ARTIFICIAL INTELLIGENCE-BASED URBAN DESIGN MULTI-PLAN GENERATION METHOD FOR REGULATORY PLOT

Abstract

The present invention discloses an artificial intelligence (AI)-based urban design multi-plan generation method for a regulatory plot. The method includes operation steps performed by the following modules: (1) plot space calculation sand table module; (2) regulatory characteristic parameter input module; (3) characteristic case intelligent learning module; (4) plot road intelligent generation module; (5) plot landscape intelligent generation module; (6) plot architecture intelligent generation module; and (7) outputted plan auxiliary drawing module. By means of the AI-based urban design multi-plan generation method for a regulatory plot of the present invention, the planners realize real-time, accurate, and efficient feedback adjustment to a plurality of urban design plans under regulatory conditions, resolving a plurality of problems such as a long working period of a conventional design plan.

Claims

1. An artificial intelligence (AI)-based urban design multi-plan generation method for a regulatory plot, the method comprising the following operation steps: S1: collecting, by a plot space calculation sand table module, two-dimensional vector data of information about architectures and roads surrounding a design plot and three-dimensional spatial information from oblique photography of surrounding architectures by using an unmanned aerial vehicle (UAV); S2: acquiring, by a regulatory characteristic parameter input module, data, in a computer aided design (CAD) format, of roads forming plots in a regulatory plan of the design plot in a master plan and regulatory plot index data of the design plot, and adjusting coordinates of the vector data to be consistent in an ultra-deep computer and connecting regulatory indexes to an architecture-road space model of the design plot; S3: acquiring, by a characteristic case intelligent learning module, a case database of a current construction plan of urban design by using an image scanner, extracting a characteristic index system of the urban design plan case, prioritizing characteristic parameters, establishing a characteristic index system decision tree, and selecting similar cases based on the decision tree, to form a characteristic learning case database; S4: intelligently generating, by a plot road intelligent generation module, a plurality of plans for a plot road network on an ultra-deep computer by using a combination of a gridding method and a shortest path method according to locations of entrances and exits of each block determined in the regulatory plan, verifying the plans by using a road rule verification model, and outputting, on a printer, road network plans that satisfy rules and characteristic parameters of the plans; S4 comprises the following four steps: S4.1: forming a hidden grid of 30 m×30 m between an entrance and an exit that are determined, automatically finding and connecting a shortest path between the entrance and the exit on the ultra-deep computer, and repeating the steps until all entrances and exits of each block of the plot in S4.1 are connected by a shortest path; S4.2: forming a road centerline of the design plan according to S4.1, and automatically widening a road on the ultra-deep computer in accordance with the specification for urban branch road; S4.3: verifying the plan generated in S4.2 in accordance with the block scale specification and the road density specification in specification parameters in GB50220-95 Code for Transport Planning on Urban Road that have different functions, and eliminating road plans that do not conform to the specifications; and S4.4: outputting, on the printer, road network plans that conform to the specifications of S4.3, and generating interactive parameters of the road network plans, wherein the interactive parameters comprise road network density and uniformity, road network accessibility, and road network connectivity; S5: intelligently generating, by a plot landscape intelligent generation module, a landscape system multi-plan on the ultra-deep computer according to a land property of each block determined in the regulatory plan and the road network plan intelligently generated in S4 by using a combination of parameter control of a basic landscape unit and an evolutionary algorithm, verifying the plan by using a landscape system rule verification model, and outputting, on the printer, landscape system plans that satisfy the rules and characteristic parameters of the plans; S5 comprises the following six steps: S5.1: setting a point in a block in a non-residential land that nears an external road by using a uniform distribution method as a starting point for generating the landscape system; S5.2: setting parameters of a basic unit for generating the landscape system: an open angle range α and a length range li; S5.3: configuring a landscape system on the ultra-deep computer to generate a fitness function; S5.4: causing the basic unit to grow by using the starting point generated for the landscape system, by using the evolutionary algorithm L-System on the ultra-deep computer; and stopping the growth of the basic unit when the landscape system in each block of the non-residential land is connected and an evolved tree structure is connected to all urban roads on boundaries of the non-residential blocks; S5.5: verifying the generated plans in accordance with parameters of the landscape system specification based on the tree structure of the landscape system; and S5.6: outputting, on the printer, the landscape system plan verified by using the specifications in S5.5, and generating interactive parameters of the landscape system plan, wherein the interactive parameters comprise a street block quantity, a street block area, and a street block shape index; S6: constructing, by a plot architecture intelligent generation module, a sample database of architectural combinations on the ultra-deep computer, intelligently matching characteristic indexes of the sample database of the architectural combinations of the design plot with a case database of the architectural combinations, to generate a plan case learning database, generating architectural combination plans with different functions by using a CVAE-GAN complementation algorithm and an adaptive algorithm, verifying the plans on the ultra-deep computer by using regulatory plot space parameters and a sunshine spacing, and outputting, on the printer, architectural combination plans that satisfy the rules and characteristic parameters of the plans; S6 comprises the following six steps: S6.1: constructing a sample database of architectural combination forms on an ultra-deep computer, comprising: collecting and vectorizing data of architectural combination samples, intercepting architectural combination image data by using the Google Map, and vectorizing architecture boundaries and roads; S6.2: extracting characteristic indexes of the architectural combination samples; S6.3: performing intelligent matching on the case database of the architectural combinations, comprising: comparing an interaction index of each street block with the case database, arranging sample architectural combinations having a matching degree of 90% according to the matching degree, and selecting first 1000 architectural combinations having the matching degree to generate a case learning database; S6.4: performing machine learning by using the first 1000 architectural combinations as data of the CVAE-GAN complementation algorithm and the architecture adaptive algorithm to generate intelligent architectural combination plans for different functions of each street block; S6.5: performing verification based on the inputted regulatory plot space parameters inputted in S2 and the residential architecture sunshine spacing, and eliminating plans that do not conform to the specifications: S6.6: outputting, on the printer, architectural combination plans that conform to the specifications of sunshine and fire protection, and generating interactive parameters of the architectural combination plans, wherein the interactive parameters comprise skyline profile volatility of the architectural combination plans; and S7: output ting, by an outputted plan auxiliary drawing module, GIS data after integration of road-landscape-architecture plans, and outputting a rule verification parameter report and a design characteristic parameter report data table on the printer, to form an engineering report drawing and printing the engineering report drawing, inputting the GIS data into a visualization platform system, and interactively displaying the GIS data to citizens by using a head-mounted VR interactive device; S7 comprises the following four steps: S7.1: hierarchically outputting and merging the road-landscape-architecture plans, classifying roads, landscapes, and architectures in the plans into layers in a GIS data format and successively naming the roads, landscape and architectures as architecture, road, and landscape, and importing the data into an AI urban design platform; S7.2: combining, in the three-dimensional data platform, the plan data extracted in S7.1 with the rule verification parameter report and the design characteristic parameters obtained in S4.4, S5.6, and S6.6, outputting a multi-plan report data table and forming an engineering report drawing, and printing the engineering report drawing in a text format by a laser printer; S7.3: combining, in the three-dimensional data platform, the plan data extracted in S7.1 with the current three-dimensional real scene data obtained in S1, adjusting the coordinates, so that the two pieces of data are in a same coordinate system, setting a plurality of observation points in a new three-dimensional model database, generating and exporting a new urban scene after planning and design in the AI urban design platform; and S7.4: wearing virtual reality glasses to perform scene roaming simulation of the observation points of the urban design plan determined in S7.3.

2. The AI-based urban design multi-plan generation method for a regulatory plot according to claim 1, wherein S1 comprises the following four steps: S1.1: collecting the two-dimensional vector data of the information about the architectures and the roads surrounding the design plot, wherein the architecture data is a closed polygon and comprises information about a quantity of architecture storeys, and the road data comprises information about a centerline and a road width of each road; S1.2: verifying the three-dimensional spatial information from the oblique photography of the architectures surrounding the design plot by means of on-site collection by using the UAV; S1.3: adjusting coordinates of the vector data to be consistent, loading, into the AI urban design platform, the two-dimensional vector data of the architectures and the roads surrounding the design plot and the three-dimensional information from the oblique photography of roads forming the regulatory and design plot and the architectures surrounding the plot, and running the data on the computer; and S1.4: performing stretching by using a storey height of 3 m based on the information about the architecture storeys, to obtain a current three-dimensional model of the architectures surrounding the design plot; and generating a three-dimensional model of the roads surrounding and forming the design plot based on information about a road centerline and a road elevation point and a road width value, so as to establish a basic sand table for plot space calculation, and

3. The AI-based urban design multi-plan generation method for a regulatory plot according to claim 1, wherein S2 comprises the following two steps: S2.1: scanning regulatory plan maps and drawings of the plot by using a scanner, to obtain the regulatory plot index data of the design plot, wherein the regulatory plot index data comprises a land property of each block, locations of the entrances and the exits in each block, a development intensity of each block, an architecture discount rate, and an architecture setback line; and S2.2: inputting the regulatory plot space parameters to the AI urban design platform, and spatially establishing a connection to the current three-dimensional model of the architectures surrounding the design plot and the three-dimensional model of the roads surrounding and forming the design plot.

4. The AI-based urban design multi-plan generation method for a regulatory plot according to claim 1, wherein S3 comprises the following two steps: S3.1. acquiring the case database of the current construction plan of urban design, and extracting the characteristic index system of the urban design plan case; and S3.2. prioritizing the characteristic parameters, establishing the characteristic index system decision tree, and selecting the similar cases based on the decision tree, to form the characteristic learning case database.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0057] FIG. 1 is an overall flowchart of a method according to an embodiment of the present invention.

DETAILED DESCRIPTION

[0058] The technical solutions of the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings of the embodiments of the present invention.

[0059] As shown in FIG. 1, an AI-based urban design multi-plan generation method for a regulatory plot includes operation steps:

[0060] S1: A plot space calculation sand table module collects two-dimensional vector data of information about architectures and roads surrounding a design plot in a specific range and three-dimensional spatial information from oblique photography of surrounding architectures.

[0061] S1 Includes the Following Four Steps:

[0062] S1.1: collecting the two-dimensional vector data of the information about the architectures and the roads surrounding a design plot in a specific range, where the architecture data is a closed polygon and includes information about a quantity of architecture storeys, and the road data includes information about a centerline and a road width of each road;

[0063] S1.2: verifying the three-dimensional spatial information from the oblique photography of the architectures surrounding the design plot by means of on-site collection by using a UAV;

[0064] S1.3: adjusting coordinates of the vector data to be consistent, loading, into the AI urban design platform, the two-dimensional vector data of the architectures and the roads surrounding the design plot and the three-dimensional information from the oblique photography of roads forming the regulatory and design plot and the architectures surrounding the plot, and running the data on a computer; and

[0065] S1.4: performing stretching by using a storey height of 3 m based on the information about the architecture storeys, to obtain a current three-dimensional model of the architectures surrounding the design plot; and generating a three-dimensional model of the roads surrounding and forming the design plot based on information about a road centerline and a road elevation point and a road width value, so as to establish a basic sand table for plot space calculation.

[0066] S2: A regulatory characteristic parameter input module acquires data, in a computer aided design (CAD) format, of roads forming plots in a regulatory plan of the design plot in a master plan and regulatory plot index data of the design plot, and adjusts coordinates of the vector data to be consistent and connect regulatory indexes to an architecture-road space model of the design plot.

[0067] S2 Includes the Following Two Steps:

[0068] S2.1: scanning regulatory plan maps and drawings of the plot by using a scanner, to obtain the regulatory plot index data of the design plot, where the regulatory plot index data includes a land property of each block, locations of entrances and exits in each block, a development intensity (an architecture density, an architecture height, and a plot ratio) of each block, an architecture discount rate, and an architecture setback line; and

[0069] S2.2: inputting the regulatory plot space parameters to an AI urban design platform, and spatially establishing a connection to the current three-dimensional model of the architectures surrounding the design plot and the three-dimensional model of the roads surrounding and forming the design plot.

[0070] S3: A characteristic case intelligent learning module acquires a case database of a current construction plan of urban design, extracts a characteristic index system of the urban design plan case, prioritizes characteristic parameters, establishes a characteristic index system decision tree, and selects similar cases based on the decision tree, to form a characteristic learning case database.

[0071] S3 Includes the Following Two Steps:

[0072] S3.1. acquiring the case database of the current construction plan of urban design, and extracting the characteristic index system of the urban design plan case; and

[0073] S3.2. prioritizing the characteristic parameters, establishing the characteristic index system decision tree, and selecting the similar cases based on the decision tree, to form the characteristic learning case database.

[0074] S4: A plot road intelligent generation module intelligently generates a plurality of plans for a plot road network by using a combination of a gridding method and a shortest path method according to locations of entrances and exits of each block determined in the regulatory plan, verifies the plans by using a road specification verification model, and outputs road network plans that conform to specifications and characteristic parameters of the plans.

[0075] S4 Includes the Following Four Steps:

[0076] S4.1: forming a hidden grid of 30 m×30 m between an entrance and an exit that are determined, automatically finding and connecting a shortest path between the entrance and the exit, and repeating the steps until all entrances and exits of each block of the plot in S4.1 are connected by a shortest path;

[0077] S4.2: forming a road centerline of the design plan according to S4.1, and automatically widening a road in accordance with the specification for urban branch road;

[0078] S4.3: verifying the plan generated in S4.2 in accordance with the block scale specification and the road density specification in specification parameters in GB50220-95 Code for Transport Planning on Urban Road that have different functions, and eliminating road plans that do not conform to the specifications; and

[0079] S4.4: outputting road network plans that conform to the specifications of S4.3, and generating interactive parameters of the road network plans, where the interactive parameters include road network density and uniformity, road network accessibility, and road network connectivity.

[0080] S5: A plot landscape intelligent generation module intelligently generates a landscape system multi-plan according to a land property of each block determined in the regulatory plan and the road network plan intelligently generated in S4 by using a combination of parameter control of a basic landscape unit and an evolutionary algorithm, verifies the plan by using a landscape system rule verification model, and outputs landscape system plans that satisfy the rules and characteristic parameters of the plans.

[0081] S5 Includes the Following Six Steps:

[0082] S5.1: setting a point in a block in a non-residential land that nears an external road by using a uniform distribution method as a starting point for generating the landscape system;

[0083] S5.2: setting parameters of a basic unit for generating the landscape system: an open angle range α, a length range li, and a range of a crossroad quantity;

[0084] S5.3: configuring a landscape system to generate a fitness function;

[0085] S5.4: causing the basic unit to grow by using the starting point generated for the landscape system, by using the evolutionary algorithm L-System; and stopping the growth of the basic unit when the landscape system in each block of the non-residential land is connected and an evolved tree structure is connected to all urban roads on boundaries of the non-residential blocks;

[0086] S5.5. verifying the generated plans in accordance with parameters of the landscape system specification based on the tree structure of the landscape system (separation of people and vehicles—a landscape road does not overlap a vehicle road, is connected to a non-residential public functional block, and is connected to a peripheral public functional block); and

[0087] S5.6: outputting the landscape system plan verified by using the specifications in S5.5, and generating interactive parameters of the landscape system plan, where the interactive parameters include a street block quantity, a street block area, and a street block shape index.

[0088] S6: A plot architecture intelligent generation module constructs a sample database of architectural combinations, intelligently matches characteristic indexes of the sample database of the architectural combinations of the design plot with a case database of the architectural combinations, to generate a plan case learning database, generates architectural combination plans with different functions by using a CVAE-GAN complementation algorithm and an adaptive algorithm, verifies the plans by using regulatory plot space parameters and a sunshine spacing, and outputs architectural combination plans that satisfy the rules and characteristic parameters of the plans.

[0089] S6 Includes the Following Six Steps:

[0090] S6.1: constructing a sample database of architectural combination forms on an ultra-deep computer, including: collecting and vectorizing data of architectural combination samples, intercepting architectural combination image data by using the Google Map, and vectorizing architecture boundaries and roads;

[0091] S6.2: extracting characteristic indexes of the architectural combination samples (including a street block shape, a street block area, an architecture density, a plot ratio, and a land property) (calculating the architecture density based on an architecture area and the street block area; recognizing an architecture height by using an architecture shadow and calculating the floor area ratio; and recognizing an architecture function and the land property according to an architecture shape):

[0092] S6.3: performing intelligent matching on the case database of the architectural combinations, including: comparing an interaction index of each street block with the case database, arranging sample architectural combinations having a matching degree of 90% according to the matching degree, and selecting first 1000 architectural combinations having the matching degree to generate a case learning database;

[0093] S6.4: performing machine learning by using the first 1000 architectural combinations as data of the CVAE-GAN complementation algorithm and the architecture adaptive algorithm to generate intelligent architectural combination plans for different functions of each street block:

[0094] S6.5: performing verification based on the inputted regulatory plot space parameters (a development intensity of each block (an architecture density, an architecture height, a plot ratio), an architecture discount rate, and an architecture setback line) inputted in S2 and a residential architecture sunshine spacing, and eliminating plans that do not conform to the specifications; and

[0095] S6.6: outputting, on the printer, architectural combination plans that conform to the specifications of sunshine and fire protection, and generating interactive parameters of the architectural combination plans, where the interactive parameters include skyline profile volatility of the architectural combination plans.

[0096] S7: An outputted plan auxiliary drawing module outputs GIS data after integration of road-landscape-architecture plans, and outputs a rule verification parameter report and a design characteristic parameter report data table, to form an engineering report drawing and printing the engineering report drawing, inputs the GIS data into a visualization platform system, and interactively displays the GIS data to citizens.

[0097] S7 Includes the Following Four Steps:

[0098] S7.1: hierarchically outputting and merging the road-landscape-architecture plans, classifying roads, landscapes, and architectures in the plans into layers in a GIS data format and successively naming the roads, landscape and architectures as architecture, road, and landscape, and importing the data into an AI urban design platform;

[0099] S7.2: combining, in the three-dimensional data platform, the plan data extracted in S7.1 with the rule verification parameter report and the design characteristic parameters obtained in S4.4, S5.6, and S6.6, outputting a multi-plan report data table and forming an engineering report drawing, and printing the engineering report drawing in a text format by a laser printer:

[0100] S7.3: combining, in the three-dimensional data platform, the plan data extracted in S7.1 with the current three-dimensional real scene data obtained in S1, adjusting the coordinates, so that the two pieces of data are in a same coordinate system, setting a plurality of observation points in a new three-dimensional model database, generating and exporting a new urban scene after planning and design in the AI urban design platform; and

[0101] S7.4: wearing virtual reality glasses to perform scene roaming simulation of the observation points of the urban design plan determined in S7.3.

[0102] The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above-mentioned embodiments. The foregoing embodiments and the description in the specification only illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, there will be various changes and improvements in the present invention, and these changes and improvements fall within the scope of the claimed invention.