Method and system for simulating spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis

Abstract

The present invention provides a method and system for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis. First, a target spatiotemporal data set containing geographic spatial distribution, architectural form changes, and cultural activity trajectories of the target block is obtained. Then, a spatiotemporal topological structure preprocessing is performed to generate a target spatiotemporal topological data set containing a sequence of spatiotemporally associated block spatial units and dynamic attribute labels. Then, based on a preset spatiotemporal evolution feature extraction model, the static structure, dynamic evolution pattern, and spatial interaction dependency characteristics of each spatial unit are extracted. The spatiotemporal evolution prediction model is called to obtain the evolution probability distribution and evolution correlation strength. Finally, a spatiotemporal evolution path map is constructed based on this, and then the spatial form simulation results and planning intervention decision parameters of the target block are generated, thereby achieving effective simulation of the spatial evolution of historical and cultural blocks and planning decision support.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and system for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis. Background Art

[0002] Historical and cultural blocks carry a city's historical memory and cultural heritage, and research on their spatial evolution is crucial for their protection and utilization. However, existing technologies often focus on static descriptions of the block's current state, or simply analyze a single factor. This makes it difficult to capture the dynamic process and inherent laws of block spatial evolution. Effective analytical methods are also lacking for the complex interactions between block spatial units, making it impossible to accurately depict the mutual influence of spatial units during evolution. Furthermore, existing spatiotemporal evolution prediction models are mostly based on simple assumptions or limited data, making it difficult to handle the complexity and uncertainty of the spatial evolution of historical and cultural blocks. This results in low prediction accuracy and an inability to provide a reliable basis for decision-making on block planning interventions.

[0003] Therefore, existing technologies are difficult to fully and accurately simulate the spatial evolution process of historical and cultural blocks, and cannot meet the actual needs of block protection and development. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis, the method comprising:

[0005] Obtaining a target spatiotemporal data set for a target block, wherein the target spatiotemporal data set includes geographic spatial distribution data, architectural form change data, and human activity trajectory data for a historical period;

[0006] Performing spatiotemporal topological structure preprocessing on the target spatiotemporal data set to generate a target spatiotemporal topological data set, wherein the target spatiotemporal topological data set includes a spatiotemporally associated block spatial unit sequence and a dynamic attribute label of each spatial unit in the block spatial unit sequence;

[0007] Based on a preset spatiotemporal evolution feature extraction model, feature extraction is performed on the target spatiotemporal topological data set to obtain the static structural features, dynamic evolution pattern features and spatial interaction dependency features of each spatial unit;

[0008] Calling the spatiotemporal evolution prediction model to perform spatiotemporal evolution prediction on the static structural features, dynamic evolution pattern features, and spatial interaction dependency features, and generate the evolution probability distribution of the spatial unit and the evolution correlation strength between adjacent spatial units;

[0009] A spatiotemporal evolution path map is constructed according to the evolution probability distribution and the evolution correlation strength, and spatial morphological simulation results and planning intervention decision parameters of the target block are generated based on the spatiotemporal evolution path map.

[0010] On the other hand, an embodiment of the present invention also provides a historical and cultural block spatial evolution simulation system based on GIS spatiotemporal analysis, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiment of the present invention constructs a target spatiotemporal data set by integrating historical geographic spatial distribution data, architectural form change data, and cultural activity trajectory data. The target spatiotemporal data set is preprocessed to obtain a spatiotemporal topological structure, generating a target spatiotemporal topological data set containing a sequence of spatiotemporally correlated block spatial units and dynamic attribute labels. This achieves a deep analysis of the complex relationships between block spaces, transforming originally discrete data into structured information with inherent logical associations, allowing the evolution of block space to be clearly presented in the spatiotemporal dimension. Feature extraction based on a preset spatiotemporal evolution feature extraction model can accurately mine the static structural features, dynamic evolution pattern features, and spatial interaction dependency features of each spatial unit. This not only reflects the inherent properties of the spatial unit itself, but also reveals its changing trends over time and its interactive relationships with surrounding spatial units. Subsequently, the spatiotemporal evolution prediction model is used to perform spatiotemporal evolution prediction on the extracted features, generating a probability distribution of spatial unit evolution and the strength of evolution correlation between adjacent spatial units. This further quantifies the evolution probability and mutual influence of block space from the perspective of probability and correlation, making the evolution process predictable and quantifiable. Finally, a spatiotemporal evolution path map is constructed based on the evolution probability distribution and evolution correlation strength. This map then generates spatial morphological simulation results and planning intervention decision parameters for the target block. This spatiotemporal evolution path map provides a comprehensive and intuitive representation of the block's spatial evolution process. The spatial morphological simulation results provide a visual reference for the block's future development, while the planning intervention decision parameters provide a precise basis for scientific planning and rational protection of the block. This allows for a comprehensive simulation and precise prediction of the spatial evolution of historical and cultural blocks. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the execution flow of the method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis provided by an embodiment of the present invention.

[0013] Figure 2It is a schematic diagram of exemplary hardware and software components of a historical and cultural block spatial evolution simulation system based on GIS spatiotemporal analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. Figure 1 This is a flow chart of a method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis provided by an embodiment of the present invention. The steps involved are introduced in detail below.

[0015] Step S110: Acquire a target spatiotemporal data set of a target block, wherein the target spatiotemporal data set includes geographic spatial distribution data, building form change data, and human activity trajectory data of a historical period.

[0016] In this embodiment, the geographic spatial distribution data, architectural form change data, and cultural activity trajectory data of the historical period are within the protection scope of the historical and cultural block, and the data collection process needs to be combined with the historical and cultural block protection scope, the core protection scope boundary, and the construction control zone boundary. By focusing on the protection scope of the historical and cultural block, the interference of irrelevant geographic information can be avoided, making the relative position relationship and layout description of buildings, streets, etc. more realistic. Within this range, the architectural form change data can clearly sort out the evolution of the unique architectural style of the block and prevent analysis deviations caused by the mixing of irrelevant architectural information. The cultural activity trajectory data is bounded by the protection scope, which can truly restore the historical and cultural scenes of the block and avoid interference from activity information in other areas.

[0017] The core protection area defines the neighborhood's most historically and culturally valuable areas. Data within this area can reflect the neighborhood's core characteristics and original features, and in-depth research will help accurately protect historical and cultural heritage. The construction control zone serves as a protected buffer zone. Data within this area can reflect the impact of the surrounding environment on the core area, providing a basis for formulating appropriate conservation strategies to ensure that the neighborhood retains its historical charm while adapting to modern needs during its development.

[0018] The geographic spatial distribution data, architectural form change data, and cultural activity trajectory data for the historical period can all be directly retrieved from a pre-built resource database. The specific process for building this resource database is not limited in detail. The following embodiment describes an exemplary implementation, but this should not be construed as a limitation on obtaining the target spatiotemporal data set for the target block.

[0019] Specifically, in the process of constructing geospatial distribution data, the geospatial information of the target block at different historical stages can be meticulously restored using historical documents, old maps, photographs, and field research. For example, by accessing historical archival materials, the relative positional relationships and general layout descriptions of various buildings, streets, and public areas in the block can be obtained at different historical periods. For example, for building X within the block, its approximate location and form in 1850 can be determined from historical documents. At the same time, referring to the urban planning drawings of the time, its topological adjacency relationship with surrounding buildings Y and Z can be determined, such as whether building X and building Y are adjacent on a certain edge. For the vertex coordinate information of the buildings, geographic information mapping tools can be used to map the current state of the block. Based on the relative positional relationships and scale information in the historical data, spatial geometry calculations and coordinate transformations can be used to reverse-infer the approximate vertex coordinates of the buildings in different historical periods. For example, in modern surveying and mapping, using a fixed reference point as a benchmark, the current vertex coordinates of Building X are determined. Based on historical data describing the changes in Building X's position and scaling over time, the vertex coordinates in 1850 are estimated to be approximately (10, 20), (30, 20), (30, 40), and (10, 40). Over time, this data collection and extrapolation process is repeated for key time points such as 1900 and 1950, yielding complete geographic spatial distribution data for different periods.

[0020] In the process of constructing data on architectural form changes, we can analyze the historical evolution of architectural form by accessing historical architectural drawings, engineering archives, local chronicles, and other materials, as well as conducting on-site inspections of existing structures and decorative details. This, combined with architectural chronology and style analysis methods, allows us to analyze the historical evolution of architectural form. Specifically, from the perspective of building density, we can calculate the building density of specific areas over time by analyzing historical records of the number of buildings in blocks at different times and estimating the block area. For example, by counting the number of buildings in a specific block in 1850 from historical documents and estimating the area based on contemporary maps, we can determine a building density of 40 buildings per square kilometer. By 1900, using a similar method to obtain the number of buildings and re-estimate the area, we find that the building density has increased to 55 buildings per square kilometer. This year-by-year record forms a detailed sequence of changes in building density. Regarding building height, the authors used historical photographs, written records, and analysis of existing building structures to infer the heights of various buildings over time. For example, by analyzing the proportional relationship between buildings and surrounding objects in historical photographs, combined with descriptions of building heights in written records, they determined that the tallest building on a block was 15 meters in 1850, and that some new buildings reached 25 meters in 1900. By accurately recording the heights of each building at different times, they constructed a sequence of building height changes. Regarding functional zoning, they studied historical land use records, commercial registration data, and records of residents' daily lives, detailing the timing and specific areas of functional transitions. For example, historical commercial archives revealed that an area was primarily registered as residential in 1850, but by 1900, significant commercial activity was observed in some areas. This allowed them to determine that the area's function gradually shifted from residential to commercial, generating a sequence of functional zoning transitions.

[0021] In the process of constructing data on cultural activity trajectories, historical documents, diaries, travelogues, and oral histories can be mined and analyzed. For example, by using important events and activities within a neighborhood as clues and carefully interpreting historical documents, we can unravel the movement routes and activity areas of people within the neighborhood during different periods. For example, by studying local chronicles and celebrity diaries, we can understand the locations and routes of gatherings, commercial activities, and other activities concentrated within the neighborhood during a specific historical period. Simultaneously, using historical maps and spatial analysis methods, we can locate and annotate this activity information in geographic space. Regarding the duration of people's stays at different locations, we analyze historical data on activity duration and event sequence, combining reasonable speculation and estimation to derive approximate stay times. Over a weekly period, we organize and summarize this collected cultural activity information from different periods, accumulating a large amount of data. Subsequently, we use algorithms to spatially rasterize this trajectory data, dividing the neighborhood into numerous small grid cells. For example, during a specific time period, a grid cell in the center of a block shows frequent activity by a large number of people. This activity intensity is represented on the human activity heat map as a high-intensity area with a darker color. Meanwhile, a grid cell on the edge of a block shows only a small number of people passing through it, indicating a low-intensity area with a lighter color. A time-series curve of activity intensity is also extracted for each spatial unit. For example, for a spatial unit, activity intensity is low between 9:00 and 11:00 a.m., peaks between 12:00 and 1:00 p.m., and then gradually decreases. This curve forms the activity intensity time-series curve for that unit. The distribution ratio of different activity types within each spatial unit is also calculated, such as leisure activities accounting for 25%, commercial activities accounting for 45%, and residential activities accounting for 30%.

[0022] Step S120: performing spatiotemporal topological structure preprocessing on the target spatiotemporal data set to generate a target spatiotemporal topological data set, wherein the target spatiotemporal topological data set includes a spatiotemporally associated block spatial unit sequence and a dynamic attribute label of each spatial unit in the block spatial unit sequence.

[0023] In this embodiment, step S120 may include:

[0024] Step S121: performing coordinate calibration processing on the geographic spatial distribution data to obtain spatial unit boundary data in a unified coordinate system, wherein the spatial unit boundary data includes the coordinates of the geometric shape vertices and the topological adjacency relationship of each spatial unit.

[0025] In this embodiment, since the geospatial distribution data acquired at different times may be based on different coordinate systems, coordinate calibration is required. For example, early surveying and mapping used a local independent coordinate system, while later surveying used the WGS84 coordinate system. Therefore, coordinate conversion software can be used to convert all geospatial distribution data to the WGS84 coordinate system based on a corresponding conversion algorithm. For example, the vertex coordinates of building X, previously (10, 20), (30, 20), (30, 40), and (10, 40) in the local independent coordinate system, become (100.123, 200.456), (100.345, 200.456), (100.345, 200.678), and (100.123, 200.678) in the WGS84 coordinate system after conversion. At the same time, the topological adjacency relationship with the surrounding buildings Y and Z is reconfirmed to ensure the accuracy of the adjacency relationship in the new coordinate system.

[0026] Step S122: performing time series alignment processing on the building form change data to obtain a building density change sequence, a building height change sequence, and a functional zoning conversion sequence that match the spatial unit boundary data.

[0027] In this embodiment, since the time intervals for obtaining building form change data at different stages are inconsistent, time series alignment is required. For example, building density data are obtained in 1850, 1900, and 1950, while building height data are obtained in 1860, 1910, and 1960. The building height data are aligned to the same time node as the building density data through methods such as linear interpolation. For the time point of 1900, the original building height data is missing. By performing linear interpolation calculations on the data of 1860 and 1910, the approximate building height data of 1900 is obtained, so that the building density change sequence and the building height change sequence are matched with the spatial unit boundary data in time, while ensuring that the functional zoning conversion sequence is also accurately corresponding on the same time scale.

[0028] Step S123: performing spatial rasterization processing on the cultural activity trajectory data to generate a cultural activity heat distribution map corresponding to the spatial unit boundary data, and extracting the activity intensity time series curve and activity type distribution ratio of each spatial unit in the cultural activity heat distribution map.

[0029] In this embodiment, spatial analysis software can be used to perform spatial gridding on human activity trajectory data. For example, a street block can be divided into square grid cells with a side length of 10 meters. The number of human activities in each grid cell within a specific time period can then be counted. For example, within an hour, Grid A, located in the center of the street block, has 50 people passing through it, while Grid B, located at the edge, has only 5 people passing through it. Based on this data, a heat distribution map of human activity is generated, with Grid A displayed as a high-activity area in red, and Grid B as a low-activity area in light blue. From the heat distribution map, a time series curve of activity intensity is extracted for each spatial cell. For example, the activity intensity in Grid C is 10 people / hour from 7:00 to 9:00 a.m., increases to 30 people / hour from 10:00 to 12:00, and then gradually decreases. Simultaneously, the distribution ratio of different activity types within each spatial cell is counted. For example, in Grid D, leisure activities account for 35%, commercial activities account for 40%, and residential activities account for 25%.

[0030] Step S124: Perform spatiotemporal topological association encoding on the spatial unit boundary data, building density change sequence, building height change sequence, functional zoning conversion sequence, activity intensity time series curve and activity type distribution ratio to generate the target spatiotemporal topological data set, wherein the dynamic attribute label of each spatial unit includes the building density change rate, functional conversion frequency and activity intensity fluctuation amplitude.

[0031] In this embodiment, taking spatial unit E as an example, its boundary data in a unified coordinate system, the building density change sequence from 1850 to 1950 (e.g., the building density was 40 buildings / square kilometer in 1850, 55 buildings / square kilometer in 1900, and 70 buildings / square kilometer in 1950), the building height change sequence (e.g., 15 meters in 1850, 25 meters in 1900, and 30 meters in 1950), the functional zoning conversion sequence (e.g., residential function in 1850, partial commercial function in 1900, and main commercial function in 1950), the activity intensity time series curve (e.g., the activity intensity was 15 people / hour from 8:00 to 10:00 in the morning, and 40 people / hour from 12:00 to 2:00), and the activity type distribution ratio (e.g., 20% leisure activities, 60% commercial activities, and 20% residential-related activities) are integrated and coded. Calculate the rate of change in building density for this spatial unit. From 1850 to 1900, the building density increased from 40 buildings / km² to 55 buildings / km² over 50 years, with a rate of change of (55-40) ÷ 50 = 0.3 buildings / km² / year. Calculate the frequency of functional conversions. From 1850 to 1950, the functional conversion from residential to commercial was 1. Calculate the fluctuation range of activity intensity. From 8:00 AM to 10:00 AM, the activity intensity was 15 people / hour, and from 12:00 PM to 2:00 AM, the activity intensity was 40 people / hour, with a fluctuation range of 40-15 = 25 people / hour. Thus, encode the above dynamic attribute tags along with other data to form the target spatiotemporal topological data set.

[0032] Step S130: Based on a preset spatiotemporal evolution feature extraction model, feature extraction is performed on the target spatiotemporal topological data set to obtain static structural features, dynamic evolution pattern features, and spatial interaction dependency features of each spatial unit.

[0033] In this embodiment, step S130 may include:

[0034] Step S131: extracting the spatial unit boundary data in the target spatiotemporal topological data set, calling the spatial morphology encoder to generate an initial structural feature vector for each spatial unit, wherein the initial structural feature vector includes shape complexity, area change trend and number of adjacent units.

[0035] For example, let's assume that the boundary data for spatial unit F is displayed as an irregular polygon with vertex coordinates (100, 100), (120, 110), (130, 130), (110, 140), and so on. The spatial morphology encoder is called to measure shape complexity by calculating parameters such as the number of sides and internal angles of the polygon. This polygon has four sides, and the algorithm calculates a shape complexity of 1.2 (the specific algorithm depends on the spatial morphology encoder settings). Furthermore, the area change trend is calculated, comparing the area of this spatial unit over time. Assuming that over the past 50 years, the area has grown from 1000 square meters to 1200 square meters, with an average annual increase of (1200 - 1000) ÷ 50 = 4 square meters, the area change trend is 4 square meters per year. Simultaneously, the number of adjacent units is counted, and spatial unit F is adjacent to three surrounding spatial units. Therefore, the shape complexity of 1.2, the area change trend of 4 square meters per year, and the number of adjacent units of 3 can be combined to generate the initial structural feature vector.

[0036] Step S132: Perform time convolution processing on the building density change sequence, building height change sequence and functional zoning conversion sequence in the target spatiotemporal topological data set to generate a first dynamic evolution feature vector, which includes the building density fluctuation period, height growth gradient and functional conversion trigger threshold.

[0037] For example, if the building density change sequence for a given spatial unit is (40, 45, 50, 55, 60) over a five-year period, the time convolution algorithm calculates a three-year building density fluctuation cycle. A similar process is performed for a building height change sequence, such as (15, 18, 20, 22, 25) over a five-year period. The height growth gradient is calculated as (25-15) ÷ 5 = 2 meters / year. For a functional zoning transition sequence, such as from residential to commercial, the conditions and relevant factors for the transition are analyzed, and the functional transition trigger threshold is determined to be when the proportion of commercial activity reaches 40%. The three-year building density fluctuation cycle, the 2-meter / year height growth gradient, and the 40% functional transition trigger threshold are concatenated to generate the first dynamic evolution feature vector.

[0038] Step S133: Perform spatiotemporal attention processing on the activity intensity time series curve and activity type distribution ratio in the target spatiotemporal topological data set to generate a second dynamic evolution feature vector, which includes the activity intensity peak phase, type mixing degree and spatial diffusion direction.

[0039] For example, assuming that the activity intensity time series curve for a given spatial unit shows a peak between 2:00 PM and 4:00 PM, the peak activity intensity phase is determined to be between 2:00 PM and 4:00 PM. The distribution ratio of activity types is calculated, e.g., leisure activities account for 30%, commercial activities account for 50%, and residential activities account for 20%. Using a set algorithm, the type mixing degree is calculated to be 0.6 (the specific algorithm depends on the processing rules). Analyzing the movement trajectories of individuals determines the spatial diffusion direction to be from the center of the block toward the periphery. Therefore, the peak activity intensity phase between 2:00 PM and 4:00 PM, the type mixing degree of 0.6, and the spatial diffusion direction from the center of the block toward the periphery can be combined to generate the second dynamic evolution feature vector.

[0040] Step S134: performing cross-scale feature concatenation on the initial structural feature vector, the first dynamically evolving feature vector, and the second dynamically evolving feature vector to obtain a static structural feature.

[0041] In this example, the previously generated initial structural feature vector (including shape complexity of 1.2, area change trend of 4 square meters / year, and number of adjacent units of 3), the first dynamic evolution feature vector (including building density fluctuation period of 3 years, height growth gradient of 2 meters / year, and functional conversion trigger threshold of 40%), and the second dynamic evolution feature vector (including activity intensity peak phase between 2:00 PM and 4:00 PM, type mixing degree of 0.6, and spatial diffusion direction from the center of the block to the periphery) can be combined for cross-scale feature splicing. According to the set splicing rules, these different dimensional features are combined to form a comprehensive static structural feature vector, which contains multiple aspects of the spatial unit's structural information.

[0042] Step S135: performing graph convolution processing on the topological adjacency relationship in the target spatiotemporal topological data set to generate spatial interaction dependency features, wherein the spatial interaction dependency features include the density conduction coefficient, functional compatibility index and activity diffusion resistance value between adjacent units.

[0043] Step S1351: construct an undirected graph structure with the spatial units in the spatial unit boundary data as nodes and the topological adjacency relationships as edges.

[0044] In this embodiment, an undirected graph structure can be constructed based on the spatial unit boundary data in the previously processed target spatiotemporal topological data set. Taking multiple spatial units in a block as an example, assuming spatial units A, B, C, D, etc., the topological adjacency relationship between them is known, such as spatial unit A is adjacent to spatial unit B, spatial unit B is adjacent to spatial unit C, spatial unit C is adjacent to spatial unit D, etc. Thus, the above-mentioned spatial units can be used as nodes of an undirected graph, and the adjacency relationship between them can be used as edges to construct an undirected graph structure. This undirected graph structure can intuitively reflect the adjacency relationship between each spatial unit, and provide an infrastructure for subsequent feature extraction and analysis. For example, in this undirected graph, spatial unit A is a node connected to the adjacent spatial unit B through an edge, and so on, forming a complete graph structure to describe the topological relationship between the spatial units in the block.

[0045] Step S1352: Initialize the initial feature vector of each node in the undirected graph structure, where the initial feature vector is generated by concatenating the shape complexity, area change trend, and number of adjacent units in the static structural features with the building density fluctuation period and function conversion trigger threshold in the dynamic evolution pattern features.

[0046] In this embodiment, the initial feature vector is initialized for each node in the constructed undirected graph structure, that is, each spatial unit. Continuing with the aforementioned spatial unit E as an example, the shape complexity in its static structural features is 1.5 (derived from the previous calculation of its boundary data, the spatial unit is a slightly complex polygonal shape), the area change trend is an increase of 3 square meters per year (obtained through analysis of area data at different times), and the number of adjacent units is 4 (it is adjacent to the four surrounding spatial units). The building density fluctuation cycle in its dynamic evolution pattern features is 4 years (derived from the analysis of the building density change sequence), and the function conversion trigger threshold is 50% (when the proportion of commercial activities reaches 50%, the function will be converted). Then, the above features are spliced to form the initial feature vector of the spatial unit E. In the same way, a corresponding initial feature vector is generated for each node in the undirected graph structure.

[0047] Step S1353: Perform multi-layer graph convolutional network processing on the undirected graph structure, aggregate the adjacent node feature vectors of each node layer by layer, and generate intermediate spatial interaction features.

[0048] In this embodiment, a multi-layer graph convolutional network is used to process an undirected graph structure with constructed and initialized node feature vectors. In the first-layer graph convolutional network, for example, spatial unit F has adjacent nodes G, H, and I. The graph convolutional network aggregates the initial feature vector of spatial unit F and the initial feature vectors of its adjacent nodes G, H, and I according to predefined rules. Assume that the initial feature vector of spatial unit G includes information such as shape complexity of 1.3, area change trend of 2 square meters / year, number of adjacent units of 3, building density fluctuation cycle of 3.5 years, and functional conversion trigger threshold of 45%. The initial feature vector of spatial unit H includes corresponding different values. Spatial unit I also has its own unique initial feature vector. Using the graph convolutional network's computational rules (e.g., based on adjacency relationships and feature vector weighting), the feature vectors of these adjacent nodes are aggregated with the feature vector of spatial unit F to generate an intermediate spatial interaction feature of spatial unit F after processing by the first-layer graph convolutional network. This intermediate spatial interaction feature incorporates partial information about spatial unit F and its adjacent nodes, preliminarily reflecting the interactive relationship between spatial units.

[0049] Then we enter the second-layer graph convolutional network. In the same way, we can further aggregate the intermediate space interaction features processed by the first layer and the features of the adjacent nodes processed by the first layer, continuously enriching and updating the intermediate space interaction features. Similarly, after processing by multiple layers of graph convolutional networks, we can obtain more comprehensive intermediate space interaction features.

[0050] Step S1354: After each layer of graph convolutional network processing, the intermediate space interaction feature is input into the gated recurrent unit, and the neighborhood feature fusion weight is generated based on the function conversion frequency in the dynamic attribute label. The intermediate space interaction feature and the current node feature vector are weighted superimposed, and the node feature vector is updated to obtain the node feature vector output by the final layer.

[0051] In this embodiment, after each layer of the graph convolutional network is processed, the generated intermediate spatial interaction features are input into the gated recurrent unit. Taking spatial unit J as an example, after processing a certain layer of the graph convolutional network, the intermediate spatial interaction features are obtained. It is also known that the function transition frequency in its dynamic attribute label is 2 (the function transitioned twice within a certain time range). The gated recurrent unit can generate neighborhood feature fusion weights based on information such as the function transition frequency. Assume that the calculation rules of the gated recurrent unit determine the weight distribution for the fusion of the current spatial unit J with the adjacent node features. For example, for the features of adjacent node K, the weight is 0.4; for the features of adjacent node L, the weight is 0.3, and so on. Then, the intermediate spatial interaction features are weightedly superimposed with the feature vector of the current node (spatial unit J) according to the above weights. In other words, the portions of the intermediate spatial interaction features from different adjacent nodes are combined with the feature vector of spatial unit J itself according to the weights to update the node feature vector of spatial unit J. After multi-layer graph convolutional network processing and such weighted superposition operations in each layer, the node feature vector output by the final layer is finally obtained, which fully integrates the information of the spatial unit itself and the adjacent units after multi-layer processing, and more comprehensively reflects the characteristics of the spatial unit in spatial interaction.

[0052] Step S1355: Perform spatial scale alignment processing on the node feature vector output by the final layer, perform spatial position encoding on the feature vector of each node and the geometric shape vertex coordinates of the spatial unit boundary data, and generate spatial interaction dependency features including density conduction gradient, functional compatibility matrix and activity diffusion direction vector.

[0053] In this embodiment, since the feature vectors of different spatial units may differ in scale, unified processing is required. For example, for the final node feature vector of spatial unit M, the values of its various dimensions are adjusted through the set algorithms and rules to make it comparable with the feature vectors of other spatial units in terms of spatial scale. Then, the feature vector of each node is spatially encoded with the vertex coordinates of the geometric shape of the spatial unit boundary data. Taking spatial unit N as an example, the geometric vertex coordinates of its boundary data are (100, 150), (120, 160), etc. The node feature vector is combined with the above vertex coordinates and encoded using the set encoding method (for example, certain dimensions of the feature vector are associated with the values of the vertex coordinates). Through such processing, a spatial interaction dependency feature including a density conduction gradient, a functional compatibility matrix, and an activity diffusion direction vector is generated. For the density conduction gradient, the density conduction gradient value from a spatial unit to its adjacent units is calculated by analyzing factors such as the changes in building density of the spatial unit and its adjacent units and the distance between them; the functional compatibility matrix is constructed based on the functional zoning conversion of different spatial units and the adjacency relationship between them to express the degree of functional compatibility between each spatial unit; the activity diffusion direction vector is a comprehensive analysis of personnel activity trajectory data and the topological adjacency relationship of spatial units to determine the diffusion direction and intensity of personnel activities between spatial units, and finally form a spatial interaction dependence feature that comprehensively reflects the interaction dependence relationship between spatial units.

[0054] Step S136: Dynamically weighting and correcting the static structural features based on the dynamic attribute labels in the target spatiotemporal topological data set to obtain dynamic evolution pattern features.

[0055] In this embodiment, taking spatial unit P as an example, its static structural features have been generated through the previous steps, including information such as shape complexity and area change trends. Simultaneously, the target spatiotemporal topological data set records the dynamic attribute tags of this spatial unit, such as a building density change rate of 2 buildings per square kilometer per year, a functional conversion frequency of 3 times (within a specific time period), and an activity intensity fluctuation amplitude of 20 people per hour. Based on these dynamic attribute tags, a dynamic weighting adjustment is performed on the static structural features. For example, for the shape complexity feature, a high functional conversion frequency indicates that the spatial unit is undergoing active functional changes, potentially impacting its future shape development. Therefore, a relatively high weight of 0.6 is assigned to this feature. For the area change trend, the weight is adjusted based on factors such as the building density change rate, assuming a weight of 0.4. In this way, weighted calculations are performed on each component of the static structural feature, and the weighted results are combined to generate a dynamic evolution pattern feature. This dynamic evolution pattern feature considers both the static structural foundation of the spatial unit and its dynamic change attributes, more accurately reflecting the spatial unit's evolutionary trend over time.

[0056] Step S140: calling the spatiotemporal evolution prediction model to perform spatiotemporal evolution prediction on the static structural features, dynamic evolution pattern features and spatial interaction dependency features, and generating the evolution probability distribution of the spatial unit and the evolution correlation strength between adjacent spatial units.

[0057] Step S141: inputting the static structural features into the first branch network of the spatiotemporal weight distribution model to generate an internal evolution weight coefficient of the spatial unit, wherein the internal evolution weight coefficient reflects the degree of influence of the historical morphology on the current evolution direction.

[0058] In this embodiment, the static structural features extracted previously are input into the first branch network of the spatiotemporal weight distribution model. The first branch network will analyze the historical morphological characteristics of the spatial unit based on the above information. If the shape of the spatial unit has been relatively stable for a long time in the past and the area changes are relatively regular, then the historical morphology may have a greater impact on the current evolution direction. The internal evolution weight coefficient indicates that the historical morphology has a greater weight in the future evolution direction of the spatial unit, which means that its future evolution may be constrained and guided by the past morphology to a certain extent. Different spatial units will generate corresponding internal evolution weight coefficients through the first branch network according to their respective static structural characteristics. The above internal evolution weight coefficients reflect the different degrees of influence of the historical morphology of each spatial unit on the current evolution direction.

[0059] Step S142: Inputting the dynamic evolution pattern characteristics into the second branch network of the spatiotemporal weight allocation model to generate a time decay adjustment coefficient, which reflects the decay rate of the contribution of data in different periods to the prediction results.

[0060] In this embodiment, the second branch network analyzes how the characteristics of the dynamic evolution pattern change over time. For example, if building density fluctuations are relatively stable, the influence of earlier data on current and future evolution may gradually decrease over time. Through an internal algorithm within the network, combining historical data and time factors, a time decay adjustment coefficient is calculated for each spatial unit. Based on the differences in the characteristics of the dynamic evolution pattern of each spatial unit, the second branch network generates its own time decay adjustment coefficient, which is used to adjust the weight of data from different periods in the predicted evolution process.

[0061] Step S143: Inputting the spatial interaction dependency feature into the third branch network of the spatiotemporal weight allocation model to generate a neighborhood coupling strength coefficient, which reflects the collaborative or competitive relationship between the evolution behaviors of adjacent spatial units.

[0062] In this embodiment, the spatial interaction dependency feature can be input into the third branch network of the spatiotemporal weight allocation model. Taking the spatial unit S and its adjacent units T and U as an example, the spatial interaction dependency feature includes information such as the density conduction gradient, the functional compatibility matrix and the activity diffusion direction vector. The third branch network analyzes the interaction between the spatial unit S and the adjacent units T and U. If the functional compatibility between the spatial unit S and the adjacent unit T is high, and the diffusion of human activities between them is relatively frequent, it means that the evolutionary behavior between them may have strong synergy; on the contrary, if there is functional competition or activity diffusion hindrance, there may be a competitive relationship. Through the analysis and calculation of the above-mentioned interaction information, the neighborhood coupling strength coefficient of the spatial unit S is generated to be 0.6. The neighborhood coupling strength coefficient represents the degree of evolutionary synergy or competition between the spatial unit S and the adjacent units. Different spatial units generate their own neighborhood coupling strength coefficients through the third branch network according to the different situations of their spatial interaction dependency features to reflect their interaction relationship with the adjacent spatial units in the evolution process.

[0063] Step S144: constructing a spatiotemporal joint weight matrix according to the internal evolution weight coefficient, the time attenuation adjustment coefficient and the neighborhood coupling strength coefficient, performing weighted fusion on the static structural features, the dynamic evolution pattern features and the spatial interaction dependency features to generate a spatiotemporal fusion feature vector.

[0064] In this embodiment, a spatiotemporal joint weight matrix can be constructed based on the previously generated internal evolution weight coefficients, time decay adjustment coefficients, and neighborhood coupling strength coefficients. Taking spatial unit V as an example, its internal evolution weight coefficient is 0.7, its time decay adjustment coefficient is 0.8, and its neighborhood coupling strength coefficient is 0.5. These coefficients are arranged into a matrix according to a predefined rule: for example, the internal evolution weight coefficient is placed in the upper left corner of the matrix, the time decay adjustment coefficient is placed in the middle, and the neighborhood coupling strength coefficient is placed in the lower right corner. (The specific arrangement depends on the model settings.) This spatiotemporal joint weight matrix is then used to perform a weighted fusion of static structural features, dynamic evolution pattern features, and spatial interaction dependency features. For each element in the static structural features, such as shape complexity and area change trend, its weight in the fusion is determined by the internal evolution weight coefficient. For dynamic evolution pattern features, such as building density fluctuation period and height growth gradient, its weight is determined by the time decay adjustment coefficient. For spatial interaction dependency features, such as density conduction gradient and functional compatibility matrix, its weight is determined by the neighborhood coupling strength coefficient. Through weighted calculation, these different features are fused to generate a spatiotemporal fusion feature vector. The spatiotemporal fusion feature vector integrates the multi-faceted characteristics of spatial units and their weight information in the spatiotemporal dimensions, providing more comprehensive data support for subsequent evolution predictions.

[0065] Step S145: inputting the spatiotemporal fusion feature vector into the evolution probability prediction layer to generate the evolution probability distribution of the spatial unit.

[0066] The generated spatiotemporal fusion feature vector is input into the evolution probability prediction layer. Taking spatial unit W as an example, its spatiotemporal fusion feature vector contains a variety of information after weighted fusion. The evolution probability prediction layer analyzes the relationship between the information contained in the feature vector and the future evolution possibility of the spatial unit. For example, considering comprehensive information such as the static structural characteristics, dynamic evolution pattern characteristics, and spatial interaction dependency characteristics of spatial unit W, combined with historical data and the laws obtained from model training, the evolution probability of the spatial unit in different future states is predicted. Different spatial units will generate their own evolution probability distribution through the evolution probability prediction layer. The above evolution probability distribution reflects the different evolution situations that may occur in each spatial unit in the future and their corresponding probabilities.

[0067] Step S146: inputting the spatiotemporal fusion feature vector into the association strength prediction layer to generate the evolving association strength between adjacent spatial units.

[0068] The spatiotemporal fusion feature vector is input into the association strength prediction layer. Taking spatial unit X and its adjacent spatial unit Y as an example, the spatiotemporal fusion feature vector contains a variety of information about them and their relationship with each other. Based on this information, the association strength prediction layer uses specific algorithms and models to analyze the degree of association between spatial units X and Y in their future evolution. For example, it considers factors such as functional compatibility and activity diffusion direction in their spatial interaction dependency characteristics, as well as their respective static structures and dynamic evolution pattern characteristics. Different adjacent spatial units will generate evolutionary association strength values between them through the association strength prediction layer. The above evolutionary association strength values reflect the mutual dependence and influence of adjacent spatial units during the evolution process.

[0069] Step S150: constructing a spatiotemporal evolution path map according to the evolution probability distribution and the evolution correlation strength, and generating spatial morphological simulation results and planning intervention decision parameters of the target block based on the spatiotemporal evolution path map.

[0070] Step S151: constructing an initial spatiotemporal evolution graph structure with each spatial unit as a node and the evolution association strength as an edge weight.

[0071] In this example, each spatial unit in a block is treated as a node. For example, the previously calculated evolutionary correlation strength between adjacent spatial units is used as the edge weight to construct an initial spatiotemporal evolution graph structure, forming a preliminary graph structure to represent the evolutionary correlation relationship between spatial units.

[0072] Step S152: updating the node states of the initial spatiotemporal evolution graph structure according to the evolution probability distribution to generate a spatiotemporal evolution graph with probability weights.

[0073] For example, in the initial spatiotemporal evolution graph structure, the state of each node is updated according to its evolution probability distribution to generate a spatiotemporal evolution graph with probability weights, which not only reflects the evolution correlation strength between spatial units, but also takes into account the probability of different evolution states of each spatial unit itself.

[0074] Step S153: performing path search processing on the spatiotemporal evolution graph with probability weights, and extracting edges with probability weights greater than a set weight to form a main evolution path set.

[0075] For example, in one possible implementation, step S153 may include:

[0076] Step S1531: performing bidirectional weight sorting on the edges in the spatiotemporal evolution graph based on the evolution association strength to generate an initial path search queue.

[0077] In this embodiment, for a spatiotemporal evolution graph with probability weights, the evolutionary correlation strength of the edges between spatial units is analyzed. For example, taking a local spatiotemporal evolution graph consisting of spatial units F, G, H, and I in a block as an example, the evolutionary correlation strength between spatial units F and G is 0.7, the evolutionary correlation strength between F and H is 0.5, and the evolutionary correlation strength between F and I is 0.6; the evolutionary correlation strength between G and H is 0.4, the evolutionary correlation strength between G and I is 0.3, and the evolutionary correlation strength between H and I is 0.65.

[0078] Bidirectional weighted sorting involves sorting the evolutionary strength of edges both from high to low and from low to high. For example, when sorting from high to low, the largest evolutionary strength value is determined first. In the above example, 0.65 between H and I is a relatively large value, followed by 0.7 between F and G, then 0.6 between F and I, and so on. When sorting from low to high, the smallest value, 0.3 between G and I, is used, followed by 0.4 between G and H, and so on.

[0079] The sorted edge information is organized into an initial path search queue. This queue contains all edge information and their corresponding weights, arranged in bidirectional order. For example, the highest-to-lowest order might be: (HI, 0.65), (FG, 0.7), (FI, 0.6), etc.; the lowest-to-highest order might be: (GI, 0.3), (GH, 0.4), etc. This initial path search queue provides an organized data foundation for subsequent path searches, facilitating the exploration of possible evolution paths based on different weight orders.

[0080] Step S1532: extract the current maximum weight edge from the initial path search queue, perform depth-first traversal along the nodes connected by the edge, and record the cumulative evolution probability of the traversal path and the sequence of spatial units passed through.

[0081] In this example, we first extract the maximum-weight edge from the initial path search queue. Assume that (FG, 0.7) is the maximum-weight edge in the current queue. Starting with spatial unit F as the starting node, we follow this edge to the connected spatial unit G and begin a depth-first traversal.

[0082] During the traversal process, the cumulative evolution probability of the traversal path needs to be recorded. The calculation of the cumulative evolution probability is based on the evolution probability distribution of each spatial unit and the weight of the connecting edge. For example, the probability of spatial unit F evolving to spatial unit G in its evolution probability distribution is 0.5, and the weight of the edge between F and G is 0.7. Then the cumulative evolution probability from F to G is the product of the two, that is, 0.5 × 0.7 = 0.35.

[0083] After reaching spatial unit G, continue checking the adjacent edges of node G. Assuming the edge weight between G and H is 0.4, and the probability of G evolving to H in its evolution probability distribution is 0.6, then the cumulative evolution probability from G to H becomes 0.084, and the spatial unit sequence passed is recorded as FGH.

[0084] Following the principle of depth-first traversal, the exploration continues along the adjacent edges of the current node, continuously updating the cumulative evolution probability and the sequence of spatial units traversed until the path cannot be further expanded. For example, during the exploration process, some nodes may be encountered whose adjacent edges have all been visited, or the weight of some edges is below the set threshold. At this time, the traversal in that direction is stopped.

[0085] Step S1533: When traversing to the end node, compare the cumulative evolution probability with the path threshold. If the cumulative evolution probability exceeds the path threshold, mark the spatial unit sequence as a candidate path.

[0086] Assuming that the path threshold set in this embodiment is 0.1, when the depth-first traversal reaches the terminal node, taking the previously obtained path from F through G to H as an example, its cumulative evolution probability is 0.084, which does not exceed the path threshold of 0.1, so this path is not marked as a candidate path.

[0087] Continuing the traversal, suppose another path starts from spatial unit A, passes through spatial units B and C, and reaches spatial unit D. During the traversal, the cumulative evolution probability is calculated to be 0.15. When reaching the terminal node D, the cumulative evolution probability of 0.15 is compared with the path threshold of 0.1. Since 0.15 is greater than 0.1, the spatial unit sequence ABCD is marked as a candidate path.

[0088] Through the above comparison process, it is possible to screen out those spatial unit sequences with higher cumulative evolution probability and greater possibility of becoming the main evolution path.

[0089] Step S1534: Verify the spatial topological continuity of the candidate path, detect whether adjacent spatial units in the candidate path meet the geometric shape vertex coordinate overlap condition in the topological adjacency relationship, and eliminate broken paths based on the detection results to obtain a verified candidate path.

[0090] For the sequence of spatial units marked as candidate paths, using ABCD as an example, we verify their spatial topological continuity. First, we check the topological adjacency of spatial units A and B to see if their geometric vertex coordinates overlap. Assuming the boundary vertex coordinates of spatial unit A are (100, 100), (120, 100), (120, 120), (100, 120), and the boundary vertex coordinates of spatial unit B are (120, 100), (140, 100), (140, 120), (120, 120), we can see that some of their vertex coordinates overlap, satisfying the topological adjacency relationship.

[0091] Next, check the topological adjacency between spatial units B and C. If the vertex coordinates of the boundary of spatial unit C overlap with the vertex coordinates of B, the adjacency condition is met. Then check the adjacency between C and D. If the condition is also met, the candidate path ABCD passes the spatial topological continuity verification.

[0092] However, if there is a candidate path, such as EFG, in which the vertex coordinates of spatial units E and F do not overlap and do not satisfy the topological adjacency relationship, then this candidate path is a broken path and needs to be eliminated.

[0093] By performing such spatial topological continuity verification on all candidate paths, the broken paths that do not meet the conditions are eliminated to obtain the verified candidate paths. The verified candidate paths are continuous in spatial topology and are more in line with the actual logic of block space evolution.

[0094] Step S1535: Arrange the verified candidate paths in descending order according to the cumulative evolution probability, and select the top N paths as the main evolution path set.

[0095] Assume that after spatial topological continuity verification, multiple verified candidate paths are obtained, for example, the cumulative evolution probability of path P1 is 0.2, the cumulative evolution probability of path P2 is 0.18, the cumulative evolution probability of path P3 is 0.16, and so on.

[0096] Then, the verified candidate paths are sorted in descending order according to the cumulative evolution probability, that is, from large to small. The order after sorting may be path P1, path P2, path P3, etc.

[0097] Based on actual needs, the first N paths are selected as the primary evolution path set. Assuming N = 3, the selected primary evolution path set includes paths P1, P2, and P3. This primary evolution path set represents the path combination with the highest probability of street spatial evolution under the current analysis, providing core data for the subsequent generation of a spatiotemporal evolution path map.

[0098] Step S154: performing loop detection on the main evolution path set, and if a closed loop is detected, performing attenuation correction on the evolution probability within the loop according to the time attenuation adjustment coefficient.

[0099] In the main evolution path set, we check whether there is a closed loop by taking the path containing spatial units M, N, O, and P as an example. After analysis, we find that there is a closed loop of MNOPM in the path.

[0100] It is known that each spatial unit has a corresponding time attenuation adjustment coefficient. Assume that the time attenuation adjustment coefficient of spatial unit M is 0.8, that of N is 0.75, that of O is 0.85, and that of P is 0.7.

[0101] For the evolution probability within a loop, take the evolution probability from M to N as an example. Assuming the original probability is 0.4, due to the existence of the loop, the time-attenuation adjustment coefficient is used to adjust the probability to 0.24. The same method is used to adjust the evolution probability of other edges within the loop. By using this time-attenuation adjustment coefficient to attenuate the evolution probability within the loop, we can more accurately reflect the actual spatial evolution of the block and avoid the problem of unreasonable amplification of evolution probabilities caused by loops.

[0102] Step S155: Overlay and render the attenuation-corrected main evolution path set with the spatial unit boundary data to generate the spatiotemporal evolution path map, which includes historical morphology preservation areas, high-probability evolution areas, and neighborhood co-evolution corridors.

[0103] In this embodiment, the attenuation-corrected main evolution path set is combined with the spatial unit boundary data. Based on the geographic information data of the block, the path information in the main evolution path set is mapped onto the map represented by the spatial unit boundary data.

[0104] For example, in the case of historical form preservation areas, if the architectural form of certain areas within the spatial unit boundary data has not changed significantly over a long period of time, these areas are marked as historical form preservation areas in the spatiotemporal evolution path map. Assuming that the building density, building height, and functional zoning in the area where spatial unit Q is located have remained relatively stable over the past few decades, then this area is classified as a historical form preservation area.

[0105] High-probability evolution zones are determined based on the cumulative evolution probabilities of each path in the main evolution path set. Spatial units traversed by paths with high cumulative evolution probabilities are considered high-probability evolution zones. For example, spatial units R, S, and T, which are traversed by path P1, are considered high-probability evolution zones due to the high cumulative evolution probability of path P1. This means that these areas are more likely to undergo spatial morphological evolution in the future.

[0106] Neighborhood coevolution corridors are determined based on the adjacency relationships and evolutionary correlation strengths between spatial units. Within the main evolutionary path set, regions connected by edges with strong evolutionary correlation strengths between adjacent spatial units form neighborhood coevolution corridors. For example, if the evolutionary correlation strength between spatial units U and V is high and they frequently appear together in the main evolutionary path, the region between them constitutes a neighborhood coevolution corridor, indicating that these regions are likely to collaborate and influence each other in future evolutionary processes.

[0107] By overlaying and rendering the above information, an intuitive spatiotemporal evolution path map is generated, which can clearly show the morphological change trend of the block in the historical development process and its possible future evolution direction.

[0108] Step S156: Mark the spatial unit sequence of the main evolution path set in the spatiotemporal evolution path map, and superimpose and match the path intersection nodes with the activity intensity peak areas in the human activity heat distribution map to generate a key evolution intervention node coordinate set.

[0109] In this embodiment, the spatial unit sequence of the main evolution path set is clearly marked on the generated spatiotemporal evolution path map. For example, if path P1 in the main evolution path set passes through spatial units ABCD, the spatial units and the connecting paths between them are clearly marked on the spatiotemporal evolution path map.

[0110] At the same time, the path intersection nodes are overlaid and matched with the human activity heat distribution map. For example, for path intersection node E, the activity intensity peak area corresponding to this node is searched in the human activity heat distribution map. Assuming that the activity intensity at node E reaches a peak within a certain time period in the human activity heat distribution map, this indicates that this area is important in terms of human activity.

[0111] By overlaying and matching all path intersection nodes with the peak activity intensity areas in the human activity heat distribution map, we identify key locations that are both on the main evolution path and have high human activity intensity. The coordinates of these locations are collected to generate a set of coordinates for key evolution intervention nodes. For example, after matching, the coordinate set of key evolution intervention nodes is {(x1, y1), (x2, y2), (x3, y3)...}. These coordinate points are very important for urban planners because they represent locations that require key attention and potential intervention during the spatial evolution of a block, helping to develop more reasonable planning strategies.

[0112] Step S157: extracting a set of spatial units within the historical morphological preservation area in the spatiotemporal evolution path map, calculating the variance value of its building density change sequence and the mean value of its functional conversion frequency, and generating a morphological stability index.

[0113] Accurately extract the set of spatial units within the historical morphological preservation area from the spatiotemporal evolution path map. Assume that this set of spatial units includes spatial units X, Y, and Z. For each spatial unit, collect the building density change sequence data. For example, the building density data for spatial unit X over the past 50 years is (30, 32, 31, 33, 32), the building density data for spatial unit Y is (40, 41, 40, 42, 41), and the building density data for spatial unit Z is (25, 26, 25, 27, 26).

[0114] Next, calculate the variance of the building density change series. For spatial unit X, first calculate the mean based on (30, 32, 31, 33, 32), then calculate the variance. Similarly, calculate the variance of spatial units Y and Z.

[0115] Next, the mean of the functional conversion frequency is calculated, and the variance of the building density change series calculated above and the mean of the functional conversion frequency are combined to generate a morphological stability index. This morphological stability index can reflect the stability of spatial units within the historical morphological preservation area.

[0116] Step S158: extracting a set of spatial units in a high-probability evolution area in the spatiotemporal evolution path map, and generating a maximum-probability evolution direction and an expected morphological transformation type according to their evolution probability distribution.

[0117] From the spatiotemporal evolution path map, identify the set of spatial units within the high-probability evolution zone, assuming these include spatial units M, N, and O. For each spatial unit, examine its evolution probability distribution. For example, the evolution probability distribution for spatial unit M shows a 0.6 probability of evolving toward an expanding form, a 0.3 probability of evolving toward a contracting form, and a 0.1 probability of maintaining its current form. Spatial unit N has a 0.7 probability of transforming its function into a commercial one, and a 0.3 probability of maintaining its current function. Spatial unit O has a 0.8 probability of increasing its building height, and a 0.2 probability of decreasing its building height.

[0118] Based on the above evolution probability distribution, the maximum probability evolution direction of each spatial unit is determined. For spatial unit M, the maximum probability evolution direction is expansion; for spatial unit N, the maximum probability evolution direction is functional transformation to commercial; and for spatial unit O, the maximum probability evolution direction is increasing building height.

[0119] At the same time, we summarize the expected morphological transformation types. Based on the above spatial units, the expected morphological transformation types include morphological expansion, functional conversion (such as residential to commercial), and building height change (increase). The above maximum probability evolution direction and expected morphological transformation types provide important information for planners to understand the development trends of high-probability evolution areas, helping to formulate corresponding planning strategies in advance.

[0120] Step S159: extracting the edge weight distribution of the neighborhood co-evolution corridor in the spatiotemporal evolution path map, and generating a corridor width adjustment parameter and a functional mixing degree improvement coefficient.

[0121] In this embodiment, the edge weight distribution of the neighborhood co-evolution corridor in the spatiotemporal evolution path map can be analyzed. Assume that the neighborhood co-evolution corridor includes the edges between spatial units A and B, B and C, and C and D, and their edge weights are 0.7, 0.8, and 0.75, respectively.

[0122] Then, a corridor width adjustment parameter is generated based on the edge weight distribution. For example, the corridor width adjustment parameter is calculated using a certain method (assuming it is based on the average or weighted average of the weights). Assuming the average value is used, (0.7 + 0.8 + 0.75) ÷ 3 = 0.75. This parameter can be used to guide the adjustment of the width of the neighborhood co-evolution corridor during planning. A higher weight may mean that a wider corridor is needed to promote coordinated development between neighborhoods. Assuming that based on past experience and planning requirements, when the corridor width adjustment parameter is between 0.7 and 0.8, it is recommended to set the corridor width to 15 meters (this is just a hypothetical correspondence; the actual relationship needs to be determined based on specific planning standards).

[0123] Regarding the functional mix improvement coefficient, this embodiment further analyzes the functions of the spatial units connected by the edges. Assume that spatial unit A is primarily residential, and spatial unit B is primarily commercial. The edge weight between the two is 0.7; the edge weight between spatial unit B and spatial unit C, whose function is leisure, is 0.8. The functional mix improvement coefficient is calculated by analyzing the functional differences and edge weights of these adjacent spatial units. For example, this embodiment provides a calculation method that takes into account factors such as edge weights and the complexity of functional type differences (assuming that functional types are divided into three categories: residential, commercial, and leisure, and different functional types are assigned different values. Here, the difference between residential and commercial is assigned a value of 2, the difference between commercial and leisure is assigned a value of 1.5, and the difference between residential and leisure is assigned a value of 1. The specific values are determined based on actual planning considerations). For the connection between A and B, the contribution to the functional mix improvement is 0.7 × 2 = 1.4; for the connection between B and C, the contribution to the functional mix improvement is 0.8 × 1.5 = 1.2. A comprehensive calculation of these contributions (assuming simple addition) yields a functional mix improvement coefficient of 1.4 + 1.2 = 2.6. This coefficient reflects the potential of neighborhood coevolution corridors in promoting the integration of different functional spatial units. A higher coefficient indicates a greater likelihood and demand for functional mix. This can be used to adjust land use planning and facility allocation to enhance functional mix and enhance the vitality and diversity of the neighborhood.

[0124] Step S1510: constructing a multidimensional decision parameter set according to the morphological stability index, the maximum probability evolution direction, the expected morphological transformation type, the corridor width adjustment parameter and the functional mixing degree improvement coefficient.

[0125] In this embodiment, the various parameters obtained by the previous calculation can be integrated, and the integrated multi-dimensional decision parameter set covers multiple key information of the target block in terms of spatial evolution.

[0126] Step S1511: input the multi-dimensional decision parameter set into the planning strategy generation model, and output the spatial morphology simulation results including the spatial control intensity level, update timing priority and functional adaptation optimization scheme.

[0127] In this embodiment, the constructed multidimensional decision parameter set can be input into the planning strategy generation model. The spatial control intensity level can be determined based on parameters such as the morphological stability index and the maximum probability evolution direction. For example, if the morphological stability index is low, indicating significant historical morphological changes in the area, a higher spatial control intensity may be required to guide its development in a reasonable direction. Conversely, for areas within high-probability evolution zones with a clear maximum probability evolution direction, targeted spatial control intensity may be required. Assuming that according to the model settings, when the morphological stability index is below 0.6, the spatial control intensity level is set to high, requiring strict restrictions on building height and density in planning. When the morphological stability index is between 0.6 and 0.8, the spatial control intensity level is set to medium, with relatively looser restrictions. When the morphological stability index is above 0.8, the spatial control intensity level is set to low. In this embodiment, the morphological stability index is 0.57, so the model outputs a high spatial control intensity level for this area.

[0128] The determination of update timing priority comprehensively considers factors such as the direction of evolution with the highest probability and the type of expected morphological transformation. Areas with a high probability of evolution and a greater impact of the expected morphological transformation on block development will be given a higher update timing priority. For example, for spatial units whose direction of evolution with the highest probability is functional transformation and involves important functional areas (such as transformation from residential to commercial center), the model will consider them to have a higher update timing priority and require priority planning and transformation. Assume that in this embodiment, areas within the high-probability evolution area involving functional transformation from residential to commercial center are determined by the model to have a high update timing priority, while some areas with relatively low evolution probability and less impact on the overall function of the block will have a lower update timing priority.

[0129] The functional adaptation optimization plan is formulated based on the corridor width adjustment parameter, the functional mixing improvement coefficient, and the functional status and expected evolution direction of each spatial unit. For example, based on the corridor width adjustment parameter of 0.75, combined with the planning standards, it is determined that the road in the neighborhood co-evolution corridor area will be widened to 15 meters to promote the flow of people and activities and enhance functional mixing. For the case where the functional mixing improvement coefficient is 2.6, the model recommends increasing the configuration of multi-functional facilities in the relevant area, such as building a comprehensive building integrating commercial, leisure, and residential services to enhance functional mixing. At the same time, functional adaptation is optimized based on the expected morphological transformation types of different spatial units. For example, for spatial units with expected building height increases, supporting infrastructure is planned, such as increasing the capacity of electricity, water supply and other facilities to adapt to the changes in demand brought about by the new building height.

[0130] By processing a multidimensional set of decision-making parameters through a planning strategy generation model, the final output is a spatial morphology simulation result that includes the spatial control intensity level, update timing priority, and functional adaptation optimization plan. This spatial morphology simulation result provides specific planning guidance for urban planners, helping them to formulate planning schemes that are more in line with the development needs of the block, and achieve scientific planning and effective intervention in the spatial morphology of historical and cultural blocks.

[0131] Furthermore, after step S120, the method further includes:

[0132] Step S201: performing integrity check on the target spatiotemporal topological data set to detect missing data spatial units and abnormal fluctuation data points.

[0133] After generating the target spatiotemporal topological data set, a completeness check is performed on it. For example, taking the spatial units in a block as an example, the target spatiotemporal topological data set contains a variety of information for each spatial unit, such as spatial unit boundary data, building density change sequence, building height change sequence, functional zoning conversion sequence, activity intensity time series curve, and activity type distribution ratio.

[0134] For spatial unit boundary data, check whether any boundary vertex coordinate information for any spatial unit is missing. For example, in the boundary data of spatial unit K, the coordinates of a vertex may not be recorded, which is missing data. For the building density change series, check whether any time period is missing data. For example, the building density change series of spatial unit L has no records between 1980 and 1990, which is a manifestation of missing data spatial units.

[0135] For example, consider a series of building height changes. Suppose the building height of spatial unit M has been steadily increasing over the past few decades, perhaps by 1-2 meters annually, but suddenly increases by 10 meters in a single year. This significant deviation from the normal trend would constitute an abnormal fluctuation data point. Anomalies can also occur in activity intensity time series curves. For example, the activity intensity of spatial unit N typically remains within a certain range on a normal working day, but suddenly reaches an extremely high level one day, far exceeding the normal fluctuation range. This constitutes an abnormal fluctuation data point in the activity intensity time series curve.

[0136] Step S202: interpolation and completion processing is performed on the detected missing data spatial units, and the spatiotemporal topological data of adjacent spatial units are called to perform feature migration filling.

[0137] For the detected missing data space units, interpolation completion and feature migration filling methods are used to process them.

[0138] For other types of data, such as missing vertex coordinates in spatial unit boundary data, they can also be migrated and filled through the topological relationship and geometric shape features of adjacent spatial units. For example, the boundary shape and topological adjacency of adjacent spatial units have certain similarities. Based on the vertex coordinates and topological relationship of adjacent spatial units, the approximate position of the missing vertex coordinates can be inferred. Assuming that a vertex coordinate of an adjacent spatial unit has a parallel or similar geometric relationship with the edge where the missing vertex coordinate is located, the possible value of the missing vertex coordinate can be calculated through this relationship, thereby realizing interpolation completion and feature migration filling of the missing data spatial unit, making the target spatiotemporal topological data set more complete and accurate.

[0139] Step S203: performing smoothing correction processing on the abnormal fluctuation data points, and replacing abnormal values that exceed the set range based on the time series trend forecast value.

[0140] For the detected abnormal fluctuation data points, taking the abnormal growth of building height in spatial unit M as an example, this embodiment first analyzes the time series trend of building height changes. Analysis of building height data over the past several years reveals that the growth trend generally conforms to a linear growth model (hypothesis), for example, an annual increase of approximately 1.5 meters.

[0141] Set a reasonable fluctuation range, assuming that the normal fluctuation range is within ±20% of the average value. In a certain year, the building height of the space unit M suddenly increased by 10 meters, which obviously exceeded the normal fluctuation range and was an outlier. This embodiment is corrected based on the time series trend forecast value. According to the previous growth trend, it is predicted that the growth value of the building height in this year should be the height of the previous year plus 1.5 meters. Assuming that the building height in the previous year was 30 meters, it is predicted that the building height in this year should be 31.5 meters.

[0142] The outlier value of 10 meters, which falls outside the set range, is replaced with the predicted value of 31.5 meters, smoothing out the abnormally fluctuating data points. A similar approach is used for the abnormally fluctuating data points in the activity intensity time series curve. Analyzing the time series trend of activity intensity, suppose it is found that activity intensity on weekdays exhibits a certain periodicity and regularity. On a particular day, activity intensity suddenly reaches an extremely high value, exceeding the normal fluctuation range. By analyzing activity intensity data from the same time period on past weekdays, a time series model (such as a moving average model) is established to predict the normal activity intensity value for that time period. Assuming the predicted value is 50 people / hour, and the actual abnormal value is 150 people / hour, the abnormal value of 150 people / hour is replaced with the predicted value of 50 people / hour. This smooths out the activity intensity time series curve and reduces the interference of abnormal data on subsequent analysis and model training.

[0143] Step S204: Divide the verified target spatiotemporal topology data set into a training set and a validation set, and use the training set to tune the parameters of the spatiotemporal evolution feature extraction model and the spatiotemporal evolution prediction model, and use the validation set to evaluate the accuracy of the spatial morphology simulation results.

[0144] After integrity verification and data processing, the target spatiotemporal topological data set is divided into two parts. According to a certain ratio (assuming 70% as training set and 30% as validation set), the data of some spatial units are randomly selected to form the training set, and the rest are used as the validation set.

[0145] For example, in one possible implementation, step S204 may include:

[0146] Step S2041, feature extraction is performed on the target spatiotemporal topological data set of each spatial unit in the training set to generate a training feature set, wherein the training feature set includes the static structural features, dynamic evolution pattern features and spatial interaction dependency features of each spatial unit in the training set.

[0147] For example, for spatial unit X in the training set, we extract its static structural features, dynamic evolution pattern features, and spatial interaction dependency features using the previously described feature extraction method to generate a training feature set. This training feature set contains multiple feature information for each spatial unit in the training set. The above feature sets for all spatial units in the training set are combined to form a complete training feature set.

[0148] Step S2042: input the training feature set into the spatiotemporal evolution prediction model to generate a predicted evolution probability distribution and predicted evolution association strength for each spatial unit in the training set.

[0149] Step S2043, calculate the first error value between the predicted evolution probability distribution and the actual evolution probability distribution of each spatial unit in the training set, and the second error value between the predicted evolution association strength and the actual evolution association strength between adjacent spatial units in the training set.

[0150] In this embodiment, the first error value can be calculated by a conventional error calculation method (such as absolute error or mean square error, etc., where absolute error is assumed to be used). The same applies to the error calculation between the predicted evolution correlation strength and the actual evolution correlation strength.

[0151] Step S2044: construct a joint loss function based on the first error value and the second error value, and use the backpropagation algorithm to adjust the convolution kernel weights of the spatiotemporal evolution feature extraction model and the fully connected layer parameters of the spatiotemporal evolution prediction model.

[0152] For example, suppose the joint loss function is the weighted sum of the two error values. Using this joint loss function, the backpropagation algorithm is used to adjust the convolution kernel weights of the spatiotemporal evolution feature extraction model and the fully connected layer parameters of the spatiotemporal evolution prediction model. Based on the value of the loss function, the backpropagation algorithm backpropagates the error and adjusts the model parameters, gradually reducing the loss function value and thus optimizing model performance.

[0153] Step S2045 , inputting the target spatiotemporal topological data set of each spatial unit in the verification set into the spatiotemporal evolution feature extraction model after parameter tuning to generate a verification feature set.

[0154] For example, after feature extraction, the spatial unit Z in the verification set obtains its verification feature set, which includes static structural features, dynamic evolution pattern features and spatial interaction dependency features.

[0155] Step S2046: Input the verification feature set into the spatiotemporal evolution prediction model after parameter tuning to generate a verification evolution probability distribution and a verification evolution correlation strength.

[0156] Specifically, the parameter-tuned spatiotemporal evolution prediction model has optimized its structure and parameters in the hope of more accurately analyzing and predicting the input data. For each spatial unit in the validation set, a comprehensive analysis is performed based on the static structural features, dynamic evolution pattern features, and spatial interaction dependency features in its validation feature set. This ultimately generates a validation evolution probability distribution for that spatial unit. The validation evolution correlation strength between that spatial unit and its adjacent spatial units is also calculated. By performing this process on all spatial units in the validation set, a complete validation evolution probability distribution and validation evolution correlation strength data are obtained.

[0157] Step S2047: Extract the true evolution probability distribution of each spatial unit in the verification set and the true evolution correlation strength between adjacent spatial units, calculate the third error value between the verification evolution probability distribution and the true evolution probability distribution, and the fourth error value between the verification evolution correlation strength and the true evolution correlation strength.

[0158] First, we accurately extract the true evolution probability distribution information for each spatial unit in the validation set from pre-collected and organized real data records. This real data may come from long-term field observations, historical record analysis, or other reliable data sources, reflecting the true probability of different evolution states of spatial units during actual development.

[0159] At the same time, the real evolution correlation strength between adjacent spatial units is obtained, which reflects the real mutual influence and synergy between the spatial units and the surrounding units during the actual evolution process.

[0160] Next, for the third error value between the verified evolution probability distribution and the true evolution probability distribution, a suitable error calculation method, such as absolute error or mean square error, is used.

[0161] The fourth error value between the verified evolution correlation strength and the true evolution correlation strength is also calculated according to the selected error calculation method.

[0162] The above error calculation is performed on each spatial unit and its adjacent units in the validation set to comprehensively evaluate the degree of deviation between the model prediction results and the actual situation.

[0163] Step S2048: Generate an accuracy evaluation index based on the third error value and the fourth error value. When the accuracy evaluation index does not reach the preset threshold, increase the number of samples in the training set and re-execute the parameter tuning step until the accuracy evaluation index reaches the preset threshold.

[0164] Specifically, the accuracy evaluation index is generated according to the actual situation and needs. For example, suppose the accuracy evaluation index is the reciprocal of the average of the two error values.

[0165] The calculated accuracy evaluation index is compared with a preset threshold. The preset threshold is a standard value set in advance based on factors such as project requirements, experience, and expectations for model accuracy.

[0166] When the accuracy evaluation index does not reach the preset threshold, it means that the prediction accuracy of the current model has not met the requirements and needs further optimization. In this case, some additional spatial unit data are randomly selected from the original data set and added to the training set to increase the number of samples in the training set.

[0167] After increasing the number of samples, the parameter tuning step is repeated until the accuracy evaluation index reaches the preset threshold. At this point, the model's prediction accuracy meets the requirements, and the current spatiotemporal evolution feature extraction model and spatiotemporal evolution prediction model are used as the final tuning model. This final tuning model can more accurately predict and analyze the spatial unit evolution of the target block.

[0168] Figure 2 A schematic diagram illustrates exemplary hardware and software components of a system 100 for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis, which can implement the concepts of the present invention, according to some embodiments of the present invention. For example, a processor 120 can be used in the system 100 for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis, and can be used to perform the functions of the present invention.

[0169] The historical and cultural block spatial evolution simulation system 100 based on GIS spatiotemporal analysis can be a general-purpose server or a special-purpose server, both of which can be used to implement the historical and cultural block spatial evolution simulation method based on GIS spatiotemporal analysis of the present invention.

[0170] For example, the historical and cultural block spatial evolution simulation system 100 based on GIS spatiotemporal analysis may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and storage media 140 in different forms, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the historical and cultural block spatial evolution simulation system 100 based on GIS spatiotemporal analysis may also include program instructions stored in ROM, RAM, or other types of non-temporary storage media, or any combination thereof. The method of the present invention can be implemented according to the above-mentioned program instructions. The historical and cultural block spatial evolution simulation system 100 based on GIS spatiotemporal analysis also includes an I / O interface 150 between the computer and other input and output devices.

[0171] In addition, an embodiment of the present invention also provides a readable storage medium, which has computer-executable instructions preset in the readable storage medium. When the processor executes the computer-executable instructions, the above-mentioned historical and cultural block spatial evolution simulation method based on GIS spatiotemporal analysis is implemented.

[0172] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, multiple features are sometimes combined into one embodiment, figure or description thereof.

Claims

1. A method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis, characterized by: The method comprises: Obtaining a target spatiotemporal data set for a target block, wherein the target spatiotemporal data set includes geographic spatial distribution data, architectural form change data, and human activity trajectory data for a historical period; Performing spatiotemporal topological structure preprocessing on the target spatiotemporal data set to generate a target spatiotemporal topological data set, wherein the target spatiotemporal topological data set includes a spatiotemporally associated block spatial unit sequence and a dynamic attribute label of each spatial unit in the block spatial unit sequence; Based on a preset spatiotemporal evolution feature extraction model, feature extraction is performed on the target spatiotemporal topological data set to obtain the static structural features, dynamic evolution pattern features and spatial interaction dependency features of each spatial unit; Calling the spatiotemporal evolution prediction model to perform spatiotemporal evolution prediction on the static structural features, dynamic evolution pattern features, and spatial interaction dependency features, and generate the evolution probability distribution of the spatial unit and the evolution correlation strength between adjacent spatial units; Constructing a spatiotemporal evolution path map according to the evolution probability distribution and the evolution correlation strength, and generating spatial morphological simulation results and planning intervention decision parameters for the target block based on the spatiotemporal evolution path map; The target spatiotemporal topological data set is subjected to feature extraction based on the preset spatiotemporal evolution feature extraction model to obtain the static structural features, dynamic evolution pattern features, and spatial interaction dependency features of each spatial unit, including: Extracting spatial unit boundary data from the target spatiotemporal topological data set, calling a spatial morphology encoder to generate an initial structural feature vector for each spatial unit, wherein the initial structural feature vector includes shape complexity, area change trend, and number of adjacent units; Performing temporal convolution processing on the building density change sequence, building height change sequence, and functional zoning conversion sequence in the target spatiotemporal topological data set to generate a first dynamic evolution feature vector, wherein the first dynamic evolution feature vector includes a building density fluctuation period, a height growth gradient, and a functional conversion trigger threshold; Performing spatiotemporal attention processing on the activity intensity time series curve and the activity type distribution ratio in the target spatiotemporal topological data set to generate a second dynamic evolution feature vector, wherein the second dynamic evolution feature vector includes the activity intensity peak phase, type mixing degree, and spatial diffusion direction; Performing cross-scale feature splicing on the initial structural feature vector, the first dynamic evolution feature vector, and the second dynamic evolution feature vector to obtain a static structural feature; Performing graph convolution processing on the topological adjacency relationship in the target spatiotemporal topological data set to generate spatial interaction dependency features, wherein the spatial interaction dependency features include a density conduction coefficient, a functional compatibility index, and an activity diffusion resistance value between adjacent units; Dynamically weighting and correcting the static structural features based on the dynamic attribute labels in the target spatiotemporal topological data set to obtain dynamic evolution pattern features; The generating of the spatial morphology simulation results and planning intervention decision parameters of the target block based on the spatiotemporal evolution path map includes: Extracting a set of spatial units within the historical morphological preservation area in the spatiotemporal evolution path map, calculating the variance value of its building density change sequence and the mean value of its functional conversion frequency, and generating a morphological stability index; Extracting a set of spatial units in a high-probability evolution area in the spatiotemporal evolution path map, and generating a maximum-probability evolution direction and an expected morphological transformation type based on their evolution probability distribution; Extracting the edge weight distribution of the neighborhood co-evolution corridor in the spatiotemporal evolution path map to generate corridor width adjustment parameters and functional mixing degree improvement coefficients; Constructing a multidimensional decision parameter set based on the morphological stability index, the maximum probability evolution direction, the expected morphological transformation type, the corridor width adjustment parameter, and the functional mixing degree improvement coefficient; The multi-dimensional decision parameter set is input into the planning strategy generation model, and the spatial morphology simulation results including the spatial control intensity level, update timing priority and functional adaptation optimization scheme are output.

2. The method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis according to claim 1 is characterized in that: The performing spatiotemporal topological structure preprocessing on the target spatiotemporal data set to generate the target spatiotemporal topological data set includes: Performing coordinate calibration processing on the geographic spatial distribution data to obtain spatial unit boundary data in a unified coordinate system, wherein the spatial unit boundary data includes geometric shape vertex coordinates and topological adjacency relationships of each spatial unit; Performing time series alignment processing on the building form change data to obtain a building density change sequence, a building height change sequence, and a functional zoning conversion sequence that match the spatial unit boundary data; Performing spatial rasterization processing on the human activity trajectory data to generate a human activity heat distribution map corresponding to the spatial unit boundary data, and extracting the activity intensity time series curve and activity type distribution ratio of each spatial unit in the human activity heat distribution map; The spatial unit boundary data, building density change sequence, building height change sequence, functional zoning conversion sequence, activity intensity time series curve and activity type distribution ratio are subjected to spatiotemporal topological association encoding to generate the target spatiotemporal topological data set, wherein the dynamic attribute label of each spatial unit includes the building density change rate, functional conversion frequency and activity intensity fluctuation amplitude.

3. The method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis according to claim 1 is characterized in that: The performing graph convolution processing on the topological adjacency relationship in the target spatiotemporal topological data set to generate spatial interaction dependency features includes: Constructing an undirected graph structure with the spatial units in the spatial unit boundary data as nodes and the topological adjacency relationships as edges; Initializing an initial feature vector for each node in the undirected graph structure, where the initial feature vector is generated by concatenating the shape complexity, area change trend, and number of adjacent units in the static structural features with the building density fluctuation period and function conversion trigger threshold in the dynamic evolution pattern features; Performing multi-layer graph convolutional network processing on the undirected graph structure, aggregating the adjacent node feature vectors of each node layer by layer, and generating intermediate spatial interaction features; After each layer of graph convolutional network processing, the intermediate spatial interaction features are input into the gated recurrent unit, and the neighborhood feature fusion weights are generated based on the function conversion frequency in the dynamic attribute label. The intermediate spatial interaction features are weighted superimposed with the current node feature vector, and the node feature vector is updated to obtain the node feature vector output by the final layer; The node feature vectors output by the final layer are spatially aligned, and the feature vector of each node is spatially encoded with the geometric shape vertex coordinates of the spatial unit boundary data to generate spatial interaction dependency features including density conduction gradient, functional compatibility matrix and activity diffusion direction vector.

4. The method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis according to claim 1 is characterized in that: The calling of the spatiotemporal evolution prediction model to perform spatiotemporal evolution prediction on the static structural features, dynamic evolution pattern features, and spatial interaction dependency features, and generate the evolution probability distribution of the spatial unit and the evolution correlation strength between adjacent spatial units, includes: Inputting the static structural features into the first branch network of the spatiotemporal weight distribution model to generate an internal evolution weight coefficient of the spatial unit, wherein the internal evolution weight coefficient reflects the degree of influence of the historical morphology on the current evolution direction; Inputting the dynamic evolution pattern characteristics into the second branch network of the spatiotemporal weight allocation model to generate a time decay adjustment coefficient, wherein the time decay adjustment coefficient reflects the decay rate of contribution of data in different periods to the prediction result; Inputting the spatial interaction dependency feature into the third branch network of the spatiotemporal weight allocation model to generate a neighborhood coupling strength coefficient, wherein the neighborhood coupling strength coefficient reflects the cooperative or competitive relationship between the evolution behaviors of adjacent spatial units; A spatiotemporal joint weight matrix is constructed according to the internal evolution weight coefficient, the time decay adjustment coefficient, and the neighborhood coupling strength coefficient, and the static structural features, the dynamic evolution pattern features, and the spatial interaction dependency features are weightedly fused to generate a spatiotemporal fusion feature vector; Inputting the spatiotemporal fusion feature vector into the evolution probability prediction layer to generate the evolution probability distribution of the spatial unit; The spatiotemporal fusion feature vector is input into the association strength prediction layer to generate the evolving association strength between adjacent spatial units.

5. The method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis according to claim 4 is characterized in that: The constructing of a spatiotemporal evolution path map according to the evolution probability distribution and the evolution correlation strength includes: Taking each spatial unit as a node and the evolution correlation strength as the edge weight, an initial spatiotemporal evolution graph structure is constructed; updating the node states of the initial spatiotemporal evolution graph structure according to the evolution probability distribution to generate a spatiotemporal evolution graph with probability weights; Performing path search processing on the spatiotemporal evolution graph with probability weights, extracting edges with probability weights greater than a set weight to form a main evolution path set; Performing loop detection on the main evolution path set, and if a closed loop is detected, performing attenuation correction on the evolution probability within the loop according to the time attenuation adjustment coefficient; The attenuation-corrected main evolution path set is superimposed and rendered with the spatial unit boundary data to generate the spatiotemporal evolution path map, which includes historical morphological preservation areas, high-probability evolution areas, and neighborhood co-evolution corridors.

6. The method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis according to claim 1 is characterized in that: The multi-dimensional decision parameter set is input into the planning strategy generation model, and the spatial morphology simulation results including the spatial control intensity level, update timing priority and functional adaptation optimization scheme are output, including: The morphological stability index is graded, and when the morphological stability index is lower than a stability threshold, a morphological protection strategy is activated to generate building height restriction parameters and material use constraint conditions; Performing conflict detection on the maximum probability evolution direction, and when it is detected that the maximum probability evolution direction conflicts with the historical context protection requirements, inserting a direction correction factor and recalculating the evolution probability distribution; Conducting a carrying capacity assessment for the expected type of morphological transformation, and generating an infrastructure expansion demand list and a traffic organization optimization plan when the assessment result indicates that the carrying capacity exceeds a preset capacity; Calculate the infiltration buffer distance between adjacent functional areas according to the corridor width adjustment parameter and generate an interface transition treatment strategy; Adjust the land compatibility rules based on the functional mixing degree improvement coefficient to generate mixed function ratio control parameters and public space configuration standards; The morphological protection strategy, direction correction factor, infrastructure expansion demand list, traffic organization optimization plan, interface transition processing strategy and mixed function ratio control parameters are sorted by strategy priority to generate the planning intervention decision parameters.

7. The method for simulating the spatial evolution of historical and cultural blocks based on GIS spatiotemporal analysis according to claim 1 is characterized in that: After performing spatiotemporal topological structure preprocessing on the target spatiotemporal data set to generate the target spatiotemporal topological data set, the method further includes: Performing integrity check on the target spatiotemporal topological data set to detect missing data spatial units and abnormal fluctuation data points; Perform interpolation and completion processing on the detected missing data spatial units, and call the spatiotemporal topological data of adjacent spatial units for feature migration and filling; Perform smoothing and correction processing on the abnormal fluctuation data points, and replace abnormal values that exceed the set range based on the time series trend forecast value; The verified target spatiotemporal topological data set is divided into a training set and a validation set, and the parameters of the spatiotemporal evolution feature extraction model and the spatiotemporal evolution prediction model are tuned through the training set, and the accuracy of the spatial morphology simulation results is evaluated through the validation set.

8. A historical and cultural block spatial evolution simulation system based on GIS spatiotemporal analysis, characterized by: It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the spatial evolution simulation method of historical and cultural blocks based on GIS spatiotemporal analysis as described in any one of claims 1 to 7.

Images