Virtual simulation method and system for village scene planning

Abstract

The invention discloses a village scene planning virtual simulation method and system, and relates to the technical field of scene planning, and the method comprises the steps: carrying out the spectral clustering of village structured graph data, obtaining a village function space, collecting village historical passenger flow data, and carrying out the optimization of the village function space through the combination of a multi-objective evolutionary algorithm, the method comprises the steps of generating an initial village scene planning scheme, simulating a behavior decision and interaction mode of village crowds by adopting a PPO model based on the initial village scene planning scheme, generating a people flow simulation evaluation report, performing dynamic feedback optimization on the initial village scene planning scheme by using the people flow simulation evaluation report, and obtaining an optimized village function planning scheme. The village planning method ensures that the village scene planning can meet the current demand and can also adapt to the future change trend, and meanwhile, the accuracy, scientificity and flexibility of planning are enhanced.

Description

Technical Field

[0001] This invention relates to the field of scene planning technology, and in particular to a virtual simulation method and system for scene planning of scenic villages. Background Technology

[0002] With the rapid development of virtual simulation technology, it has been widely applied in urban planning, architectural design, and cultural tourism. In particular, in the planning of scenic villages (a combination of landscape and village), GIS technology provides rich spatial data analysis capabilities, enabling planners to understand the topography, land use, and transportation network elements of the region from a macro perspective. This transforms scenic village planning from a two-dimensional static display to a three-dimensional dynamic simulation, providing decision-makers with a more intuitive and comprehensive perspective. In addition, the fusion of multi-source heterogeneous data further enhances the accuracy and practicality of the model, enabling the planning scheme to not only reflect the current situation but also predict future trends, thereby improving the quality of planning decisions.

[0003] Nevertheless, existing methods for planning scenic villages still have room for improvement. First, most current virtual simulation tools fail to fully integrate real-world images with GIS data, resulting in models that lack sufficient detail and accuracy. This also limits the depth of understanding of structured map data of scenic villages and ignores the impact of multi-dimensional factors such as historical visitor flow data and peak-hour pedestrian flow distribution on the planning results, making it impossible to optimize the planning scheme in real time based on actual operating conditions. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a virtual simulation method for planning scenic villages to solve the problem of not being able to optimize the planning scheme based on the operation of the scenic village.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a virtual simulation method for planning scenic village scenes, comprising, Collect real-world images and GIS data of the scenic village, and preprocess them to generate multi-source heterogeneous data; Based on multi-source heterogeneous data, the Graphormer model is used for semantic parsing to generate Jingcun structured graph data; Spectral clustering was performed on the structured map data of Jingcun to obtain the functional space of Jingcun. Historical visitor flow data of the scenic village was collected and combined with a multi-objective evolutionary algorithm to optimize the functional space of the scenic village and generate an initial scenic village scene planning scheme. Based on the initial village scene planning scheme, the PPO model is used to simulate the behavioral decision-making and interaction patterns of the village crowd, and a pedestrian flow simulation evaluation report is generated. Using the pedestrian flow simulation evaluation report, the initial scenic village scene planning scheme is dynamically optimized based on feedback, resulting in an optimized scenic village functional planning scheme.

[0007] As a preferred embodiment of the virtual simulation method for village scene planning described in this invention, the step of collecting real-world village images and GIS data, and preprocessing them to generate multi-source heterogeneous data, specifically involves... Drones were used to capture aerial images of Jingcun and Jingcun GIS data was extracted from the OpenStreetMap platform. Distortion correction was performed on the real-world images of Jingcun, and attribute standardization was performed on the GIS data of Jingcun to form a multi-source heterogeneous dataset.

[0008] As a preferred embodiment of the virtual simulation method for scene planning of the scenic village described in this invention, the step of generating structured graph data of the scenic village based on multi-source heterogeneous data and using the Graphormer model for semantic parsing specifically involves... The multi-source heterogeneous dataset is transformed into a graph structure and input into the Graphormer model. The global self-attention mechanism of the Graphormer model integrates different spatial semantic information in the graph structure. Multi-head attention calculates the correlation between real-world images and GIS data. Finally, in the encoding stage, the Graphormer model dynamically aggregates the multi-source features of real-world images and GIS data to generate structured graph data of the landscape village.

[0009] As a preferred embodiment of the virtual simulation method for scene planning of scenic villages described in this invention, the step of performing spectral clustering on the structured map data of the scenic villages to obtain the functional space of the scenic villages specifically involves: Node adjacency matrix and node feature matrix are constructed based on Jingcun structured graph data; A Gaussian kernel function is used to calculate the similarity matrix from the node adjacency matrix and the node feature matrix. The similarity matrix is ​​then normalized to form a Laplace matrix. The Laplacian matrix is ​​decomposed into eigenvalues, and the eigenvectors corresponding to the k smallest non-zero eigenvalues ​​are selected to form a feature matrix. The K-means algorithm is then used to perform vector clustering on the feature matrix to generate the Jingcun scene space.

[0010] As a preferred embodiment of the virtual simulation method for scenic village scene planning described in this invention, the step of collecting historical visitor flow data of the scenic village and optimizing the functional space of the scenic village using a multi-objective evolutionary algorithm to generate an initial scenic village scene planning scheme specifically involves: Collect historical visitor flow data for the scenic area to extract visitor flow distribution, peak periods, and length of stay, and establish multi-objective optimization directions based on the functional spaces of the scenic area; Based on the multi-objective optimization approach, a multi-objective evolutionary algorithm is used to perform cross- and mutation optimization operations on the functional space layout of the scenic village to generate an initial scenic village scene planning scheme.

[0011] As a preferred embodiment of the virtual simulation method for village scene planning described in this invention, the step of using a PPO model to simulate the behavioral decision-making and interaction patterns of village crowds based on an initial village scene planning scheme, and generating a crowd flow simulation evaluation report, specifically involves... Based on the initial scenic village functional planning scheme, after constructing the state space, action space and reward function for the PPO model, individual tourists are modeled as intelligent agents in the PPO model. The PPO model simulates the behavioral decision-making and interaction patterns of intelligent agents in the scenic village functional space through policy network and value network. Based on the intelligent agent's behavioral decisions and interaction patterns in the functional spaces of the scenic village, the density of tourist distribution, path overlap rate, and facility utilization rate are recorded, and congestion analysis is conducted to generate a pedestrian flow simulation assessment report.

[0012] As a preferred embodiment of the virtual simulation method for village scene planning described in this invention, the step of using a pedestrian flow simulation evaluation report to dynamically optimize the initial village scene planning scheme and obtain an optimized village functional planning scheme specifically involves: Based on the pedestrian flow simulation evaluation report, the planning shortcomings in the initial scenic village scene planning scheme were identified and dynamically modified to obtain the modified scenic village scene planning scheme. The modified Jingcun functional planning scheme was combined with the PPO model for iterative simulation to verify whether the modification could alleviate congestion and improve overall mobility. Through multiple dynamic feedback loops, the distribution of people and functional needs were gradually balanced to form an optimized Jingcun functional planning scheme.

[0013] Secondly, the present invention provides a virtual simulation system for scenic village planning, comprising, The data acquisition module collects real-world images and GIS data of the scenic village, and performs preprocessing to generate multi-source heterogeneous data; The semantic parsing module uses the Graphormer model to perform semantic parsing based on multi-source heterogeneous data, generating Jingcun structured graph data; The clustering module performs spectral clustering on the Jingcun structured graph data to obtain the Jingcun functional space; The solution module collects historical visitor flow data of the scenic village and combines it with a multi-objective evolutionary algorithm to optimize the functional space of the scenic village and generate an initial scenic village scene planning scheme. The evaluation report module, based on the initial village scene planning scheme, uses the PPO model to simulate the behavioral decision-making and interaction patterns of the village crowd and generates a pedestrian flow simulation evaluation report. The optimization module uses the pedestrian flow simulation evaluation report to dynamically optimize the initial scenic village scene planning scheme and obtain the optimized scenic village functional planning scheme.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the virtual simulation method for scene planning as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the virtual simulation method for scene planning of a scenic village as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: The Graphormer model performs semantic parsing, deeply integrates real-world images and GIS data, enhances the understanding of the structured map data of the village, supports more accurate functional area division, improves the quality of planning decisions, automatically identifies potential functional areas using spectral clustering analysis, reduces human interference, makes the planning results more objective and reasonable, and improves the scientific nature and planning efficiency of zoning. Finally, through a dynamic feedback optimization mechanism based on the pedestrian flow simulation evaluation report, the initial planning scheme is continuously improved, effectively solving the congestion problem in actual operation and improving overall mobility. This ensures that the planning scheme can meet current needs and adapt to future changing trends, thereby improving the sustainability of the village planning method and enhancing the accuracy, scientific nature, and flexibility of village scene planning. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart for planning a virtual simulation method for the Jingcun scene.

[0019] Figure 2 A flowchart for planning the virtual simulation methodology architecture for the Jingcun scene.

[0020] Figure 3 A flowchart for generating the functional spaces of the scenic village.

[0021] Figure 4 A flowchart for generating the initial village scene planning scheme. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 As an embodiment of the present invention, this embodiment provides a virtual simulation method for planning a scenic village scene, comprising the following steps: S1. Collect real-world images and GIS data of the scenic village, and preprocess them to generate multi-source heterogeneous data; Drones were used to capture aerial images of Jingcun and Jingcun GIS data was extracted from the OpenStreetMap platform. It should be noted that when collecting aerial images of Jingcun using drones, a multi-rotor drone equipped with a high-definition oblique photography camera was used to conduct low-altitude aerial photography of the entire Jingcun area along a five-directional flight path, acquiring high-resolution images of the village landscape, including building facades, road networks, and vegetation distribution. Simultaneously, vector-format GIS data (including highway=*road network, building=*building outlines, and landuse=*land use type) was downloaded from the OpenStreetMap platform via the Overpass API.

[0026] Distortion correction was performed on the real-world images of Jingcun, and attribute standardization was performed on the GIS data of Jingcun to form a multi-source heterogeneous dataset; Furthermore, based on the radial and tangential distortion coefficients from the high-definition oblique photography camera, a Brown-Conrad distortion model was used to model the Jingcun real-scene image. This involved substituting the radial and tangential distortion coefficients from the high-definition oblique photography camera into the Brown-Conrad distortion model. The coordinate transformation of each pixel in the Jingcun real-scene image was performed using the inverse mapping algorithm of the Brown-Conrad distortion model to eliminate radial and tangential distortion, finally outputting the distortion-corrected Jingcun real-scene image. The GIS data extracted from the OpenStreetMap platform was reconstructed using the ogr2ogr tool and unified to the CGCS2000 coordinate system. For example, highway=*road network was converted to line features, retaining road attributes (such as road type, name, number of lanes, and speed limit); building=*building outline was converted to polygon features, retaining building attributes (such as usage, area, and floor height). The distortion-corrected and attribute-reconstructed Jingcun real-scene image was integrated with the Jingcun GIS data into a multi-source heterogeneous dataset.

[0027] S2. Based on multi-source heterogeneous data, use the Graphormer model for semantic parsing to generate Jingcun structured graph data; The multi-source heterogeneous dataset is transformed into a graph structure and input into the Graphormer model. The global self-attention mechanism of the Graphormer model integrates different spatial semantic information in the graph structure. Multi-head attention calculates the correlation between real-world images and GIS data. Finally, in the encoding stage, the Graphormer model dynamically aggregates the multi-source features of real-world images and GIS data to generate structured graph data of the scenic village. It should be noted that the real-world image of the scenic village is segmented into regular grid cells, with each regular grid cell serving as a node. An edge structure is formed based on the nodes and the GIS data within them; for example, edges are established between a node and the buildings within its range.

[0028] The structures of adjacent village scenes are collected and standardized before being used as the training set for the Graphormer model. During training, the global self-attention mechanism of the Graphormer model learns the association weights between the texture features of real-world images and the geometric features of GIS data in the structures of adjacent village scenes. The multi-head attention layer computes the node relationships in different semantic spaces in parallel. The parameter matrix of the Graphormer model is updated in each iteration. Finally, the Graphormer model can accurately predict the spatial semantic associations of buildings, roads and vegetation in the village scene, resulting in the trained Graphormer model.

[0029] The village map structure is input into the trained Graphormer model. The Graphormer model's global self-attention mechanism calculates the cross-modal attention weights between nodes in the graph structure and the village's GIS data. Multi-head attention layers capture node association patterns in different semantic spaces in parallel. Visual features (real-world images of the village) and geographic features (GIS data) are dynamically fused through residual connections and layer normalization. The encoder of the Graphormer model outputs the village's structured map data after aggregation by the graph readout layer. The nodes in the village's structured map data contain road and building attributes reconstructed from the GIS data.

[0030] S3. Perform spectral clustering on the Jingcun structured map data to obtain the Jingcun functional space; Node adjacency matrix and node feature matrix are constructed based on Jingcun structured graph data; It should be noted that, based on the nodes and edges recorded in the Jingcun structured graph data, each node is assigned a unique number. All nodes in the Jingcun structured graph data are connected to form an N×N node adjacency matrix (N represents the total number of nodes). If two nodes in the Jingcun structured graph data are connected, the corresponding position in the node adjacency matrix is ​​set to 1; otherwise, it is set to 0. This is to transform the topological structure of the Jingcun structured graph data into a numerical matrix representation. The connection between two nodes can be understood as follows: for example, if nodes represent building outlines and are actually adjacent, they are considered connected. Simultaneously, based on the attributes within each node in the Jingcun structured graph data, an N×D node feature matrix is ​​formed in node number order, where D represents the number of features (attributes).

[0031] A Gaussian kernel function is used to calculate the similarity matrix from the node adjacency matrix and the node feature matrix. The similarity matrix is ​​then normalized to form a Laplace matrix. It should be noted that a Gaussian kernel function is applied to each row vector in the node feature matrix to calculate the similarity between each pair of nodes, generating a similarity matrix, which can be expressed by the formula: ; in, Represents the similarity matrix. This represents the natural exponential function. This represents the squared Euclidean distance between the two nodes. This represents the bandwidth parameter of the Gaussian kernel; The degree value of each node in the similarity matrix is ​​calculated based on the similarity matrix. That is, the sum of the elements in each row of the similarity matrix is ​​the degree value of the corresponding node. The degree values ​​of all nodes are arranged in order on a diagonal, and the remaining positions are 0, thus obtaining the degree matrix. The similarity matrix is ​​symmetrically normalized, i.e., calculated using the degree matrix and the similarity matrix, to obtain the Laplacian matrix, which is expressed by the formula: ; ; in, Represents the Laplace matrix, Represents the identity matrix. Degree matrix, Represents the similarity matrix; The Laplacian matrix is ​​subjected to eigenvalue decomposition. The eigenvectors corresponding to the k smallest non-zero eigenvalues ​​are selected to form a feature matrix. The K-means algorithm is used to perform vector clustering on the feature matrix to generate the Jingcun scene space. Furthermore, eigenvalue decomposition is performed based on the normalized Laplace matrix to solve for the eigenvalues ​​and eigenvectors of the Laplace matrix. The eigenvalues ​​of the Laplace matrix are obtained by solving the characteristic equation, expressed by the formula: ; in, Represents the determinant of the Laplace matrix. Represents the Laplace matrix, Represents eigenvalues. Represents the identity matrix; The eigenvectors of the Laplace matrix are obtained by solving a system of linear equations, expressed by the following formula: ; in, Represents the Laplace matrix, Represents eigenvalues. Represents the identity matrix. Represents the eigenvector; After sorting the eigenvalues ​​of the Laplacian matrix in ascending order, the eigenvectors corresponding to the k smallest non-zero eigenvalues ​​are selected. These selected eigenvectors are then combined column-wise to form a multidimensional feature matrix, with each row corresponding to a node in the Jingcun structured map data. K-means clustering analysis is then performed on the rows of the multidimensional feature matrix, specifically, k cluster centers are randomly initialized. The Euclidean distance from each node's row vector to each cluster center is calculated, and the node is assigned to the cluster corresponding to the nearest Euclidean distance cluster center. During the iteration phase, the mean of the row vectors corresponding to all nodes in each cluster is recalculated as the new cluster center, and the node assignment process is repeated until the maximum number of iterations is reached. Finally, each row of the multidimensional feature matrix is ​​assigned to a corresponding cluster, forming k non-overlapping clusters. Each cluster represents a functional area of ​​Jingcun, ultimately forming a Jingcun scene space with k functional areas.

[0032] S4. Collect historical visitor flow data of the scenic village and combine it with a multi-objective evolutionary algorithm to optimize the functional space of the scenic village and generate an initial scenic village scene planning scheme. Collect historical visitor flow data for the scenic area to extract visitor flow distribution, peak periods, and length of stay, and establish multi-objective optimization directions based on the functional spaces of the scenic area; It should be noted that after collecting historical visitor flow data for the scenic area, daily visitor flow distribution was extracted through time series analysis to identify peak visitor periods and the average dwell time of tourists in each functional space. Spatial overlay analysis was then performed on the visitor flow distribution and the layout of the scenic area's functional spaces. Specifically, based on the degree of tourist concentration in the dining area, viewing area, and rest area at different times during the visitor flow distribution, the attraction range of each functional space was divided using Voronoi diagrams. Spatial autocorrelation tests were used to identify the matching relationship between visitor flow distribution and functional space capacity. After discovering the spatiotemporal coupling patterns between high-density visitor areas and areas with narrow passages and insufficient service facilities, a multi-objective optimization direction for the scenic area was established, focusing on "reducing the congestion index during peak hours," "balancing the dwell time in each functional space," and "improving the accessibility of core attractions."

[0033] Based on the multi-objective optimization direction, a multi-objective evolutionary algorithm is used to perform cross and mutation optimization operations on the functional space layout of the scenic village to generate an initial scenic village scene planning scheme. Furthermore, a multi-objective evolutionary algorithm is used to initialize the population of functional areas within the scenic village's functional space. Each individual in the population encodes a layout scheme for the scenic village's functional space. Parent individuals (selected layout schemes) are chosen from all layout schemes through non-dominated sorting and congestion calculation. A simulated binary crossover operator is used to perform path crossover and recombination on the selected layout schemes. A multinomial mutation operator is used to randomly perturb the location and area parameters of the dining area, viewing area, and rest area within the selected layout schemes to enhance diversity. After crossover and mutation, the multi-objective evolutionary algorithm simultaneously calculates the quantified values ​​of three objectives for each layout scheme during peak hours: congestion index, functional space dwell time balance (using the Gini coefficient), and accessibility to core attractions. The algorithm then outputs a Pareto front solution set. The layout scheme with the best overall performance is selected from the Pareto front solution set as the initial scenic village scene planning scheme.

[0034] S5. Based on the initial village scene planning scheme, use the PPO model to simulate the behavioral decision-making and interaction patterns of the village crowd and generate a crowd flow simulation evaluation report. Based on the initial scenic village functional planning scheme, after constructing the state space, action space and reward function for the PPO model, individual tourists are modeled as intelligent agents in the PPO model. The PPO model simulates the behavioral decision-making and interaction patterns of intelligent agents in the scenic village functional space through policy network and value network. It should be noted that, based on the functional zoning data and path topology of the initial scenic village functional planning scheme, the state space (including tourist location coordinates, target attraction direction vector, and surrounding real-time pedestrian density), action space (tourist movement direction and tourist dwell time), and reward function (designed as a weighted combination of visitor comfort, functional matching degree, and movement energy consumption) of the PPO model are first constructed. Subsequently, historical tourist behavior data of the scenic village is collected and standardized. The standardized historical tourist behavior data of the scenic village is input into the PPO model for training. The PPO model's policy network outputs the action space probability distribution based on the state space, and the value network evaluates the state value. During the training process, the PPO model continuously optimizes the parameters of the policy network and the value network based on the action space probability distribution and the state value, and finally obtains the trained PPO model.

[0035] The trained PPO model instantiates individual tourists as agents and inputs the functional zoning data and spatial constraint rules (viewing, resting, and taking photos) of the initial scenic village functional planning scheme into the PPO model for simulation. During the simulation, the policy network of the PPO model generates the behavioral probability distribution of the agent based on the state space, which guides the agent's behavioral decisions and interaction patterns in the scenic village functional space. At the same time, the PPO model's value network evaluates the state value to optimize long-term benefits (so that the agent's behavioral decisions can maximize its overall benefits in multiple future time steps).

[0036] Based on the intelligent agent's behavioral decisions and interaction patterns in the functional space of the scenic village, the density of tourist distribution, path overlap rate and facility utilization rate are recorded, and congestion analysis is conducted to generate a pedestrian flow simulation evaluation report. Furthermore, based on the behavioral decision-making and interaction patterns of the intelligent agent within the functional space of the scenic village, the location and behavior of the intelligent agent are monitored in real time, recording tourist distribution density, path overlap rate, and facility utilization rate. Tourist distribution density is generated as a tourist density heatmap through spatial gridding to identify high-density tourist gathering areas. Based on path overlap rate, the frequency of repeated passage on key road sections is obtained by analyzing the spatial overlap of tourist behavioral trajectories. Specifically, after denoising and Kalman filtering smoothing of the collected tourist behavioral trajectory data, the scenic village is divided into 5m×5m grid units using a spatial gridding method. Based on the processed behavioral trajectory data, a dynamic time warping algorithm is used to calculate the similarity of behavioral trajectories of different tourists passing through the same grid unit within the same time period. Two behavioral trajectories are considered to have valid overlap when their spatial overlap exceeds 65% in three or more consecutive grid units. Frequency statistics are performed on all overlapping trajectories within a certain time period (e.g., 15 minutes). Combined with high-density areas in the tourist distribution density heatmap, road segments with a repetition frequency exceeding 30 times / hour and a spatial overlap rate greater than 40% are selected as key road segments. The repetition frequency of these key road segments is then calculated. A spatial overlap rate exceeding 65% is considered valid overlap because when two tourist behavior trajectories achieve a 65% overlap within three consecutive 5m×5m grid cells, over 85% of actual same-direction travel is accurately captured, and the false positive rate for random path intersections is controlled below 7%. Road segments with an overlap rate greater than 40% are considered key road segments because when pedestrian path overlap exceeds 40%, the probability of intersection conflicts between two-way and multi-way pedestrian flows significantly increases, leading to a decrease in traffic efficiency of over 20%. Facility utilization statistics are used to count the number of visitors and dwell time in each functional area to assess facility capacity pressure. The final result is a pedestrian flow simulation evaluation report including a tourist density heatmap, the repetition frequency of key road segments, and facility capacity pressure.

[0037] S6. Use the pedestrian flow simulation evaluation report to dynamically optimize the initial scenic village scene planning scheme and obtain the optimized scenic village functional planning scheme. Based on the pedestrian flow simulation evaluation report, the planning shortcomings in the initial scenic village scene planning scheme were identified and dynamically modified to obtain the modified scenic village scene planning scheme. It should be noted that, based on the visitor density heatmap, key road segment repetition frequency, and facility carrying capacity pressure data in the pedestrian flow simulation evaluation report, a threshold comparison method was used to identify the planning shortcomings of the initial scenic village scene planning scheme. Specifically, areas exceeding the density threshold (e.g., 0.8 people / ㎡) in the visitor density heatmap were marked as over-crowding shortcomings; key road segments with repetition frequency exceeding the path capacity threshold (e.g., 60%) were identified as traffic conflict shortcomings; and based on facility carrying capacity pressure, facilities with utilization rates exceeding (e.g., 90%) and those with utilization rates below (e.g., 30%) were identified as configuration imbalance shortcomings. Based on the results of the planning shortcomings identification, the scenic village scene planning scheme was dynamically modified: for high-density areas, crowding pressure was reduced by adding diversion paths and expanding functional zones (functional layout adjustment); for road segments with high overlap rates, traffic conflicts were alleviated by optimizing the path topology and adding alternative routes (path optimization strategy); for overloaded facilities, the number of similar functional facilities was increased, and for inefficient facilities, their functions were converted and their scale reduced (facility configuration change), ultimately resulting in the modified scenic village scene planning scheme. The density threshold is based on the human comfort space requirement of 0.67-0.83 people / square meter, so 0.8 people / square meter is the critical warning value; the path capacity threshold is based on maintaining a path utilization rate of ≤60% to ensure emergency evacuation capacity; facilities with utilization rates exceeding (e.g., 90%) and those with utilization rates below (e.g., 30%) are identified as configuration imbalances because: when the facility utilization rate exceeds 90%, the equipment failure rate increases exponentially (more than 3 times the benchmark value), and when the facility utilization rate is below 30%, the unit operating cost exceeds the revenue by 220%.

[0038] The modified Jingcun functional planning scheme was combined with the PPO model for iterative simulation to verify whether the modification effect could alleviate congestion and improve overall mobility. Through multiple dynamic feedback loops, the distribution of people flow and functional needs were gradually balanced to form an optimized Jingcun functional planning scheme. Furthermore, the functional layout adjustments, path optimization strategies, and facility configuration changes involved in the revised scenic village functional planning scheme are transformed into parameter input conditions recognizable by the PPO model. These input conditions are then fed into the PPO model (which has already been trained and requires no further training). Based on the input conditions, the PPO model uses the strategy network and value network to run a new round of pedestrian flow simulation, simulating the flow of tourists within the revised scenic village functional planning scheme. It then outputs new tourist distribution density, path overlap rate, and facility utilization rate. Based on the newly output tourist distribution density, path overlap rate, and facility utilization rate, the revised scenic village functional planning scheme is evaluated to assess whether it has improved the mitigation of high-density gatherings, reduced path conflicts, and optimized facility utilization. If local congestion and functional configuration imbalances still exist, the scenic village functional planning scheme is modified again. This process is repeated until, through multiple rounds of dynamic feedback loops, the tourist distribution density, path overlap rate, and facility utilization rate gradually approach a reasonable range, ultimately forming an optimized scenic village functional planning scheme. The reasonable range for tourist density is 0.5 to 0.8 people per square meter; the reasonable range for path overlap is 30% to 60%, where less than 30% indicates insufficient path utilization, and more than 60% indicates frequent path intersections and potential traffic conflicts; the reasonable range for facility utilization is 40% to 90%, where less than 40% indicates redundant resource allocation and unreasonable facility layout, and more than 90% indicates excessive pressure on facilities.

[0039] This embodiment also provides a virtual simulation system for scenic village planning, including: The data acquisition module collects real-world images and GIS data of the scenic village, and performs preprocessing to generate multi-source heterogeneous data; The semantic parsing module uses the Graphormer model to perform semantic parsing based on multi-source heterogeneous data, generating Jingcun structured graph data; The clustering module performs spectral clustering on the Jingcun structured graph data to obtain the Jingcun functional space; The solution module collects historical visitor flow data of the scenic village and combines it with a multi-objective evolutionary algorithm to optimize the functional space of the scenic village and generate an initial scenic village scene planning scheme. The evaluation report module, based on the initial village scene planning scheme, uses the PPO model to simulate the behavioral decision-making and interaction patterns of the village crowd and generates a pedestrian flow simulation evaluation report. The optimization module uses the pedestrian flow simulation evaluation report to dynamically optimize the initial scenic village scene planning scheme and obtain the optimized scenic village functional planning scheme.

[0040] This embodiment also provides a computer device applicable to the virtual simulation method for scene planning in a scenic village, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the virtual simulation method for scene planning in a scenic village as proposed in the above embodiment.

[0041] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0042] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the virtual simulation method for scene planning of a scenic village as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0043] In summary, this invention utilizes the Graphormer model for semantic parsing, deeply integrating real-world images and GIS data to enhance the understanding of structured map data of scenic villages. This supports more precise functional zoning, improves the quality of planning decisions, and automatically identifies potential functional areas using spectral clustering analysis, reducing human interference and making planning results more objective and reasonable. This improves the scientific nature of zoning and planning efficiency. Finally, through a dynamic feedback optimization mechanism based on pedestrian flow simulation evaluation reports, the initial planning scheme is continuously improved, effectively solving congestion problems in actual operation and improving overall mobility. This ensures that the planning scheme can meet current needs and adapt to future trends, thereby improving the robustness and sustainability of the scenic village planning method, while also enhancing the accuracy, scientific nature, and flexibility of scenic village scene planning.

[0044] It should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for virtual simulation of a village scene planning, characterized in that: The application relates to a method for generating a scenic village function planning scheme, and belongs to the field of scenic village planning. The method comprises the following steps: collecting scenic village real scene images and GIS data, and performing preprocessing to generate multi-source heterogeneous data; based on the multi-source heterogeneous data, performing semantic analysis by using a Graphormer model to generate scenic village structured graph data; performing spectral clustering on the scenic village structured graph data to obtain scenic village function space; collecting scenic village historical passenger flow data and combining a multi-objective evolutionary algorithm to optimize the scenic village function space to generate an initial scenic village scene planning scheme; based on the initial scenic village scene planning scheme, simulating behavior decision and interaction mode of scenic village crowds by using a PPO model to generate a people flow simulation evaluation report; 2. The method of claim 1, wherein: using the people flow simulation evaluation report to perform dynamic feedback optimization on the initial scenic village scene planning scheme to obtain an optimized scenic village function planning scheme. The method comprises the following steps: collecting scenic village real scene images and GIS data, and performing preprocessing to generate multi-source heterogeneous data, specifically, 3. The method of claim 2, wherein: collecting scenic village real scene images by using a UAV aerial photography, and extracting scenic village GIS data from an OpenStreetMap platform; performing distortion correction on the scenic village real scene images, and performing attribute standardization on the scenic village GIS data to form a multi-source heterogeneous data set.

4. The method of claim 3, wherein: The method comprises the following steps: based on the multi-source heterogeneous data, performing semantic analysis by using a Graphormer model to generate scenic village structured graph data, specifically, transforming the multi-source heterogeneous data set into a graph structure to input the Graphormer model, the global self-attention mechanism of the Graphormer model fuses different spatial semantic information in the graph structure, multi-head attention calculates the correlation of the real scene images and the GIS data, and finally the Graphormer model dynamically aggregates multi-source features of the real scene images and the GIS data in the encoding stage to generate scenic village structured graph data. The method comprises the following steps:

5. The method of claim 4, wherein: performing spectral clustering on the scenic village structured graph data to obtain scenic village function space, specifically, constructing a node adjacency matrix and a node feature matrix based on the scenic village structured graph data; performing Gaussian kernel function calculation on the node adjacency matrix and the node feature matrix to generate a similarity matrix, and performing normalization processing on the similarity matrix to form a Laplacian matrix; 6. The method of village scene planning virtual simulation according to claim 5, characterized in that: performing eigenvalue decomposition on the Laplacian matrix, selecting the feature vectors corresponding to the first k smallest non-zero eigenvalues to form a feature matrix, and using a K-means algorithm to perform vector clustering on the feature matrix to generate a scenic village scene space. The method comprises the following steps: collecting scenic village historical passenger flow data and combining a multi-objective evolutionary algorithm to optimize the scenic village function space to generate an initial scenic village scene planning scheme, specifically, collecting scenic village historical passenger flow data to extract passenger flow distribution, peak period and stay duration, and combining the scenic village function space to establish a multi-objective optimization direction; based on the multi-objective optimization direction, using a multi-objective evolutionary algorithm to perform cross and mutation optimization operation on the scenic village function space layout to generate an initial scenic village scene planning scheme. The method comprises the following steps: based on the initial scenic village scene planning scheme, simulating behavior decision and interaction mode of scenic village crowds by using a PPO model to generate a people flow simulation evaluation report, specifically, According to the initial village function planning scheme, the state space, action space and reward function are constructed for the PPO model. The individual tourists are modeled as agents in the PPO model, and the PPO model simulates the behavior decision and interaction mode of the agents in the village function space through the policy network and value network. According to the behavior decision and interaction mode of the agent in the village function space, the tourist distribution density, path overlap rate and facility usage rate are recorded, and congestion analysis is performed to generate a people flow simulation evaluation report.

7. The method of village scene planning virtual simulation according to claim 6, characterized in that: The people flow simulation evaluation report is used to dynamically feedback and optimize the initial village scene planning scheme, and an optimized village function planning scheme is obtained. Specifically, Based on the people flow simulation evaluation report, the planning short board in the initial village scene planning scheme is identified and dynamically modified to obtain a modified village scene planning scheme. The modified village function planning scheme is iteratively simulated again combined with the PPO model to verify whether the modification effect alleviates congestion and improves overall liquidity. Through multiple dynamic feedback cycles, the people flow distribution and functional demand are gradually balanced to form an optimized village function planning scheme.

8. A scenic village scene planning virtual simulation system based on the scenic village scene planning virtual simulation method according to any one of claims 1-7, characterized in that: It includes, The data acquisition module acquires real scene images and GIS data of the village and performs preprocessing to generate multi-source heterogeneous data. The semantic analysis module uses the Graphormer model to perform semantic analysis based on the multi-source heterogeneous data to generate structured graph data of the village. The clustering module performs spectral clustering on the structured graph data of the village to obtain the village function space. The scheme module collects historical passenger flow data of the village and optimizes the village function space combined with a multi-objective evolutionary algorithm to generate an initial village scene planning scheme. The evaluation report module uses the PPO model to simulate the behavior decision and interaction mode of the village crowd based on the initial village scene planning scheme to generate a people flow simulation evaluation report. The optimization module uses the people flow simulation evaluation report to dynamically feedback and optimize the initial village scene planning scheme to obtain an optimized village function planning scheme. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the village scene planning virtual simulation method of any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the village scene planning virtual simulation method of any one of claims 1-7.

Images