An ai intelligent layout and landscape optimization system for landscape planning and design

Abstract

This invention discloses an AI-powered intelligent layout and landscape optimization system for landscape planning and design. The system comprises: a multi-source data preprocessing module for acquiring multi-source data for landscape planning and design, performing data cleaning and spatial data fusion; a basic rule constraint module for analyzing industry standards for landscape planning and design, establishing site-specific constraints, and transforming functional requirements into quantifiable constraint indicators; an ecological and humanistic dual-driven collaborative optimization module including a gene pool evolution sub-module and a narrative script arrangement sub-module; a simulation verification module using CFD numerical simulation technology to simulate ecological processes within the site, analyze spatial use performance indicators, and output simulation results; and a scheme output and iterative optimization module for generating visualized schemes, conducting multi-dimensional comparisons, providing users with an interactive editing interface, and performing iterative optimization. This invention enables automated landscape planning, achieves long-term stable evolution of plant communities, and boasts high design efficiency.

Description

Technical Field

[0001] This invention relates to the field of landscape design technology, and in particular to an AI-powered intelligent layout and landscape optimization system for landscape planning and design. Background Technology

[0002] Garden layout planning and design refers to a systematic project that, within a specific geographical space, comprehensively utilizes material elements such as topography, water bodies, plants, architecture, and garden features. Through scientific analysis and artistic conception, it creates a living environment that combines ecological benefits, practical functions, and aesthetic value. Its core principle follows the basic rule of "adapting to local conditions and expressing emotions through scenery," emphasizing respect for and exploration of the site's natural endowments and cultural context. In spatial organization, planning often employs axial symmetry to highlight solemn order, or utilizes traditional gardening techniques such as anticipation and winding paths to create a changing landscape experience. In terms of element allocation, it focuses on the ecological stability and seasonal changes of multi-layered plant communities of trees, shrubs, and grasses, combining the dynamic and static flow of water systems to construct a flexible landscape framework. Modern garden layout further emphasizes "people-oriented" principles, deeply integrating sponge city concepts, barrier-free design, and low-carbon technologies to meet multiple functions such as recreation, leisure, and protection. As an important link between people and nature, garden layout planning and design is not only a material expression of spatial aesthetics and ecological science, but also a key way to inherit regional history, improve urban microclimate and enhance the quality of public space.

[0003] However, existing landscape architecture planning and design techniques rely heavily on the experience and judgment of designers. Ecological layouts often deviate from the original ecosystem of the site, making it difficult to achieve the long-term stable evolution of plant communities. Furthermore, the optimization of ecological indicators requires repeated manual adjustments, which is inefficient. At the same time, regional culture lacks a continuous spatial narrative logic, making it difficult to accurately match the implicit experience needs of users. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an AI-powered intelligent layout and landscape optimization system for landscape planning and design, addressing the shortcomings of existing technologies.

[0005] The technical solution adopted by this invention to solve its technical problem is: This invention provides an AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design. The system includes: a multi-source data preprocessing module, a basic rule constraint module, a simulation verification module, a scheme output and iterative optimization module, and an ecological and humanistic dual-driven collaborative optimization module; wherein: The output of the multi-source data preprocessing module is simultaneously connected to the first input of the ecological and humanistic dual-drive collaborative optimization module and the first input of the simulation verification module; the multi-source data preprocessing module is used to acquire multi-source data for landscape planning and design, and to perform data cleaning and spatial data fusion to obtain data in a standard format for output; The output of the basic rule constraint module is connected to the second input of the ecological and humanistic dual-drive collaborative optimization module; the basic rule constraint module is used to analyze industry standards for landscape architecture planning and design, establish site baseline constraints, and transform functional requirements into quantifiable constraint indicators. The main output of the ecological and humanistic dual-drive collaborative optimization module is connected to the first input of the scheme output and iterative optimization module, and the secondary output of the ecological and humanistic dual-drive collaborative optimization module is connected to the second input of the simulation verification module. The output of the simulation verification module is connected to the second input of the scheme output and iterative optimization module; the simulation verification module is used to calculate the microclimate values ​​of the site at different times using CFD numerical simulation technology, simulate the ecological processes in the site, analyze the space use performance indicators, and output various simulation results. The output of the scheme and the feedback output of the iterative optimization module are connected back to the feedback input of the ecological and humanistic dual-drive collaborative optimization module. The scheme output and iterative optimization module is used to generate a visual scheme based on the simulation results, perform multi-dimensional comparisons, provide users with an interactive editing interface, and perform iterative optimization according to users' optimization needs.

[0006] Furthermore, the ecological and humanistic dual-drive synergistic optimization module of the present invention internally includes: a gene pool evolution submodule and a narrative script arrangement submodule; wherein: The ecological constraint output of the ecological gene bank adaptive evolution submodule is connected to the spatial constraint input of the scene narrative dynamic arrangement submodule, and the cultural feature output of the scene narrative dynamic arrangement submodule is connected to the external guidance input of the ecological gene bank adaptive evolution submodule. The core scheme output terminal of the ecological gene bank adaptive evolution submodule and the spatial script output terminal of the scene narrative dynamic arrangement submodule are respectively connected to the first input terminal and the second input terminal of the internal collaborative integration unit of the ecological and humanistic dual-drive collaborative optimization module.

[0007] Furthermore, the multi-source data preprocessing module of the present invention internally comprises a multi-source data access unit, a data cleaning and calibration unit, a spatial data fusion unit, and a standard format output unit; wherein: The multi-source data access unit connects to various types of raw data from the site, including: measured topographic data, high-resolution satellite imagery, continuous meteorological monitoring data, and user requirement text materials; The data cleaning and calibration unit identifies missing fields and abnormal values ​​in the original data, uses interpolation to supplement the missing data, unifies the coordinate system of all spatial data to the National Geodetic Coordinate System 2000, and converts unstructured user requirement text into a set of structured tags. The spatial data fusion unit uses spatial registration technology to overlay and integrate different types of geospatial data to construct a unified database containing geographic information of the entire site. The standard format output unit converts the processed dataset into a format that subsequent modules of the system can directly read.

[0008] Furthermore, the basic rule constraint module of the present invention internally includes an industry standard parsing unit, a site baseline constraint unit, a functional requirement translation unit, and a rule consistency verification unit; wherein: The industry standard analysis unit is used to sort out relevant laws, regulations and technical standards, including the Urban and Rural Planning Law, the Urban Green Space Design Standards, and the Ecological Protection Red Line Management Measures, and extract the mandatory requirements for site plot ratio, green space ratio and building height restrictions, as well as the guiding principles for plant configuration and facility layout. The site baseline constraint unit is used to delineate prohibited construction areas, restricted construction areas, and suitable construction areas within a site based on natural attributes, including site topographic slope, soil bearing capacity, and hydrogeological conditions. Based on the project positioning and the user's specific needs, the functional requirements translation unit transforms the area ratio, spatial scale, and facility configuration requirements of the functional zones into quantifiable constraints. The functional zones include leisure areas, sports areas, and ecological conservation areas. The rule consistency verification unit is used to cross-check three types of rules: industry standards, site conditions, and functional requirements, identify and mark possible rule contradictions, and generate a rule consistency report.

[0009] Furthermore, the simulation verification module of the present invention internally includes a microclimate numerical calculation unit, an ecological process simulation unit, a spatial performance quantification unit, and a simulation result processing unit; wherein: The microclimate numerical calculation unit uses CFD numerical simulation technology, inputting site topography, vegetation distribution, and building layout data to calculate the distribution of sunshine duration, wind speed and direction, and air temperature and humidity during typical periods of the four seasons of spring, summer, autumn and winter. The ecological process simulation unit uses an ecosystem dynamics model to simulate the annual growth of plant communities, species succession trends, changes in biodiversity, and annual changes in carbon sequestration capacity within the site. The spatial performance quantification unit uses a spatial syntax model to analyze the spatial use performance indicators within the site, including pedestrian activity paths, visual accessibility, and spatial integration. The simulation results processing unit organizes the simulation results of three types—microclimate, ecology, and spatial performance—into standardized numerical reports and visualization charts such as heat maps and contour maps, and stores them in the system database.

[0010] Furthermore, the solution output and iterative optimization module of the present invention internally includes a solution visualization generation unit, a multi-dimensional comparison unit, an interactive adjustment and editing unit, and an optimization instruction generation unit; wherein: Based on the output layout scheme data, the scheme visualization generation unit automatically generates site visualization results, including master plan layout, 3D bird's-eye view, and walking walkthrough animation. The multi-dimensional comparison unit quantifies and ranks multiple candidate solutions generated at different iteration stages based on four dimensions: ecological indicators, spatial performance, cultural compatibility, and functional satisfaction. The interactive adjustment and editing unit provides a visual scheme editing interface, allowing users to directly drag and modify the functional zoning layout, plant species configuration, and facility location settings in the scheme; The optimization instruction generation unit transforms the user's adjustment requirements and the results of multiple scheme comparisons into optimization parameters and constraints that the system can recognize, and initiates a new round of scheme optimization process.

[0011] Furthermore, the gene pool evolution submodule of the present invention divides the site into grid units of a certain size, and the landscape elements of each unit are encoded with ecological gene sequences: the sequence length matches the total number of grids, and each gene position contains three dimensional encoded values: the plant dimension uses one-hot vectors to mark native tree / shrub / ground cover species, the road dimension uses continuous values ​​to represent road width, and the spatial scale dimension uses discrete values ​​to mark site functions.

[0012] Furthermore, the ecological gene sequence encoding formula within the gene pool evolution submodule of the present invention is as follows:

[0013] Among them, G represents the complete ecological gene sequence, which corresponds to a complete garden layout scheme; N is the total number of site grid cells; The plant species encoding vector for the i-th grid is labeled with a one-hot format to identify native plant types; The value represents the continuous road width of the i-th grid cell, ranging from 0 to 5. The i-th grid is used as a discrete marker for the site function, where 0 represents a green area, 1 represents a leisure plaza, and 2 represents a sports area. The formula for the dynamic multi-dimensional reward function is:

[0014] in, This is the total reward value for the PPO agent in each iteration, used to evaluate the overall performance of the layout scheme; These are dynamic weighting coefficients that are adaptively adjusted based on the site type. S represents the solar sunshine uniformity of the current scheme, with a value ranging from 0 to 1; The theoretical maximum uniformity of solar radiation on the site is calculated from the site boundaries and geographical location. V represents the ventilation efficiency of the current scheme, with a value ranging from 0 to 1; The theoretical maximum ventilation efficiency of the site is calculated based on the prevailing wind direction and topography. B represents the biodiversity index of the current scheme, with a value ranging from 0 to 10; C represents the annual carbon sequestration capacity of the current plan, expressed in kg / year; F represents the functional requirement matching degree of the current solution, with a value range of 0-1; PPO Strategy Update Formula:

[0015] in, These are the policy network parameters after the (k+1)th iteration; The network parameters for the current policy in the k-th iteration; The learning rate, ranging from 0.001 to 0.01; For the strategy in the kth round, in state The probability of generating gene sequence G; For the (k-1)th round policy in state The probability of generating gene sequence G; The advantage function value of gene sequence G is equal to the difference between the current reward value and the historical average reward value.

[0016] Furthermore, the narrative script arrangement submodule of the present invention calls the large language model to perform in-depth semantic analysis on the regional cultural characteristics of the site and the implicit experience needs of users, and extracts core cultural symbols and experience keywords; then LLM logically integrates cultural symbols and experience needs to generate a linear landscape script with humanistic narrative, while marking each node with quantitative layout constraints, and finally transforming the natural language script into a structured layout rule that can be parsed by the ecological gene bank submodule.

[0017] Furthermore, the formula for calculating the regional cultural element matching degree of the narrative script arrangement submodule of the present invention is as follows:

[0018] Where M represents the matching degree of regional cultural elements, with a value ranging from 0 to 1. The higher the value, the stronger the fit between the script's cultural symbols and the site's regional culture. n is the total number of cultural symbols contained in the script; The weight of the i-th cultural symbol is assigned based on the symbol's representativeness of the local culture of the site. For the i-th cultural symbol The semantic similarity with the site's regional cultural data H is calculated by the large language model using vector cosine similarity. The formula for calculating narrative coherence is:

[0019] C represents the narrative coherence score, ranging from 0 to 10. A higher value indicates a smoother connection between the spatial logic of the node sequence and the rhythm of the experience. m is the total number of landscape nodes contained in the script; The spatial connectivity between the j-th node and the (j+1)-th node ranges from 0 to 1 and is calculated from the road connectivity and line-of-sight accessibility between nodes. The complementarity of experiences from node j to node j+1, with a value ranging from 0 to 10, is calculated based on the differences in node functions and the degree of relevance to the theme.

[0020] The beneficial effects of this invention are: 1. In this invention, the ecological gene bank adaptive evolution submodule can extract core gene features from the site's native ecosystem and perform adaptive evolution, ensuring that the generated layout scheme closely matches the site's natural background conditions, guaranteeing the long-term stability of the plant community and the ecosystem's self-regulating ability. Simultaneously, it can automatically complete the quantitative optimization of ecological indicators, replacing the process of repeated manual adjustments, shortening the scheme generation cycle, and thus improving overall efficiency.

[0021] 2. In this invention, the scene narrative dynamic arrangement submodule, through the deep semantic analysis capabilities of a large language model, accurately uncovers the regional cultural heritage of the site and the implicit experiential needs of users. It integrates the humanistic storyline into the logic of the landscape layout, ensuring that the generated scheme is not merely a collection of isolated landscape nodes, but rather forms a spatial sequence with a continuous emotional experience, enhancing the site's cultural identity and emotional resonance. This allows the scheme to deeply resonate with regional cultural characteristics while also meeting users' personalized requirements for spatial experience, thereby increasing user acceptance of the design scheme. Attached Figure Description

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is an overall system block diagram of the present invention; Figure 2 This is a system block diagram of the ecological and humanistic dual-drive collaborative optimization module in this invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0024] Reference Figure 1-2 The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design includes: a multi-source data preprocessing module, a basic rule constraint module, a simulation verification module, a scheme output and iterative optimization module, and an ecological and humanistic dual-drive collaborative optimization module. The output of the multi-source data preprocessing module is simultaneously connected to the first input of the ecological and cultural dual-drive collaborative optimization module and the first input of the simulation verification module. The output of the basic rule constraint module is connected to the second input of the ecological and humanistic dual-drive collaborative optimization module; The main output of the ecological and humanistic dual-drive collaborative optimization module is connected to the first input of the scheme output and iterative optimization module, and the secondary output of the ecological and humanistic dual-drive collaborative optimization module is connected to the second input of the simulation verification module. The output of the simulation verification module is connected to the second input of the iterative optimization module; The output of the solution and the feedback output of the iterative optimization module are connected back to the feedback input of the ecological and humanistic dual-drive collaborative optimization module.

[0025] The ecological and cultural dual-driven synergistic optimization module includes: a gene pool evolution submodule and a narrative script arrangement submodule; The ecological constraint output of the ecological gene bank adaptive evolution submodule is connected to the spatial constraint input of the scene narrative dynamic arrangement submodule, and the cultural feature output of the scene narrative dynamic arrangement submodule is connected to the external guidance input of the ecological gene bank adaptive evolution submodule. The core scheme output end of the ecological gene bank adaptive evolution submodule and the spatial script output end of the scene narrative dynamic arrangement submodule are respectively connected to the first and second input ends of the internal collaborative integration unit of the ecological and humanistic dual-drive collaborative optimization module.

[0026] The multi-source data preprocessing module internally comprises a multi-source data access unit, a data cleaning and calibration unit, a spatial data fusion unit, and a standard format output unit. The multi-source data access unit interfaces with various raw data sources, including measured topographic data, high-resolution satellite imagery, continuous meteorological monitoring data, and user-required text documents, supporting direct import of multiple formats such as SHP, TIFF, and TXT. The data cleaning and calibration unit identifies missing fields and outliers in the raw data, uses interpolation to fill in the missing data, unifies the coordinate system of all spatial data to the National Geodetic Coordinate System 2000, and converts unstructured user-required text into a structured tag set. The spatial data fusion unit uses spatial registration technology to overlay and integrate different types of geospatial data, such as topography, vegetation, and hydrology, to construct a unified database containing comprehensive geographic information of the entire site. The standard format output unit converts the processed dataset into JSON and GeoJSON formats that can be directly read by subsequent modules of the system, enabling seamless data transfer between modules.

[0027] The basic rule constraint module is internally structured with an industry standard analysis unit, a site baseline constraint unit, a functional requirement translation unit, and a rule consistency verification unit. The industry standard analysis unit reviews relevant laws, regulations, and technical standards such as the Urban and Rural Planning Law, Urban Green Space Design Standards, and Ecological Protection Red Line Management Measures, extracting mandatory requirements regarding site floor area ratio, green space ratio, and building height restrictions, as well as guiding principles for plant configuration and facility layout. The site baseline constraint unit, based on natural attributes such as site topography, slope, soil bearing capacity, and hydrogeological conditions, delineates prohibited construction areas, restricted construction areas, and suitable construction areas within the site. The functional requirement translation unit, based on project positioning and clearly defined user needs, transforms the area ratio, spatial scale, and facility configuration requirements of functional zones such as recreational areas, sports areas, and ecological conservation areas into quantifiable constraint indicators. The rule consistency verification unit cross-checks the three types of rules—industry standards, site conditions, and functional requirements—identifying and marking potential rule contradictions and generating a rule consistency report.

[0028] The simulation verification module includes a microclimate numerical calculation unit, an ecological process simulation unit, a spatial performance quantification unit, and a simulation results processing unit. The microclimate numerical calculation unit uses CFD numerical simulation technology, inputting data such as site topography, vegetation distribution, and building layout to calculate the distribution of sunshine duration, wind speed and direction, and air temperature and humidity during typical periods of the four seasons. The ecological process simulation unit uses an ecosystem dynamics model to simulate the annual growth of plant communities, species succession trends, biodiversity changes, and annual variations in carbon sequestration capacity within the site. The spatial performance quantification unit uses a spatial syntax model to analyze spatial use performance indicators such as pedestrian activity paths, visual accessibility, and spatial integration within the site. The simulation results processing unit organizes the results of the microclimate, ecological, and spatial performance simulations into standardized numerical reports and visual charts such as heat maps and contour maps, storing them in the system database.

[0029] The scheme output and iterative optimization module internally includes a scheme visualization generation unit, a multi-dimensional comparison unit, an interactive adjustment and editing unit, and an optimization command generation unit. The scheme visualization generation unit automatically generates site master plans, 3D bird's-eye view renderings, and walking animations based on the layout scheme data output by the core optimization module, supporting export in JPG, PNG, MP4, and other formats. The multi-dimensional comparison unit quantitatively scores and ranks multiple candidate schemes generated at different iteration stages based on four dimensions: ecological indicators, spatial performance, cultural compatibility, and functional satisfaction. The interactive adjustment and editing unit provides a visual scheme editing interface, allowing users to directly drag and modify elements such as functional zoning layouts, plant species configurations, and facility locations. The optimization command generation unit transforms user adjustment requirements and multi-scheme comparison results into system-recognizable optimization parameters and constraints, transmitting them to the core optimization module to initiate a new round of scheme optimization.

[0030] The gene pool evolution submodule divides the site into 10m×10m grid units. Landscape elements in each unit are encoded using ecological gene sequences: the sequence length matches the total number of 100 grid units, and each gene locus contains three dimensional encoded values—the plant dimension uses one-hot vectors to label native tree / shrub / ground cover species (e.g., [1,0,0][1,0,0] represents Chinese scholar tree), the road dimension uses continuous values ​​to represent road width (0 corresponds to non-road areas, 1-5 correspond to widths of 1-5m), and the spatial scale dimension uses discrete values ​​to label site functions (0=green area, 1=leisure plaza, 2=sports area). The system uses the PPO algorithm as the core optimization framework. Each iteration of the agent generates a complete gene sequence corresponding to a layout scheme. The simulation module obtains four quantitative indicators: solar uniformity, ventilation efficiency, biodiversity index, and carbon sequestration capacity. Simultaneously, a dynamic reward function is constructed by combining functional requirement matching degree (e.g., the compliance rate of the sports area area ratio). The agent continuously adjusts the coding combination of gene sequences by interacting with the simulation environment. When the reward value increases by less than 1% in 5 consecutive iterations, the evolution is terminated and the layout scheme corresponding to the optimal gene sequence is output.

[0031] The coding formula for the ecological gene sequence within the gene pool evolution submodule is as follows:

[0032] G represents a complete ecological gene sequence, corresponding to a complete garden layout scheme; N is the total number of site grid cells; The plant species encoding vector for the i-th grid is labeled with a one-hot format to identify native plant types; The value is a continuous value for the road width of the i-th grid, ranging from 0 (non-road) to 5 (5m wide road); The i-th grid is used as a discrete marker for the site function, where 0 represents a green area, 1 represents a leisure plaza, and 2 represents a sports area. The formula for the dynamic multi-dimensional reward function is:

[0033] This is the total reward value for the PPO agent in each iteration, used to evaluate the overall performance of the layout scheme; These are dynamic weighting coefficients that are adaptively adjusted based on the site type. S represents the solar sunshine uniformity of the current scheme, with a value ranging from 0 to 1; The theoretical maximum uniformity of solar radiation on the site is calculated from the site boundaries and geographical location. V represents the ventilation efficiency of the current scheme, with a value ranging from 0 to 1; The theoretical maximum ventilation efficiency of the site is calculated based on the prevailing wind direction and topography. B represents the biodiversity index of the current scheme, with a value ranging from 0 to 10; C represents the annual carbon sequestration capacity of the current plan, expressed in kg / year; F represents the functional requirement matching degree of the current solution, with a value range of 0-1 (1 represents that the user's functional requirements are fully met). PPO Strategy Update Formula (Adapted to Landscape Layout Scenarios):

[0034] These are the policy network parameters after the (k+1)th iteration; The network parameters for the current policy in the k-th iteration; The learning rate, ranging from 0.001 to 0.01; For the strategy in the kth round, in state The probability of generating gene sequence G; For the (k-1)th round policy in state The probability of generating gene sequence G; The advantage function value of gene sequence G is equal to the difference between the current reward value and the historical average reward value.

[0035] The narrative script arrangement submodule invokes a large language model to conduct in-depth semantic analysis of the site's regional cultural characteristics and users' implicit experience needs, extracting core cultural symbols and experience keywords. Subsequently, LLM logically integrates cultural symbols and experience needs to generate a linear landscape script with humanistic narrative, while labeling each node with quantified layout constraints. Finally, the natural language script is transformed into structured layout rules that can be parsed by the ecological gene bank submodule.

[0036] The formula for calculating the matching degree of regional cultural elements in the narrative script arrangement submodule is as follows:

[0037] In the formula: M represents the matching degree of regional cultural elements, with a value ranging from 0 to 1. The higher the value, the stronger the fit between the script's cultural symbols and the site's regional culture. n is the total number of cultural symbols contained in the script; The weight of the i-th cultural symbol is assigned based on the symbol's representativeness of the local culture of the site. For the i-th cultural symbol The semantic similarity with the site's regional cultural data H is calculated by the large language model using vector cosine similarity. The formula for calculating narrative coherence is:

[0038] In the formula: C is the experience narrative coherence score, with a value range of 0 to 10. The higher the value, the smoother the connection between the spatial logic of the node sequence and the rhythm of the experience. m is the total number of landscape nodes contained in the script; The spatial connectivity between the j-th node and the (j+1)-th node ranges from 0 to 1 and is calculated from the road connectivity and line-of-sight accessibility between nodes. The complementarity of experiences from node j to node j+1, with a value ranging from 0 to 10, is calculated based on the differences in node functions and the degree of relevance to the theme.

[0039] In summary, the ecological gene bank adaptive evolution submodule of this invention can extract core gene features from the site's native ecosystem and perform adaptive evolution, ensuring that the generated layout scheme closely matches the site's natural background conditions, guaranteeing the long-term stability of the plant community and the ecosystem's self-regulating ability. Simultaneously, it can automatically complete the quantitative optimization of ecological indicators, replacing the process of repeated manual adjustments, shortening the scheme generation cycle, and thus improving overall efficiency.

[0040] In this invention, the scene narrative dynamic arrangement submodule, through the deep semantic analysis capabilities of a large language model, accurately uncovers the regional cultural heritage of the site and the implicit experiential needs of users. It integrates the humanistic storyline into the logic of the landscape layout, ensuring that the generated scheme is not merely a collection of isolated landscape nodes, but rather forms a spatial sequence with a continuous emotional experience, enhancing the site's cultural identity and emotional resonance. This allows the scheme to deeply resonate with regional cultural characteristics while also meeting users' personalized requirements for spatial experience, thereby increasing user acceptance of the design scheme.

[0041] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0042] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0043] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. An AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design, characterized in that, The system includes: a multi-source data preprocessing module, a basic rule constraint module, a simulation verification module, a scheme output and iterative optimization module, and an ecological and humanistic dual-driven collaborative optimization module; among which: The output of the multi-source data preprocessing module is simultaneously connected to the first input of the ecological and humanistic dual-drive collaborative optimization module and the first input of the simulation verification module; the multi-source data preprocessing module is used to acquire multi-source data for landscape planning and design, and to perform data cleaning and spatial data fusion to obtain data in a standard format for output; The output of the basic rule constraint module is connected to the second input of the ecological and humanistic dual-drive collaborative optimization module; the basic rule constraint module is used to analyze industry standards for landscape architecture planning and design, establish site baseline constraints, and transform functional requirements into quantifiable constraint indicators. The main output of the ecological and humanistic dual-drive collaborative optimization module is connected to the first input of the scheme output and iterative optimization module, and the secondary output of the ecological and humanistic dual-drive collaborative optimization module is connected to the second input of the simulation verification module. The output of the simulation verification module is connected to the second input of the scheme output and iterative optimization module; the simulation verification module is used to calculate the microclimate values ​​of the site at different times using CFD numerical simulation technology, simulate the ecological processes in the site, analyze the space use performance indicators, and output various simulation results. The output of the scheme and the feedback output of the iterative optimization module are connected back to the feedback input of the ecological and humanistic dual-drive collaborative optimization module. The scheme output and iterative optimization module is used to generate a visual scheme based on the simulation results, perform multi-dimensional comparisons, provide users with an interactive editing interface, and perform iterative optimization according to users' optimization needs.

2. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 1, characterized in that, The ecological and humanistic dual-driven synergistic optimization module internally includes: a gene pool evolution submodule and a narrative script arrangement submodule; wherein: The ecological constraint output of the ecological gene bank adaptive evolution submodule is connected to the spatial constraint input of the scene narrative dynamic arrangement submodule, and the cultural feature output of the scene narrative dynamic arrangement submodule is connected to the external guidance input of the ecological gene bank adaptive evolution submodule. The core scheme output terminal of the ecological gene bank adaptive evolution submodule and the spatial script output terminal of the scene narrative dynamic arrangement submodule are respectively connected to the first input terminal and the second input terminal of the internal collaborative integration unit of the ecological and humanistic dual-drive collaborative optimization module.

3. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 1, characterized in that, The multi-source data preprocessing module internally includes a multi-source data access unit, a data cleaning and calibration unit, a spatial data fusion unit, and a standard format output unit; wherein: The multi-source data access unit connects to various types of raw data from the site, including: measured topographic data, high-resolution satellite imagery, continuous meteorological monitoring data, and user requirement text materials; The data cleaning and calibration unit identifies missing fields and abnormal values ​​in the original data, uses interpolation to supplement the missing data, unifies the coordinate system of all spatial data to the National Geodetic Coordinate System 2000, and converts unstructured user requirement text into a set of structured tags. The spatial data fusion unit uses spatial registration technology to overlay and integrate different types of geospatial data to construct a unified database containing geographic information of the entire site. The standard format output unit converts the processed dataset into a format that subsequent modules of the system can directly read.

4. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 1, characterized in that, The basic rule constraint module internally includes an industry standard parsing unit, a site baseline constraint unit, a functional requirement translation unit, and a rule consistency verification unit; wherein: The industry standard analysis unit is used to sort out relevant laws, regulations and technical standards, including the Urban and Rural Planning Law, the Urban Green Space Design Standards, and the Ecological Protection Red Line Management Measures, and extract the mandatory requirements for site plot ratio, green space ratio and building height restrictions, as well as the guiding principles for plant configuration and facility layout. The site baseline constraint unit is used to delineate prohibited construction areas, restricted construction areas, and suitable construction areas within a site based on natural attributes, including site topographic slope, soil bearing capacity, and hydrogeological conditions. Based on the project positioning and the user's specific needs, the functional requirements translation unit transforms the area ratio, spatial scale, and facility configuration requirements of the functional zones into quantifiable constraints. The functional zones include leisure areas, sports areas, and ecological conservation areas. The rule consistency verification unit is used to cross-check three types of rules: industry standards, site conditions, and functional requirements, identify and mark possible rule contradictions, and generate a rule consistency report.

5. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 1, characterized in that, The simulation verification module internally includes a microclimate numerical calculation unit, an ecological process simulation unit, a spatial performance quantification unit, and a simulation result processing unit; wherein: The microclimate numerical calculation unit uses CFD numerical simulation technology, inputting site topography, vegetation distribution, and building layout data to calculate the distribution of sunshine duration, wind speed and direction, and air temperature and humidity during typical periods of the four seasons of spring, summer, autumn and winter. The ecological process simulation unit uses an ecosystem dynamics model to simulate the annual growth of plant communities, species succession trends, changes in biodiversity, and annual changes in carbon sequestration capacity within the site. The spatial performance quantification unit uses a spatial syntax model to analyze the spatial use performance indicators within the site, including pedestrian activity paths, visual accessibility, and spatial integration. The simulation results processing unit organizes the simulation results of three types—microclimate, ecology, and spatial performance—into standardized numerical reports and visualization charts such as heat maps and contour maps, and stores them in the system database.

6. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 1, characterized in that, The scheme output and iterative optimization module internally includes a scheme visualization generation unit, a multi-dimensional comparison unit, an interactive adjustment and editing unit, and an optimization instruction generation unit; wherein: Based on the output layout scheme data, the scheme visualization generation unit automatically generates site visualization results, including master plan layout, 3D bird's-eye view, and walking walkthrough animation. The multi-dimensional comparison unit quantifies and ranks multiple candidate solutions generated at different iteration stages based on four dimensions: ecological indicators, spatial performance, cultural compatibility, and functional satisfaction. The interactive adjustment and editing unit provides a visual scheme editing interface, allowing users to directly drag and modify the functional zoning layout, plant species configuration, and facility location settings in the scheme; The optimization instruction generation unit transforms the user's adjustment requirements and the results of multiple scheme comparisons into optimization parameters and constraints that the system can recognize, and initiates a new round of scheme optimization process.

7. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 2, characterized in that, The gene pool evolution submodule divides the site into grid units of a certain size. The landscape elements of each unit are encoded with ecological gene sequences: the sequence length matches the total number of grids, and each gene locus contains three dimensional encoded values: the plant dimension uses one-hot vectors to mark native tree / shrub / ground cover species, the road dimension uses continuous values ​​to represent road width, and the spatial scale dimension uses discrete values ​​to mark site functions.

8. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 7, characterized in that, The encoding formula for the ecological gene sequence within the gene pool evolution submodule is as follows: Among them, G represents the complete ecological gene sequence, which corresponds to a complete garden layout scheme; N is the total number of site grid cells; The plant species encoding vector for the i-th grid is labeled with a one-hot format to identify native plant types; The value represents the continuous road width of the i-th grid cell, ranging from 0 to 5. The i-th grid is used as a discrete marker for the site function, where 0 represents a green area, 1 represents a leisure plaza, and 2 represents a sports area. The formula for the dynamic multi-dimensional reward function is: in, This is the total reward value for the PPO agent in each iteration, used to evaluate the overall performance of the layout scheme; These are dynamic weighting coefficients that are adaptively adjusted based on the site type. S represents the solar sunshine uniformity of the current scheme, with a value ranging from 0 to 1; The theoretical maximum uniformity of solar radiation on the site is calculated from the site boundaries and geographical location. V represents the ventilation efficiency of the current scheme, with a value ranging from 0 to 1; The theoretical maximum ventilation efficiency of the site is calculated based on the prevailing wind direction and topography. B represents the biodiversity index of the current scheme, with a value ranging from 0 to 10; C represents the annual carbon sequestration capacity of the current plan, expressed in kg / year; F represents the functional requirement matching degree of the current solution, with a value range of 0-1; PPO Strategy Update Formula: in, These are the policy network parameters after the (k+1)th iteration; The network parameters for the current policy in the k-th iteration; The learning rate, ranging from 0.001 to 0.01; For the strategy in the kth round, in state The probability of generating gene sequence G; For the (k-1)th round policy in state The probability of generating gene sequence G; The advantage function value of gene sequence G is equal to the difference between the current reward value and the historical average reward value.

9. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 2, characterized in that, The narrative script arrangement submodule calls the large language model to conduct in-depth semantic analysis of the site's regional cultural characteristics and users' implicit experience needs, extracting core cultural symbols and experience keywords; then LLM logically integrates cultural symbols and experience needs to generate a linear landscape script with humanistic narrative, while marking each node with quantitative layout constraints, and finally transforming the natural language script into a structured layout rule that can be parsed by the ecological gene bank submodule.

10. The AI-powered intelligent layout and landscape optimization system for landscape architecture planning and design according to claim 9, characterized in that, The formula for calculating the matching degree of regional cultural elements in the narrative script arrangement submodule is as follows: Where M represents the matching degree of regional cultural elements, with a value ranging from 0 to 1. The higher the value, the stronger the fit between the script's cultural symbols and the site's regional culture. n is the total number of cultural symbols contained in the script; The weight of the i-th cultural symbol is assigned based on the symbol's representativeness of the local culture of the site. For the i-th cultural symbol The semantic similarity with the site's regional cultural data H is calculated by the large language model using vector cosine similarity. The formula for calculating narrative coherence is: C represents the narrative coherence score, ranging from 0 to 10. A higher value indicates a smoother connection between the spatial logic of the node sequence and the rhythm of the experience. m is the total number of landscape nodes contained in the script; The spatial connectivity between the j-th node and the (j+1)-th node ranges from 0 to 1 and is calculated from the road connectivity and line-of-sight accessibility between nodes. The complementarity of experiences from node j to node j+1, with a value ranging from 0 to 10, is calculated based on the differences in node functions and the degree of relevance to the theme.

Images