library is connected with the green space layout scheme generated in S11 through four types of operations including display, selection, calling and modification, and is mapped in the display device with interactive and display functions through the projection device.

16. The method of claim 15, wherein, The step S14 specifically comprises: The green space layout scheme generated in step S11 is imported into the city three-dimensional space digital model using 3D holographic projection equipment, VR glasses and virtual reality data gloves, and the decision-making instruction library constructed in step S13 is used to display, select, call and modify the green space layout scheme; The city three-dimensional space digital model is generated after the unified city three-dimensional vector data is converted to the 2000 national geodetic coordinate system, and contains city geographic elevation, road network, city water system and city mountain information; The modification in the instruction library is to adjust the parameter values of green space level and spatial distribution position. The green space level adjustment means that the computer calculates the green space levels L1, L2 and L3 through the classification index, and then calculates the average error using the weight formula, and automatically corrects the error by computer. The green space layout adjustment means that the computer calculates the spatial distribution of each green space through the force iteration of attraction and repulsion by multi-agent algorithm, and the user adjusts the force iteration interval according to the decision-making demand, and automatically corrects the error by computer.

17. The method of claim 16, wherein, The step S15 specifically comprises: The data integration and translation equipment is used to integrate the three types of data of green space layout scheme plan, classification index and weight coefficient, and the data is displayed in the holographic sand table. The three-dimensional model of the scheme is printed by industrial 3D printer. The planar graph of the green space layout scheme, the classification index file of the scheme, and the weight coefficient file of the scheme are output by the drawing data integration equipment at a scale of 1:1000, 1:1000 and 1:500 respectively, and the above contents are printed into a design manual by the printing equipment.

Images