Historical and cultural oriented space classification and regeneration method and system

Abstract

The invention relates to the technical field of urban planning, in particular to a historical and cultural oriented space classification and regeneration method and system. The method comprises the following steps: acquiring historical culture data, and performing multi-modal perception on the historical culture data to obtain historical culture spatial data; performing deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; performing spatial semantic map construction on the historical and cultural element data to obtain spatial semantic map data; performing age evolution trajectory classification on the spatial semantic map data to obtain historical and cultural spatial classification data; performing space regeneration generation according to the historical and cultural space classification data to obtain space regeneration data; and acquiring urban planning demand data, and performing spatial regeneration fusion according to the urban planning demand data and the spatial regeneration data to obtain spatial regeneration fusion data. Space regeneration is carried out in combination with city planning, and dynamic balance between historical and cultural protection and modern city development is effectively achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of urban planning, and in particular to a history and culture-oriented space classification and regeneration method and system. Background Art

[0002] With the acceleration of urbanization and the growing demand for cultural heritage preservation, typical historical and cultural spaces, such as traditional blocks, traditional architectural complexes, and traditional style districts, face the real challenges of complex spatial structures and diverse functional reuse requirements during urban renewal. However, existing technologies for identifying and managing such spaces rely heavily on manual research and template rules, lacking the ability to perceive and represent heterogeneous historical data such as images, text, and GIS in a unified, structured manner. This makes it difficult to automatically extract and model spatial morphology, functional semantics, and cultural evolution. Artificial intelligence (AI) is a new technological discipline that studies and develops theories, methods, techniques, and application systems for simulating, extending, and expanding human intelligence. AI is a branch of computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in ways similar to human intelligence. The challenge is how to integrate AI with the generation of uses for historical and cultural spaces and intelligent reasoning for spatial regeneration strategies. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes a history and culture-oriented space classification and regeneration method and system to solve at least one of the above technical problems.

[0004] This application provides a historical and culturally oriented spatial classification and regeneration method, including the following steps: Step S1: Acquire historical and cultural data, perform multimodal perception on the historical and cultural data, and obtain historical and cultural spatial data; perform deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; Step S2: constructing a spatial semantic map of the historical and cultural element data to obtain spatial semantic map data; classifying the spatial semantic map data by chronological evolution trajectory to obtain historical and cultural spatial classification data; Step S3: performing spatial regeneration based on the historical and cultural space classification data to obtain spatial regeneration data; Step S4: Acquire urban planning demand data, and perform spatial regeneration fusion according to the urban planning demand data and the spatial regeneration data to obtain spatial regeneration fusion data.

[0005] The present invention uses multimodal perception and deep feature extraction to achieve refined identification of historical and cultural spaces and precise extraction of cultural elements. Compared with traditional single-mode perception technology, it significantly improves the recognition accuracy of cultural spaces and the integrity of element preservation. Through the construction of spatial semantic maps and the classification of chronological evolution trajectories, it can not only reveal the time series characteristics of spatial evolution, but also automatically classify spatial groups with different cultural value evolution patterns, providing a scientific basis for spatial regeneration. Combining spatial regeneration fusion with urban planning demand data effectively achieves a dynamic balance between historical and cultural protection and modern urban development, avoiding the problem of disconnection between regeneration plans and actual planning needs in existing methods.

[0006] Preferably, step S1 is specifically: Access to historical and cultural data; Perform multimodal feature extraction on historical and cultural data to obtain multimodal feature data; Perform feature splicing on the multimodal feature data to obtain feature splicing data; Perform self-attention encoding on the feature splicing data to obtain feature attention data; Perform fusion reasoning on feature attention data to obtain historical and cultural space data; Perform deep feature extraction on historical and cultural spatial data to obtain deep feature data; The cultural attributes of historical and cultural space data are annotated based on the deep feature data to obtain historical and cultural element data.

[0007] In the historical and cultural data processing stage of the present invention, multi-source information such as images, texts and spatial structures is integrated through multi-modal feature extraction and feature splicing, effectively improving the integrity and information richness of historical and cultural space identification. The self-attention encoding mechanism is used to assign importance weights to multi-modal features, thereby enhancing the ability to extract key cultural information. Through the fusion reasoning process, deep correlation and semantic consistency inference of different modal features are achieved, significantly improving the representation quality of historical and cultural space data. On this basis, deep feature extraction and cultural attribute labeling are carried out to achieve fine-grained recognition and accurate classification of historical and cultural elements.

[0008] Preferably, the fusion reasoning is specifically: Perform cross-modal correlation processing on the feature attention data to obtain cross-modal correlation data; Perform modal consistency optimization on cross-modal correlation data to obtain characteristic modal optimization data; Perform semantic completion reasoning on the feature modality optimization data to obtain feature modality completion data; The spatial structure consistency of the characteristic modal completion data is fused to obtain historical and cultural spatial data.

[0009] By performing cross-modal correlation processing on feature attention data, the present invention can accurately model the deep connections between different modal features, thereby improving the accuracy of multi-source feature fusion of historical and cultural data. Through modal consistency optimization processing, information redundancy and conflict between modalities are further eliminated, and the consistency and coordination of the overall feature expression are enhanced. Combined with semantic completion reasoning, it effectively solves the problem of missing or incomplete semantics of some modal data, and improves the completeness of spatial semantic reasoning. Through spatial structure consistency fusion, the reasonable coherence of the fusion results in spatial distribution and cultural structure logic is ensured.

[0010] Preferably, step S2 is specifically: Element nodes and spatial relationship edges are constructed for historical and cultural element data to obtain element node data and spatial relationship edge data respectively; Generate a spatial semantic graph based on element node data and spatial relationship edge data to obtain spatial semantic graph data; Extract the chronological evolution trajectory based on the spatial semantic atlas data to obtain the chronological evolution trajectory data; Perform trajectory clustering on the chronological evolution trajectory data to obtain trajectory cluster data; The cultural space classification of the chronological evolution trajectory data is carried out according to the trajectory clustering data to obtain the historical and cultural space classification data.

[0011] In the present invention, by constructing element nodes and spatial relationship edges for historical and cultural element data, the core elements and their spatial associations in the historical and cultural space are systematically extracted, thereby improving the structured expression ability of cultural information organization. Based on the spatial semantic map generated in this way, not only the bidirectional association between cultural elements and spatial positions is realized, but also rich spatiotemporal evolution clues are provided for the subsequent extraction of evolution trajectories. By clustering the trajectory data of the evolution trajectory of the era, it is possible to effectively identify the cultural space change groups under different historical evolution models, avoiding the local one-sidedness brought about by the traditional single time axis analysis. Finally, cultural space classification is carried out based on the trajectory clustering results, and the spatial types and cultural evolution characteristics are accurately divided, which greatly improves the scientific nature of historical and cultural space management and regeneration decision-making.

[0012] Preferably, the spatial semantic graph data includes first spatial semantic graph data and second spatial semantic graph data, the time parameter corresponding to the first spatial semantic graph data is not greater than the time parameter corresponding to the second spatial semantic graph data, and the chronological evolution trajectory extraction is specifically as follows: Performing time-constrained node association pairing on the first spatial semantic graph data and the second spatial semantic graph data to obtain time-constrained node pair data; Determine the node pair evolution relationship of the time-constrained node pair data to obtain the node evolution data; The spatial connectivity is concatenated according to the node evolution data to obtain the chronological evolution trajectory data.

[0013] The present invention effectively ensures the logical continuity of historical and cultural spatial nodes in the time series through node association pairing based on time parameter constraints, avoiding the problems of arbitrary node association and broken evolution chain in existing methods. By determining the evolutionary relationship between nodes, it is possible to identify the evolutionary patterns between different time nodes in a fine-grained manner, such as functional continuity, cultural transformation or spatial disappearance, which significantly improves the accuracy and explanatory power of cultural evolution trajectory modeling. By connecting spatial connectivity in series, not only the evolutionary path between nodes is reconstructed, but also chronological evolution trajectory data with spatial consistency and cultural logical coherence is formed.

[0014] Preferably, the trajectory clustering is specifically: Perform spatial node extraction and cultural semantic node extraction on the chronological evolution trajectory data to obtain spatial node data and cultural semantic node data; Pairing spatial node data and cultural semantic node data to obtain node pair data; Generate evolution trajectory of data according to nodes to obtain evolution trajectory data; Perform local trajectory repair on the evolving trajectory data to obtain trajectory repair data; The spatial path similarity matrix and the time span similarity matrix are constructed for the trajectory repair data to obtain the spatial similarity matrix data and the time span similarity matrix data respectively; The trajectory repair data is hierarchically clustered according to the spatial similarity matrix data and the time span similarity matrix data to obtain trajectory clustering data.

[0015] The present invention fully preserves the spatial evolution characteristics and cultural evolution semantics in the trajectory by extracting spatial nodes and cultural semantic nodes from the chronological evolution trajectory data, thereby improving the integrity of the trajectory data and the ability to express multi-dimensional information. By pairing nodes and generating evolutionary trajectories, a trajectory chain based on temporal, spatial and semantic continuity is reconstructed, avoiding the evolutionary fault problem caused by node isolation in traditional trajectory analysis. In response to noise or abnormal jumps in the generated trajectory, the consistency and availability of the trajectory data are improved through local trajectory repair processing. By constructing the spatial path similarity matrix and the time span similarity matrix respectively, the dual similarity measurement of the trajectory in the spatial dimension and the temporal dimension is realized, and the multi-scale recognition ability of the trajectory clustering is enhanced. Through hierarchical clustering based on the dual matrix, the cultural space groups under different evolutionary modes are accurately distinguished.

[0016] Preferably, step S3 is specifically: Extract cultural evolution characteristics based on historical and cultural space classification data to obtain cultural evolution characteristic data, where the cultural evolution characteristic data includes spatial function evolution characteristic data, architectural form change characteristic data, cultural symbol evolution characteristic data, and time evolution trajectory characteristic data; The preset historical and cultural space evolution generation model is used to generate multi-path natural evolution samples of cultural evolution feature data to obtain candidate spatial regeneration data. The preset historical and cultural space evolution generation model is obtained by generative adversarial training based on pre-stored historical and cultural space classification experience data and spatial regeneration label data. Perform spatial coherence test on the candidate spatial regeneration data to obtain spatial regeneration data.

[0017] The present invention systematically identifies multi-dimensional features such as spatial functional evolution, architectural form changes, cultural symbol evolution, and time trajectory changes by extracting cultural evolution features based on historical and cultural space classification data, greatly enriching the evolutionary expression capabilities of cultural space. Multi-path natural evolution sample generation is performed through a preset historical and cultural space evolution generation model, effectively simulating the possible regeneration paths of space under different evolutionary scenarios, and improving the diversity and rationality of space regeneration solutions. The generation model is based on adversarial training of local historical experience data and spatial regeneration label data, ensuring that the generated results have both historical consistency and practical adaptability. Further, through spatial coherence testing, candidate samples with spatial logical breaks or functional conflicts are eliminated, and finally high-quality spatial regeneration data that conforms to the laws of spatial evolution and cultural protection requirements is output.

[0018] Preferably, the step of constructing the preset historical and cultural space evolution generation model includes the following steps: Perform feature expansion layer processing on the historical and cultural space classification experience data stored locally to obtain feature expansion data; Performing deep feature transformation processing on the feature expansion data to obtain deep feature data; Generating structured features on the deep feature data to obtain structured feature data; Performing feature upsampling processing according to the structured feature data to obtain feature upsampling data; Perform convolution processing on the feature upsampled data to obtain candidate space regeneration sample data; Performing convolutional coding block processing on the candidate spatially regenerated sample data and the spatially regenerated label data to obtain convolutional coding block data; Performing feature compression layer processing on the convolutional coding block data to obtain feature compression layer data; Perform authenticity scoring layer processing based on the feature compression layer data to obtain sample authenticity scoring data; The candidate space regeneration sample data is iteratively trained according to the sample authenticity score data to obtain the historical and cultural space evolution generation model.

[0019] In the present invention, by performing feature expansion layer processing and deep feature transformation on the locally pre-stored historical and cultural space classification experience data, the high-order feature relationships hidden in the spatial evolution process are fully excavated, and the expressive ability of evolution feature learning is improved. Through structured feature generation and feature upsampling, a multi-dimensional feature tensor with spatial semantics is reconstructed, laying the foundation for spatial layout reasoning. Combined with convolution processing to generate candidate space regeneration sample data, the local features of the space and the overall cultural evolution trend are effectively modeled. Further, through the hierarchical processing of convolutional coding blocks, feature compression layers and authenticity scoring layers, the semantic differences between spatial regeneration samples and real samples are accurately captured. Through adversarial iterative training based on sample authenticity scores, the generative model can adaptively optimize the generation strategy to obtain a historical and cultural space evolution generation model that has both historical and cultural rationality and spatial evolution naturalness.

[0020] Preferably, step S4 is specifically: Obtain urban planning demand data; Perform attribute mapping based on urban planning demand data and spatial regeneration data to obtain spatial planning mapping data; Performing conflict detection on the spatial planning mapping data to obtain spatial conflict detection data; Coordinately adjust the spatial conflict detection data to obtain spatial adjustment data; The spatial layout fusion is performed on the spatial adjustment data to obtain the spatial regeneration fusion data.

[0021] By acquiring urban planning demand data, the present invention can capture the dynamic requirements of urban development for spatial layout in real time, ensuring that the regeneration process of historical and cultural spaces is closely connected with the current urban functional planning. Through attribute mapping processing based on urban planning demand data and spatial regeneration data, the semantic correspondence and transformation of historical and cultural space functions and modern urban uses are effectively achieved. Through spatial conflict detection, potential functional contradictions, layout conflicts or resource overlap problems can be identified in a timely manner, thereby improving the pre-risk perception capability of spatial regeneration plans. Combined with collaborative adjustment processing, the adaptability between historical and cultural spaces and urban planning goals is dynamically optimized. Through spatial layout fusion operations, the coordinated unification of historical and cultural protection goals and urban space renewal needs is achieved.

[0022] Preferably, the present application further provides a historical and culturally oriented space classification and regeneration system for executing the above-mentioned historical and culturally oriented space classification and regeneration method, the historical and culturally oriented space classification and regeneration system comprising: The historical and cultural data processing module is used to obtain historical and cultural data, perform multimodal perception on the historical and cultural data, and obtain historical and cultural spatial data; perform deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; The spatial semantic modeling and trajectory classification module is used to construct a spatial semantic map of historical and cultural element data to obtain spatial semantic map data; and to classify the chronological evolution trajectory of the spatial semantic map data to obtain historical and cultural spatial classification data; A space regeneration generation module is used to generate space regeneration based on historical and cultural space classification data to obtain space regeneration data; The spatial regeneration fusion module is used to obtain urban planning demand data, and perform spatial regeneration fusion based on the urban planning demand data and spatial regeneration data to obtain spatial regeneration fusion data.

[0023] The beneficial effects of the present invention are: by performing multimodal perception and deep feature extraction on historical and cultural data in the initial stage, multi-source information such as images, texts and spatial structures is systematically integrated, which greatly improves the integrity and fine-grained expression capabilities of historical and cultural space identification. By constructing a spatial semantic map and performing chronological evolution trajectory classification, not only the dynamic modeling of historical and cultural elements in the space-time dimension is achieved, but also the context and pattern of spatial evolution can be revealed in detail, and spatial units with different cultural attributes and evolution characteristics can be accurately divided. Spatial regeneration is generated based on the spatial classification results, which effectively simulates multi-path natural evolution samples that conform to the laws of historical evolution, avoiding the limitations of traditional spatial regeneration methods that sever historical continuity. Attribute mapping, conflict detection and coordinated adjustment are carried out in combination with urban planning demand data to achieve a high degree of coordination and unity between historical and cultural protection goals and modern urban development needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 A flowchart showing the steps of a history and culture-oriented space classification and regeneration method according to one embodiment is shown; Figure 2 A flowchart showing the steps of a method for processing historical and cultural data according to an embodiment is shown; Figure 3 A flowchart showing the steps of a spatial semantic modeling and trajectory classification method according to an embodiment is shown; Figure 4 A flowchart showing the steps of a spatial regeneration generation method according to an embodiment is shown; Figure 5 A flowchart of the steps of a spatial regeneration and fusion method according to an embodiment is shown. DETAILED DESCRIPTION

[0025] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative work are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0027] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0028] See also Figures 1 to 5 , this application provides a historical and cultural oriented spatial classification and regeneration method, comprising the following steps: Step S1: Acquire historical and cultural data, perform multimodal perception on the historical and cultural data, and obtain historical and cultural spatial data; perform deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; In one embodiment, data collection and processing is performed. The sources of historical and cultural data include, but are not limited to, historical documents and archives, geographic information system (GIS) data, imagery (such as satellite imagery and historical maps), local chronicles, and oral history materials. Data collection methods include manual entry, batch scanning combined with text recognition using optical character recognition (OCR), imagery collection based on satellite remote sensing, and on-site street view photography. After data collection is completed, the system performs in-depth perception and processing of the historical and cultural data based on multimodal perception technology. This includes the following aspects: spatial perception: for data with geographic coordinate attributes, spatial registration is performed using methods such as affine transformation to spatially align historical maps with existing maps based on landmark points; text perception: for historical documents and archives, natural language processing (NLP) techniques are applied to perform keyword extraction and named entity recognition (NER) to identify place names, building names, and other geographical or cultural entity information in the documents; and image perception: for historical imagery, image segmentation techniques (such as a segmentation model based on a U-Net neural network) are used to process images and extract visual elements such as historical buildings, block boundaries, and natural landscapes. Based on the historical and cultural spatial data obtained through multimodal fusion perception, in-depth feature extraction is performed. This includes spatial feature extraction, which calculates the basic spatial geometric attributes of each cultural spatial unit, including plot area, shape complexity (defined as the ratio of perimeter to area), and the coordinates of the plot center point; semantic feature extraction, which comprehensively classifies the types of historical and cultural elements into categories such as architectural ruins, commercial areas, residential areas, traditional buildings, and transportation nodes based on the extracted text, images, and spatial features; and temporal feature extraction, which annotates the temporal attributes corresponding to each historical and cultural element based on documentary records, image timestamps, or morphological evolution inference methods. Based on the above feature extraction results, a structured historical and cultural spatial dataset is formed, in which each data record contains at least an element identifier (ID), element type, spatial range description, corresponding era annotation, and detailed feature description.

[0029] Step S2: constructing a spatial semantic map of the historical and cultural element data to obtain spatial semantic map data; classifying the spatial semantic map data by chronological evolution trajectory to obtain historical and cultural spatial classification data; In one embodiment, a spatial semantic graph is constructed. Each historical and cultural element is defined as a node in the graph, with each node corresponding to a specific cultural element entity, such as a building site, residential area, or traditional architecture. Spatial edges are constructed based on the relationships between nodes. When two nodes are spatially close (for example, the Euclidean distance between their centroids is less than 100 meters) and have a historical functional connection (for example, a temple and a market have a historical functional connection) (determined using a preset function table / parameter table, with expert knowledge assisting in table construction), an edge is established between the two nodes. Node and edge attributes are annotated, including: type (for example, traditional architecture, commercial facility, etc.); age; area (units can be standardized in square meters); and preservation status (intact, partially damaged, or severely damaged). Edge attributes are annotated: relationship type (for example, proximity, functional association, evolutionary inheritance, etc.); and weight (determined based on the spatial distance between nodes or the importance of the documented information; the closer the spatial distance or the more explicit the historical connection, the higher the weight). Through the above method, the spatial semantic map of historical and cultural elements is constructed, forming a complete map structure containing nodes, edges and their attribute information.

[0030] After the spatial semantic map is generated, a chronological evolution trajectory classification process is performed. This process specifically involves the following steps: Based on the chronological attributes of historical and cultural elements, all nodes are divided into corresponding time segments to form a temporally hierarchical data structure. Node evolution paths are identified and constructed based on evolution trajectory extraction rules. If a historical and cultural element exhibits a continuous existence or evolutionary relationship across different time segments (for example, a market gradually transformed into a marketplace and ultimately into a modern commercial center), a corresponding chronological evolution trajectory is established. This evolution trajectory is recorded using trajectory encoding, sequentially connecting the starting node, intermediate change nodes, and end node to form a clear evolutionary chain. For example, a trajectory encoding might represent the evolution of "marketplace" to "marketplace" and then to "modern commercial center." Trajectory similarity is calculated based on trajectory features (such as node change category and time span). Trajectory data is automatically classified using the K-Means clustering algorithm or the density-based clustering algorithm (DBSCAN) to extract representative groups of cultural evolution patterns. The output is historical and cultural spatial classification data, which includes detailed descriptions of each typical evolutionary pattern.

[0031] Step S3: performing spatial regeneration based on the historical and cultural space classification data to obtain spatial regeneration data; In one embodiment, spatial regeneration generation rules are formulated to clarify spatial function assignment and spatial layout optimization strategies. Regarding spatial function assignment, corresponding regeneration uses are assigned to different types of historical and cultural elements based on trajectory information extracted from historical and cultural space classification data. Specific assignment rules include, but are not limited to, assigning the regeneration use of traditional building trajectories as cultural exhibition halls, preserving their spiritual and cultural functions while enhancing their public accessibility; assigning the regeneration use of residential trajectories as cultural and creative industrial parks, integrating the characteristics of the historical residential environment to develop cultural and creative industries; and assigning the regeneration use of commercial node trajectories as distinctive commercial blocks, strengthening the existing commercial atmosphere and emphasizing the integration of historical block style preservation and commercial vitality. In terms of spatial layout optimization, based on the preservation status and spatial evolution characteristics reflected in the historical and cultural space classification data, the following regional categories are subdivided: Conservation Priority Zones, which target areas with a high degree of historical preservation (e.g., structural preservation rates exceeding 95%) and are designated as Conservation Priority Zones. These zones adhere to the principle of minimal intervention, with only necessary repairs and environmental improvements being carried out. Redevelopment Zones, which target spatial units with low levels of historical preservation and conveniently located in areas with convenient transportation, are designated as Key Redevelopment Zones, encouraging functional remodeling and spatial structural renewal. Transitional Coordination Zones, located between Conservation Priority Zones and Redevelopment Zones, are designated as Transitional Coordination Zones. Urban green spaces, outdoor exhibition facilities, and other measures are employed to flexibly connect conservation and development areas, alleviating the sense of spatial discontinuity. In the process of spatial regeneration, spatial generation technology is applied to support efficient layout optimization. Specifically, a Geographic Information System (GIS) platform was used to generate a reclaimed land zoning map based on the functional assignments, clarifying the geographic scope and boundaries of each functional area. Generative design algorithms were employed to optimize spatial layout, specifically using morphological transformation techniques to regularize and fine-tune land forms to improve land use efficiency. A Voronoi diagram partitioning algorithm was applied to generate a rational distribution pattern of building clusters centered around historical nodes, ensuring spatial coherence and visual order. Standardized spatial reclaimed data was generated, with each spatial unit containing information such as the reclaimed use label, geographic scope, and design guidance notes, using a format such as {"Area ID": "A001","Reclaimed Use": "Cultural and Creative Park","Area": "2.5 hectares","Design Notes": {...}}.

[0032] Step S4: Acquire urban planning demand data, and perform spatial regeneration fusion according to the urban planning demand data and the spatial regeneration data to obtain spatial regeneration fusion data.

[0033] In one embodiment, urban planning demand data collection and processing is performed. The urban planning demand data is mainly derived from official planning materials such as urban renewal plan documents, transportation hub layout maps, land use planning redline maps, and public service facility layout requirements issued by the government. The system standardizes the collected original planning data, specifically mapping the textual descriptive planning requirements into quantifiable specific indicators, such as the green space ratio should be no less than 30%; the public facility service coverage radius should not exceed 500 meters; the commercial functional area volume ratio should not exceed 2.0; and the graphic data (such as land use redline and transportation layout) are uniformly converted into the standard format of the Geographic Information System (GIS) to ensure consistency with the coordinate system of the historical and cultural spatial data. After standardizing planning requirements, spatial regeneration integration rules are formulated, encompassing priority regulation and spatial coordination adjustments. Regarding use priority regulation, if a conflict exists between historical and cultural regeneration uses and urban planning uses (e.g., a former cultural area is planned for a future transportation hub), the conflict is addressed according to pre-defined use priority criteria: If the regenerated area possesses high cultural value (e.g., meets Level 1 or 2 cultural heritage protection standards), cultural uses are prioritized. If the cultural value is low, or if there is a significant conflict between the regeneration use and urban development needs, urban planning uses are prioritized. Cultural symbols (e.g., small monuments and cultural landmarks) are incorporated to preserve a certain degree of historical memory. Regarding spatial coordination adjustments, the system implements the following specific rules: Building density in regenerated areas is adjusted based on the planned floor area ratio to ensure that construction intensity meets the requirements of the overall plan; necessary public nodes, such as plazas, bus stops, and community service centers, are added based on the layout requirements of urban public service facilities; and transportation connections in regenerated areas are optimized to ensure that the connectivity rate between regenerated units and the city's main transportation network reaches or exceeds 80%, enhancing accessibility and ease of use. After completing the aforementioned integration and adjustment operations, the system generates spatial regeneration integration data. This integration data includes updated regeneration unit attributes, such as the actual determined use category, the adjusted land area, information on allocated public facility nodes, and indicators such as the cultural retention rate under the integration plan. For example, a fused area might be recorded as: Integration Zone Number F001, Actual Use: Mixed Cultural Exhibition and Commercial Blocks, Green Space Ratio of 32%, and Main Transportation Connection Point: Exit C of Metro Line 3.

[0034] Preferably, step S1 is specifically: Step S11: Acquire historical and cultural data; In one embodiment, historical and cultural data includes historical archives, local chronicles, cultural relic protection catalogs, oral materials; satellite imagery, historical maps, street view photographs; GIS vector data (e.g., building points, road lines, and land parcels); and timestamp data (e.g., construction year, repair record time). Public data interfaces are captured (e.g., the National Cultural Heritage Administration API and data from local government platforms); OCR is used to recognize old documents, using text error correction and layout structure recognition; and remote sensing imagery is downloaded and spatially registered (e.g., landmark control point correction error < 2 meters).

[0035] Step S12: extracting multimodal features from the historical and cultural data to obtain multimodal feature data; In one embodiment, for the text modality, word segmentation is performed (such as Jieba, BERT word segmenter); key entities (such as building type, purpose, place name) are extracted; Embedding encoding (such as Word2Vec / sentence vector SBERT, the dimension is unified to 256 dimensions). For the image modality, image segmentation (such as DeepLabV3+) is performed to extract the building outline; visual feature extraction (ResNet, EfficientNet extract feature vectors, the dimension is such as 512 dimensions); local descriptors are extracted (such as SIFT, ORB). For the spatial modality, geographic attribute features are extracted, such as area, perimeter, location center, shape compactness ( , where C is the shape compactness, A is the area enclosed by the outline, and P is the perimeter of the outline); according to the spatial position relationship, analyze spatial structural characteristics such as proximity (such as shortest distance measurement) and connectivity (such as topological graph connected component analysis).

[0036] Step S13: performing feature splicing on the multimodal feature data to obtain feature splicing data; In one embodiment, features from different modalities are uniformly normalized. Direct vector concatenation is performed in a fixed order, arranged as follows: [text features | image features | spatial features]. Low-rank mapping is employed (e.g., reducing the feature space to a unified 512-dimensional feature space through linear transformation or sparse coding). The concatenated feature vectors maintain consistent dimensions (e.g., unified to 512 or 1024 dimensions). If missing features are present (e.g., missing images in historical data), mean filling or zero vector filling is employed.

[0037] Step S14: performing self-attention encoding on the feature splicing data to obtain feature attention data; In one embodiment, the self-attention encoding process includes the construction of query vectors, key vectors, and value vectors, the calculation of attention weights, and subspace feature encoding and fusion. For each concatenated feature vector, a linear transformation is performed to generate the corresponding Q (query vector), K (key vector), and V (value vector): 、 、 To train the learnable parameter matrix; for example, Q = concatenated features , K = splicing features , V = splicing feature Based on the similarity between the query vector and the key vector, the attention weight between each pair of features is calculated. The calculation formula of the attention weight is as follows ,in is the square root of the key vector dimension. The features are divided into several subspaces, self-attention is performed on each, and then concatenated. High-weight features indicate stronger historical and cultural relevance; weakly relevant features can be weakened. This yields a feature vector set after feature attention encoding.

[0038] Step S15: Perform fusion reasoning on the feature attention data to obtain historical and cultural space data; In one embodiment, a fusion network (e.g., a feedforward neural network (FFN)) is used to perform nonlinear transformations (ReLU activation) on attention features. Alternatively, a spatial adjacency graph is constructed for cultural units with spatial relationships (e.g., building plots and street blocks), and a graph neural network (GNN) is applied to propagate and fuse features within this adjacency structure. During the inference process, the fused high-order feature vectors are assigned to each cultural spatial unit, including building plots, street blocks, and street districts, giving each spatial unit a deep representation of historical and cultural characteristics. Based on the inferred and assigned spatial unit features, regional groups within the space belonging to the same cultural evolution chain are identified, revealing the natural evolutionary sequence of historical and cultural space. Furthermore, based on feature similarity, spatial units are clustered (by mapping a preset spatial function label table based on cluster center values) into several distinctive spatial landscape areas (e.g., traditional commercial blocks, modern industrial heritage sites, etc.).

[0039] Step S16: performing deep feature extraction on the historical and cultural space data to obtain deep feature data; In one embodiment, a multi-layer neural network (such as a three-layer MLP or ResNet Block) is designed to further extract information from the inferred data. The extracted information includes cultural temporal evolution characteristics (e.g., modern and classical styles); structural complexity (e.g., densely built-up areas versus open blocks); and architectural features (e.g., color, texture, and sense of scale). After processing, feature dimensionality reduction (e.g., using PCA or t-SNE) is performed to retain at least 95% of the information. After dimensionality reduction, a highly discriminative deep feature representation is output, with each feature vector accompanied by a label identifying the cultural evolution category, architectural feature category, or structural feature category.

[0040] Step S17: label the historical and cultural space data with cultural attributes based on the depth feature data to obtain historical and cultural element data.

[0041] In one embodiment, cultural attribute labels are assigned based on a preset dictionary or expert knowledge base, such as type labels (residential / temple / commercial); style labels (Class B architectural style, Class C architectural style); preservation status (intact, partially damaged, abandoned); and protection level (national, provincial, municipal, or generally registered). Based on deep feature vectors, a classifier (such as XGBoost, LightGBM, or a Transformer classification head) automatically determines the identity of the building and generates a confidence score: each label is assigned a confidence level (e.g., 90% confidence indicates "Class A style building"). If the confidence levels of multiple attributes are close (difference less than 5%), an expert system review decision is made (determined by preset knowledge rules or a manual review system) or a weighted attribute combination is performed (based on the confidence ratio, assigning a composite label or prioritizing the highest confidence category).

[0042] Preferably, the fusion reasoning is specifically: Perform cross-modal correlation processing on the feature attention data to obtain cross-modal correlation data; In one embodiment, the correlation between features of different modalities (text, image, and space) is analyzed to establish cross-modal connections. The attention features of each modality are mapped to the same dimension (e.g., unified to 512 dimensions) using a linear transformation or a trainable projection matrix. Cross-modal similarity is calculated. In the unified feature space, the system calculates the cosine similarity between text and image, text and space, and image and space: , for The feature similarity between is the first modal eigenvector, is the second modal feature vector, the first modal feature vector and the second modal feature vector are one of text-image, text-space, and image-space. Construct the similarity matrix between the three groups of modalities, each matrix size is (n is the number of data points). For feature pairs with high similarity scores (based on a threshold, such as a value greater than 0.5), their feature weights are increased; for feature pairs with low similarity scores, their feature weights are decreased. Through dynamic weighted optimization, the correlations between features from different modalities are highlighted, thereby generating cross-modal correlation data that reflects cross-modal connections.

[0043] Perform modal consistency optimization on cross-modal correlation data to obtain characteristic modal optimization data; In one embodiment, the system minimizes the statistical differences between the distributions of different modal features based on the maximum mean difference method. The average of all sample feature vectors in the first modal feature set is processed by the kernel mapping function, and the average of all sample feature vectors in the second modal feature set is processed by the kernel mapping function, and the square of the Euclidean norm of the difference between the two mean vectors is calculated. Perform local consistency regularization. In spatially adjacent units, the system requires that the consistency of feature vectors of different modalities in the local range meets the preset standard, that is, the Euclidean distance between the cross-modal feature vectors corresponding to any adjacent units is calculated, and the distance is controlled to be lower than the set threshold (for example, less than 0.1). The system performs weight adjustment based on the importance index of each modality. Specifically, the system adopts a weight mechanism driven by modal confidence to set a fusion weight for each modality, and the weight changes dynamically according to the credibility of the modality. For example, if the credibility of the text modality is , image modality credibility , the weight is mapped through the preset weight mapping table, then the weight is set to , , and the rest are assigned to other modalities.

[0044] Perform semantic completion reasoning on the feature modality optimization data to obtain feature modality completion data; In one embodiment, missing semantic regions are detected. The system compares and analyzes the feature vectors of each feature unit to detect units that are missing key semantic dimensions (e.g., chronological information, usage category labels). Feature missingness is determined by the absence of valid values for the relevant dimension or a confidence score below a set threshold. Multiple inference completion methods are implemented, including those based on nearest neighbor reasoning. The system employs a k-nearest neighbor (KNN) strategy to retrieve the k nearest neighbor samples (e.g., k is set to 5) with the closest semantic features for a feature unit with missing semantics. Based on the mean or weighted average of the features of the retrieved neighbor samples, missing attribute information is inferred and supplemented, thereby completing feature completion. For textual data, the system utilizes a domain-specific fine-tuned deep language model (e.g., BERT or T5) to predict missing semantic components based on the available semantic context, automatically inferring historical usage categories or descriptive details. Methods based on graph structure inference are also implemented. The system constructs a graph structure based on a spatial semantic adjacency graph, where nodes represent spatial or object units and edges represent semantic similarity or spatial proximity. Through label propagation algorithms or graph neural network (GNN) methods, missing labels or attributes are inferred in the graph structure to achieve semantic completion across nodes.

[0045] The spatial structure consistency of the characteristic modal completion data is fused to obtain historical and cultural spatial data.

[0046] In one embodiment, the system detects the consistency relationship between spatial units (such as blocks and plots) in terms of cultural type labels, age category labels, etc. If it is detected that there is an obvious abnormal break in the above labels between adjacent units, for example, a unit labeled as type E architectural style appears in a group of buildings with type B architectural style, the system performs spatial smoothing processing, corrects the abnormal labels through neighborhood reclassification or local label correction, and restores the overall spatial continuity. Performs spatial attribute weighted fusion processing. The system considers the following three factors: spatial proximity score (denoted as ), indicating the unit The degree of spatial proximity to surrounding units; modal similarity score (denoted as ), indicating the unit The similarity in multimodal feature expression; historical relevance score (denoted as ), indicating the unit The strength of the association with the surrounding historical and cultural elements. The system performs weighted fusion on the above score items according to the set weights and calculates the fusion score. Defined as a unit The spatial proximity score of is multiplied by a weight coefficient of 0.4, plus the modal similarity score multiplied by a weight coefficient of 0.4, plus the historical relevance score multiplied by a weight coefficient of 0.2. The attribute category with the highest score is assigned to the unit As its fusion label. Based on the fused spatial label, the system performs a connected region extraction operation. This operation extracts a set of spatially connected units with the same fusion label, forming a continuous historical and cultural area, such as a continuously distributed area of Class B architectural style blocks. The extracted connected regions are marked as spatial regeneration units.

[0047] Preferably, step S2 is specifically: Step S21: constructing element nodes and spatial relationship edges for the historical and cultural element data to obtain element node data and spatial relationship edge data respectively; In one embodiment, element nodes are constructed, such as treating each historical and cultural element (e.g., a single building, a block unit, or a road node) as a node. Node basic attributes include an ID (unique identifier, such as "Node_001"); category (e.g., traditional architecture, residential building, or public facility); era; spatial coordinates (center of mass or surface outline); and preservation status (intact, partially damaged, or renovated). Spatial relationship edges are constructed, such as defining an edge as follows: if the distance between the centers of mass of two nodes is less than a certain threshold (e.g., 150 meters), an adjacent edge is established; or if the two nodes belong to the same plot / block, a co-domain edge is established. Specifically, if there is a documented functional association (e.g., "temple and market"), a semantically enhanced edge is established. Edge attributes are defined as distance (Euclidean distance); association type (proximity, functional association, or chronological continuity); and association weight (based on distance and functional relevance scores, calculated based on a preset weight, or threshold mapping based on a preset rule engine; for example, proximity + functional relevance = high weight). Element node data and spatial relationship edge data are output.

[0048] Step S22: Generate a spatial semantic graph based on the element node data and the spatial relationship edge data to obtain spatial semantic graph data; In one embodiment, standard graph data structures (such as adjacency lists or adjacency matrices) are used to organize nodes and edges. Nodes are connected by spatial relationship edges to form a complete spatial semantic graph. Each node contains a dictionary of attributes (such as category, age, and protection level); each edge contains weight, distance, and association descriptions. Multiple node connections are allowed (i.e., a node can have relationships with multiple nodes). Within the adjacency list data structure, the spatial semantic graph is organized as key-value pairs, where the key is the unique identifier of the element node; the value is a list of target nodes to which the node is connected. Each connection record contains the following fields: target node identifier; internode distance; and relationship type description.

[0049] In one embodiment, the system can group input data into time series based on the time parameters (such as construction year, usage period, etc.) carried in the element node data and spatial relationship edge data, and construct spatial semantic graph data for different periods. Alternatively, the system can first construct a complete spatial semantic graph containing nodes and edges from all historical stages, and then perform subgraph extraction based on the time attributes of the edges or nodes in subsequent processing. Specifically, by setting a time condition (such as the age field of the node or the time label of the edge), a subgraph within a specified time period can be extracted from the overall graph and used as the data source for the first or second graph.

[0050] Step S23: extracting the chronological evolution trajectory based on the spatial semantic atlas data to obtain chronological evolution trajectory data; In one embodiment, an evolutionary trajectory refers to a path that begins at one historical node and evolves to another historical node through changes in spatial or functional relationships. Nodes must meet the requirement of chronological progression, meaning the starting node is older than the ending node. Nodes must maintain spatial or functional connectivity, meaning valid spatial or functional edge connections exist between them. Evolutionary trajectory extraction is based on a traversal operation of spatial semantic graph data, employing either a depth-first search (DFS) or breadth-first search (BFS) algorithm. Trajectory extraction rules include selecting an older node as the starting node and a younger node as the ending node. All nodes on the path between the starting and ending nodes must maintain spatial or functional connectivity, and the association weight of the edges connecting the nodes must be below a set threshold (e.g., an edge weight less than 0.5). Each extracted evolutionary trajectory is annotated with the following features: the starting node identifier; the ending node identifier; the sequence of nodes passed through; the trajectory length (number of steps); and the chronological span (the time difference between the starting and ending eras). The following constraints are set during the trajectory extraction process: the number of trajectory steps is generally controlled between 3 and 10 steps to avoid diluting the evolutionary significance due to excessively long trajectories; trajectories with obvious time spans are extracted, such as the evolution chain from architectural style B to architectural style E; for trajectories with shorter spans (such as architectural style B3 to architectural style D1), trajectories with shorter links and clearer evolutionary relationships are preferentially extracted.

[0051] Step S24: performing trajectory clustering on the chronological evolution trajectory data to obtain trajectory cluster data; In one embodiment, features of each trajectory are extracted, including starting point type; end point type; time span; trajectory length; and trajectory path morphology (e.g., presence of loops, jumps, etc.). Trajectory similarity is calculated (e.g., edit distance, dynamic time warping (DTW)), or similarity is measured using Euclidean distance based on trajectory feature vectors. K-Means, DBSCAN, or spectral clustering is used, with the number of clusters set based on task requirements (e.g., k = 5 clusters for a typical cultural evolution pattern). Each trajectory is assigned a category label, such as "traditional architecture evolution," "residential transformation," or "commercial regeneration," based on its cluster group (data based on group centers is mapped using a preset label library). Trajectory clustering data is output, with each trajectory having a cluster category.

[0052] Step S25: Perform cultural space classification on the chronological evolution trajectory data according to the trajectory clustering data to obtain historical and cultural space classification data.

[0053] In one embodiment, the spatial classification rules are defined such as classifying the spatial units covered by the starting node, the end node and all the passing nodes of the trajectory into the same cultural space classification according to the clustering category of each trajectory; each clustering category corresponds to a cultural evolution theme, such as a traditional architectural style area, a residential evolution area or a commercial prosperity area. During the classification process, a spatial continuity check is performed to ensure that the spatial connectivity rate between nodes in the same classification unit is not less than 80%. The spatial connectivity rate is defined as the ratio of the number of node pairs that are directly or indirectly spatially connected between any two nodes within the classification unit to the number of all possible node pairs. If the spatial connectivity rate is detected to be lower than 80%, local reclassification or node supplementation operations are performed based on the actual connection conditions of the nodes. For each cultural space classification unit, the following cultural attribute information is integrated and labeled, such as summarizing the dominant cultural category to which the spatial elements within the unit belong (the area ratio is calculated based on the preset electronic map, and the labels of the spatial elements whose proportion exceeds the threshold are used as the dominant cultural category), such as traditional buildings, traditional dwellings, commercial facilities, etc.; extracting the dominant period of the node age data within the unit as the time marker of the unit; based on the trajectory evolution characteristics, describing the main change paths of the internal spatial evolution of the unit, such as the evolution from traditional buildings to dwellings, or the transformation from traditional commercial areas to modern commercial complexes.

[0054] Preferably, the spatial semantic graph data includes first spatial semantic graph data and second spatial semantic graph data, the time parameter corresponding to the first spatial semantic graph data is not greater than the time parameter corresponding to the second spatial semantic graph data, and the chronological evolution trajectory extraction is specifically as follows: Performing time-constrained node association pairing on the first spatial semantic graph data and the second spatial semantic graph data to obtain time-constrained node pair data; In one embodiment, during the node pairing process, the following pairing constraints are satisfied at the same time, such as the Euclidean distance between the centroids of the nodes in the first spatial semantic graph and the nodes in the second spatial semantic graph does not exceed a preset threshold (e.g., 100 meters), or the spatial overlap rate between the two (i.e., the intersection-over-union ratio IOU) reaches or exceeds a preset threshold (e.g., 0.3). The category evolution relationship of the paired nodes should conform to the preset evolution logic (judged by an expert engine built based on expert knowledge). For example, "residential residences" can evolve into "shops," and "official offices" can evolve into "modern office buildings." If the category evolution logic is weak or not encouraged (e.g., "temples" evolve into "factories"), a low confidence mark should be added to the pairing results. The time attribute of the node in the first spatial semantic graph must be earlier than or equal to the time attribute of the corresponding node in the second spatial semantic graph. For each node in the first spatial semantic graph, retrieve a set of candidate nodes in the second spatial semantic graph that meet all the above pairing constraints; Determine the node pair evolution relationship of the time-constrained node pair data to obtain the node evolution data; In one embodiment, the system performs node pair evolution relationship determination on time-constrained node pair data to generate node evolution data. Specifically, based on changes in node attributes, the system determines the type of evolution relationship between each pair of nodes and records the corresponding change indicators. Evolution relationship types include functional evolution, where a node's functional category changes and the category change conforms to the permitted evolution path defined in a preset evolution knowledge base mapping table, such as evolving from "residential" to "commercial shop." Morphological evolution, where a node's physical morphological characteristics (such as area, perimeter, and compactness) undergo significant changes, such as a significant increase in building area or an increase in structural complexity. Functional and morphological dual evolution, where a node undergoes both functional category and physical morphological changes. No significant evolution, where a node undergoes no significant changes in either functional category or physical morphological characteristics. Specific determination rules include: For functional evolution determination, if a node category changes and the category change path conforms to the permitted evolution relationship defined in the evolution knowledge base mapping table, then it is determined to be functional evolution. For morphological evolution determination, the system compares the morphological indicators of the node pair, such as area, perimeter, and compactness, and calculates the change rate, which is defined as (the index value of the subsequent node - the index value of the previous node) divided by the index value of the previous node. For example, if the area change rate exceeds a set threshold (e.g., 30%), it is considered morphological evolution. Perimeter and compactness can also be evaluated for change significance using similar methods. Comprehensive evolution assessment is performed. If a node pair meets the above change criteria in both functional category and morphological characteristics, it is considered to have dual functional and morphological evolution. If a node pair does not meet the change assessment threshold in either functional category or morphological characteristics, it is marked as having no significant evolution.

[0055] The spatial connectivity is concatenated according to the node evolution data to obtain the chronological evolution trajectory data.

[0056] In one embodiment, the geographic spatial regions corresponding to multiple evolutionary nodes can form a continuous path or connected area, and the connections between nodes meet predefined spatial proximity and evolutionary consistency conditions. In a specific concatenation method, the system constructs a node evolution graph. Evolved nodes serve as node elements in the graph. If the Euclidean distance between the centers of mass of any two nodes is less than or equal to a predefined threshold (e.g., 100 meters), an edge is established between the two nodes. The edge weight is determined based on a combination of the spatial distance between the nodes and their evolutionary similarity. Evolutionary similarity can be quantified based on the consistency of changes in node functional categories. The system extracts chronological evolutionary trajectories from the node evolution graph according to the following trajectory extraction rules: Nodes with high cultural weight or preservation value are selected from the first spatial semantic graph (Graph 1) as trajectory starting points. Cultural weight can be determined based on a comprehensive score of node area, age, and historical and cultural level. The node corresponding to the starting point in the second spatial semantic graph (Graph 2) is selected as the trajectory endpoint. A shortest path algorithm (e.g., Dijkstra's algorithm) is used to search for connected paths within the node evolution graph. The paths must satisfy evolutionary direction consistency, meaning that the time parameters of the nodes in the path show an increasing trend. The Euclidean distance between the centroids of any two consecutive nodes on the path must not exceed a set maximum jump threshold (e.g., 100 meters). Trajectory validity checks are performed, such as ensuring that the path length is at least a set number of steps (e.g., at least 2 steps) and that the temporal and spatial spans fall within reasonable ranges (e.g., at least 30 years and at least 500 meters²). Chronological trajectory data is output.

[0057] Preferably, the trajectory clustering is specifically: Perform spatial node extraction and cultural semantic node extraction on the chronological evolution trajectory data to obtain spatial node data and cultural semantic node data; In one embodiment, spatial node extraction involves traversing each chronological trajectory and extracting the spatial location attributes of each node in the trajectory (center of mass coordinates (x, y), area, outline shape, etc.). Furthermore, cultural semantic attributes of each node (category, such as residential / temple / shop; chronological label; functional characteristics) are extracted. Standardized cultural labels (using a unified classification system, such as a standard code table for architectural categories) are then generated, along with cultural node data. The resulting output is a list of spatial node data and a list of cultural semantic node data.

[0058] Pairing spatial node data and cultural semantic node data to obtain node pair data; In one embodiment, based on the matching between spatial nodes and cultural semantic nodes, such as when the spatial distance between the two nodes is less than a set threshold (e.g., 100 meters), or when categories are similar or there is an evolutionary relationship (e.g., residential buildings to residential buildings, traditional architecture to cultural spaces), an expert knowledge engine based on expert knowledge is used to determine the matching. For each node in the trajectory, the next node that meets these conditions is searched to form a node pair. The spatial distance, category change type, and time span of the node pair are recorded.

[0059] Generate evolution trajectory of data according to nodes to obtain evolution trajectory data; In one embodiment, the evolution trajectory is constructed by concatenating node pairs in chronological order to form a directed trajectory path. The trajectory format is a node sequence (starting point → intermediate node → end point). The chronological order must be unidirectionally increasing; spatial jumps are controlled (the distance between consecutive nodes must not exceed a set threshold, such as 150 meters). Trajectory features recorded include total number of steps, spatial span (Euclidean distance from start to end point), chronological span, and evolution type chain (e.g., culture → commerce). A preliminary evolution trajectory data list is output.

[0060] Perform local trajectory repair on the evolving trajectory data to obtain trajectory repair data; In one embodiment, local repair of trajectories is mainly aimed at correcting interruptions or mutations that occur in the evolution trajectory. Trajectory anomaly scenarios include interruptions, which refer to missing nodes in the trajectory path, resulting in a break in the trajectory chain and the inability to form a complete evolution sequence; mutations, which refer to excessive changes in node categories in the trajectory path and an unreasonable evolution process, such as the discontinuity of jumping directly from traditional functions to industrial functions. In response to the above anomalies, the system adopts the following repair strategies: at the location where a broken link appears in the trajectory, the system searches for nodes that can be supplemented in adjacent time segments and spatial ranges; it adopts a K-nearest neighbor (KNN) method, combined with a temporal consistency screening criterion, to give priority to nodes that are similar in cultural category to the previous and next nodes as supplementary nodes; spatial distance and chronological proximity are used as node priority retrieval indicators, and category similarity is used as the main weighting factor for node completion selection. When the categories of continuous nodes in the trajectory change abnormally (such as evolving directly from traditional buildings to industrial plants), the system attempts to insert reasonable intermediate transition category nodes between the two; for example, by inserting nodes with cultural functional attributes, the evolution chain is corrected to "traditional → cultural → commercial" to restore the rationality and coherence of the trajectory evolution; the screening of inserted nodes is also based on spatial proximity, temporal continuity and cultural category matching evaluation.

[0061] The spatial path similarity matrix and the time span similarity matrix are constructed for the trajectory repair data to obtain the spatial similarity matrix data and the time span similarity matrix data respectively; In one embodiment, during the construction of the spatial path similarity matrix, the system calculates the spatial path similarity by comparing the spatial distribution similarity of two trajectories. Trajectory Hausdorff distance or dynamic time warping (DTW) is used; smaller trajectory distances indicate higher spatial path similarity; normalized to [0, 1], similarity = 1 indicates perfect match. The time span similarity matrix compares the time span differences between two trajectories. Time span similarity is defined as: , For the trajectory and trajectory The time span similarity between For the trajectory The span of years, For the trajectory The span of years, is the maximum value of the two trajectory spans and is also normalized to [0,1], where 1 indicates that the time span is completely consistent. Output spatial similarity matrix; time span similarity matrix.

[0062] The trajectory repair data is hierarchically clustered according to the spatial similarity matrix data and the time span similarity matrix data to obtain trajectory clustering data.

[0063] In one embodiment, spatial similarity and temporal similarity are combined to define similarity: ,in For the trajectory and trajectory The fusion similarity between is the spatial similarity weight coefficient, For the trajectory and trajectory The similarity in spatial paths, is the time similarity weight coefficient, For the trajectory and trajectory Similarity over time. Initially, each trajectory is considered an independent category. During each merging round, the system selects the pair of trajectories or trajectory clusters with the highest similarity and performs the merging operation. The clustering process continues until the preset clustering stop conditions are met, including allowing merging only when the combined similarity between the trajectory groups to be merged is at least a set threshold (for example, 0.8). The number of trajectories within each cluster group must not exceed a set maximum limit (for example, 20 trajectories / category) to prevent excessive clustering and excessive heterogeneity within the group. After these steps, the system outputs trajectory clustering data.

[0064] Preferably, step S3 is specifically: Step S31: extracting cultural evolution features based on the historical and cultural space classification data to obtain cultural evolution feature data, wherein the cultural evolution feature data includes spatial function evolution feature data, architectural form change feature data, cultural symbol evolution feature data, and time evolution trajectory feature data; In one embodiment, spatial function evolution feature data is extracted to extract the function type corresponding to each trajectory node to form a spatial function evolution sequence (such as: "residential" → "shop" → "office building"). To encode the function change chain, a sparse coding vector can be constructed for each function category, or a function category embedding method can be used to map the function category to a low-dimensional dense vector (such as a 64-dimensional embedding representation). Architectural form change feature data is extracted to calculate the following change rate indicators for the architectural form characteristics between trajectory nodes, including area change rate, outline compactness change rate (the later outline compactness minus the earlier outline compactness, divided by the earlier outline compactness) and height change rate. Cultural symbol evolution feature data is extracted to extract the change process of architectural style labels over time (such as "Class B style building" → "Class C style building"). Symbol change features are calculated and defined as follows: ,in For the characteristic data of cultural symbol evolution, Style similarity can be inferred through pre-trained models (such as the CLIP image-text matching model), and style retention (such as the degree of element detail retention, based on the ratio of later elements to earlier elements). Temporal trajectory feature data extraction involves extracting the following temporal features from the trajectory evolution process: decade span, such as the latest node age minus the earliest node age; temporal continuity indicators, such as the average age interval between adjacent nodes in the trajectory; and node density change trends, such as the trend in the number of nodes per unit age. Simple linear regression can be used to fit the slope of node density change.

[0065] Step S32: Using a preset historical and cultural space evolution generation model to generate multi-path natural evolution samples for the cultural evolution feature data to obtain candidate spatial regeneration data, wherein the preset historical and cultural space evolution generation model is obtained by generative adversarial training using pre-stored historical and cultural space classification experience data and spatial regeneration label data; In one embodiment, the historical and cultural spatial evolution generative model utilizes a generative adversarial network (GAN) structure or its improved versions (e.g., conditional GAN, diffusion-based generator). The training data consists of locally stored empirical historical and cultural spatial classification data (real historical evolution samples) and spatial regeneration label data (labels of actual renovation cases, such as preservation, renewal, and replacement). The cultural evolution feature vector serves as the input to the generator (G). Generator G receives the evolution features and outputs candidate spatial regeneration schemes (e.g., new functional categories, new architectural forms, and new cultural symbol states). A discriminator (D) determines whether the output conforms to the rationality of historical evolution and empirical rules of spatial regeneration (based on empirical learning from the training set). Random noise is introduced to generate natural evolution samples along different paths each time. Each input evolution feature generates N candidate samples (e.g., N = 5 different evolution paths). The loss function includes adversarial loss, feature matching loss, and path coherence loss. Through the above generation process, each cultural evolution feature input outputs N natural evolution paths, forming a candidate spatial regeneration dataset.

[0066] Step S33: Performing a spatial coherence check on the candidate spatial regeneration data to obtain spatial regeneration data.

[0067] In one embodiment, the generated spatial regeneration scheme ensures consistency and rationality in terms of geographic spatial distribution, cultural evolution logic, and morphological structure. Spatial contiguity between trajectory nodes is ensured (the Euclidean distance between center of mass is continuously less than a set threshold, such as 100 meters). Isolated nodes or large jumps are not permitted (the jump rate is controlled within 5%). The functional evolution chain is logically sound, without abrupt breaks (for example, if residential buildings jump directly to industrial plants, a commercial / office transition should be inserted). Changes in area, outline, and building elevation should be gradual (i.e., the rate of change is within a preset threshold). Thresholds are set, such as a single-step area growth rate of <50%. Architectural style transitions are reasonable (for example, a transition from a B-style to a C-style is acceptable, but a B-style to a D-style has low confidence and should be eliminated, based on a pre-set expert engine). Each candidate spatial regeneration trajectory is tested; samples that fail the test are discarded or repaired. Spatial regeneration data that meets all test criteria is output as the official spatial regeneration generation result.

[0068] Preferably, the step of constructing the preset historical and cultural space evolution generation model includes the following steps: Perform feature expansion layer processing on the historical and cultural space classification experience data stored locally to obtain feature expansion data; In one embodiment, the system performs feature expansion processing on locally stored historical and cultural space classification experience data to generate feature expansion data. The historical and cultural space classification experience data originates from a locally stored classification experience database, which records the functional category, time span, morphological change characteristics, and cultural symbol change attributes of each spatial unit. The feature expansion rules specifically include decomposing the architectural function change field recorded in the original classification data into two basic fields: the starting functional category and the ending functional category; decomposing the time evolution field into three basic fields: the starting year, the ending year, and the time span, where the time span is defined as the difference between the ending year and the starting year; decomposing the cultural symbol change field into two basic fields: the starting style and the ending style; and decomposing the spatial morphological change field into three basic fields: the area change rate, the height change rate, and the compactness change rate. The area change rate is defined as the ending area minus the starting area divided by the starting area, and the height change rate and the compactness change rate are calculated similarly. Continuous features are numerically normalized to the [0,1] interval. Structured and uniformly dimensional feature expansion data is output.

[0069] Performing deep feature transformation processing on the feature expansion data to obtain deep feature data; In one embodiment, the system constructs a deep feature extraction structure by stacking multiple layers of nonlinear mapping units, wherein the nonlinear mapping unit adopts a multi-layer perceptron (MLP) structure. Each mapping layer is followed by an activation function for nonlinear transformation, and the activation function is selected to be either a rectified linear unit (ReLU) or a Gaussian error linear unit (GELU). In terms of setting the number of layers, the system preferably sets the deep feature transformation network to 3 to 5 layers. The feature dimensions are gradually compressed between layers, for example, from the initial 1024 dimensions to 512 dimensions, and then to 256 dimensions. A dropout mechanism is introduced after the output of each hidden layer. The dropout operation randomly discards some neuron outputs with a zero ratio of 0.2.

[0070] Generating structured features on the deep feature data to obtain structured feature data; In one embodiment, the system rearranges the one-dimensional deep feature vector corresponding to each sample into a two-dimensional matrix structure. For example, if the original deep feature vector is 256 in length, it is rearranged into a two-dimensional feature map with 16 rows and 16 columns, thereby forming a spatial distribution of 16×16 feature units. During the rearrangement process, the system preferably maintains local correlation between feature units, that is, the feature content represented by adjacent units maintains semantic continuity. For example, this can correspond to different attribute changes during the evolution of the same building, such as area expansion, functional migration, and cultural symbolism.

[0071] Performing feature upsampling processing according to the structured feature data to obtain feature upsampling data; In one embodiment, the system uses an upsampling operation to increase the size of the feature map. Preferably, a deconvolution operation or sub-pixel convolution technology is used to achieve feature amplification. The deconvolution operation expands the size of the feature map by inverse modeling the standard convolution operation; the sub-pixel convolution is performed by rearranging the sub-pixel structure. The amplification ratio is set to 2 times or 4 times. For example, when the input feature map size is 16×16, it can be increased to a spatial size of 32×32 or 64×64 after upsampling. During the upsampling process, the system pays special attention to avoiding the checkerboard effect caused by the upsampling operation, that is, the local artifact phenomenon caused by uneven upsampling. To this end, the system introduces a smoothing convolution layer as a post-processing operation after upsampling, and further smoothes the feature map through a small-step standard convolution to eliminate the structural noise that occurs.

[0072] Perform convolution processing on the feature upsampled data to obtain candidate space regeneration sample data; In one embodiment, the system uses a multi-layer stacked two-dimensional convolution (Conv2D) structure to extract and reconstruct features based on the input feature upsampling data. Each convolution layer uses a convolution kernel of size 3×3, the stride is set to 1, and the boundary padding (Padding) is set to 1. The convolution network is preferably stacked in 3 to 5 layers to gradually extract high-level spatial features. After each convolution operation, the system introduces batch normalization processing and then applies the rectified linear unit (ReLU) activation function. At the output of the convolution network, the system generates multiple feature maps. Each feature map corresponds to a candidate space regeneration state, specifically including but not limited to candidate space layout features, functional configuration features or cultural symbol distribution features. The feature maps together constitute a candidate sample set for the rationality of preliminary spatial evolution.

[0073] Performing convolutional coding block processing on the candidate spatially regenerated sample data and the spatially regenerated label data to obtain convolutional coding block data; In one embodiment, the system uses a convolutional encoder structure with shared weights to encode candidate sample data and true label data separately. The so-called shared weights refer to the fact that the same set of convolutional encoder parameters are applied to the encoding process of both sample data and label data, ensuring that the two types of data are expressed in the same feature space. The structure of the convolutional encoder includes the following processing units connected in series, performing a two-dimensional convolution (Conv2D) operation with a convolution kernel size of 3×3, a stride of 1, and a padding of 1; applying batch normalization to normalize the feature distribution; then applying the rectified linear unit (ReLU) activation function; performing another layer of two-dimensional convolution operation with the same convolution kernel size of 3×3, a stride of 1, and a padding of 1; performing a flattening operation to flatten the two-dimensional feature map into a one-dimensional feature vector, and unifying the length of the encoded output feature vector, for example, to 512 dimensions.

[0074] Performing feature compression layer processing on the convolutional coding block data to obtain feature compression layer data; Performing feature compression layer processing on the convolutional coding block data to obtain feature compression layer data; In one embodiment, the system incorporates one or more fully connected layers as feature compression units. The input convolutional encoded feature vector (e.g., 512-dimensional) undergoes a linear transformation and is compressed to a lower dimension, such as 128 or 64. After each fully connected layer output, the system applies a nonlinear activation function, preferably a rectified linear unit (ReLU) or a Gaussian error linear unit (GELU). The system may incorporate a dropout mechanism after the fully connected layer. Dropout works by randomly setting some neuron outputs to zero at a preset rate during training.

[0075] Perform authenticity scoring layer processing based on the feature compression layer data to obtain sample authenticity scoring data; In one embodiment, the system introduces a single-node fully connected output layer to generate the final authenticity score. The output layer uses a Sigmoid activation function for nonlinear processing. This function compresses input features to the closed interval [0, 1], ensuring that the output score has standardized probabilistic interpretability. In the authenticity score results, a score close to 1 indicates that the generated sample is judged to be highly similar to a real sample; a score close to 0 indicates that the generated sample is judged to be a forged or distorted sample. The system uses this score to quantitatively evaluate the authenticity of the candidate sample.

[0076] The candidate space regeneration sample data is iteratively trained according to the sample authenticity score data to obtain the historical and cultural space evolution generation model.

[0077] In one embodiment, a dual-model architecture is established, where the generator (G) aims to generate spatially reconstructed samples that are closer to reality, aiming to maximize the sample authenticity score; the discriminator (D) aims to correctly distinguish between real and generated samples, aiming to accurately determine the authenticity of the input data. During training, the system uses the standard GAN loss function as the optimization objective. Specifically, the generator G attempts to minimize, while the discriminator D attempts to maximize, the following loss function: ,in To optimize the target area for adversarial purposes, we maximize the influence of the discriminator and minimize the influence of the generator based on game optimization. It is the expectation of the logarithmic prediction value when the true sample x is expected to obey the true data distribution, plus the expectation of the logarithmic anti-prediction value of the forged sample z generated by the generator G, is the expectation of the real sample loss term, To sample x from the true data distribution, is the discriminator’s predicted value for x, is the real data distribution, sample The true probability distribution of for Log-probability value, To generate the expectation of the sample loss term, is the “true” prediction probability of the discriminator for the generated sample. An alternating update strategy is adopted, that is, after every 5 training updates of the discriminator D, the generator G is updated once to maintain the dynamic balance of generation and discrimination training; the optimizer uses the Adam optimizer, an adaptive learning rate optimization algorithm based on first-order moment estimation and second-order moment estimation, and the initial value of the learning rate is set to During the training process, the mean authenticity score of the generated samples is monitored. When the mean reaches or exceeds the set threshold (for example, 0.85), the model is considered to have converged.

[0078] Preferably, step S4 is specifically: Step S41: Acquire urban planning demand data; In one embodiment, the acquired urban planning demand data includes the following contents, such as the latest urban planning text materials, such as land use planning, functional zoning, transportation system layout and public facilities layout; planning map data in vector format, such as land red line map, road system map; planning index requirements, such as green space ratio not less than 30%, building height limit within 80 meters, and commercial land ratio requirement between 20% and 40%. The acquired urban planning demand data is unified into a structured format, specifically including attribute data tables, such as CSV or JSON file formats; spatial data files, such as GeoJSON or Shapefile format files. Each planning unit must be accompanied by the following necessary attribute fields, such as land use type, such as residential land, commercial land, cultural and educational land, transportation land, green land, etc.; building height limit, which stipulates the maximum allowable height of a single building; volume ratio range, that is, the range of the ratio of building area allowed for development and construction to land area; public service facility supporting indicators, which record the configuration standards or ratio requirements of various types of public facilities.

[0079] Step S42: performing attribute mapping according to the urban planning demand data and the spatial regeneration data to obtain spatial planning mapping data; In one embodiment, the goal of the mapping process is to match the attributes of each regeneration space unit with the corresponding urban planning unit and to attach the relationship. During the attribute mapping process, the following rules are followed, such as spatial position matching, the center point or main range of the regeneration space unit must completely fall within a certain urban planning land area as the basic basis for spatial correspondence; attribute field matching, the main attribute fields of the corresponding planning unit, including land type, building height limit and volume ratio range, are assigned and attached to the regeneration unit; functional adaptation detection, recording the original functional attributes of the regeneration unit, and comparing them with the planning function requirements of the corresponding urban planning unit, marking the matching status label based on the comparison results, and dividing them into two categories: Matched and Conflict. When dealing with situations where the regeneration unit spans multiple urban planning units, the following detailed rules are adopted, including area ratio priority, with the planning unit with the largest coverage area of the regeneration unit as the main attribute source; secondary attribute recording, recording the secondary coverage area and its corresponding planning attributes at the same time as an auxiliary reference.

[0080] Step S43: performing conflict detection on the spatial planning mapping data to obtain spatial conflict detection data; In one embodiment, field overlap is calculated for urban planning demand data and spatial regeneration data, and data in the spatial regeneration data that does not exist in the urban planning demand data is marked to obtain spatial conflict detection data.

[0081] In one embodiment, the system maps urban planning data and regeneration data for each spatial region using structured fields based on a pre-defined electronic map template. A field comparison is performed for each mapped region. If there are inconsistencies between key attribute fields (based on field names, codes, or pre-defined semantic standards), a spatial conflict is identified. Comparison fields and conflict triggering conditions are configured and adjusted using a pre-defined attribute mapping rule table.

[0082] In one embodiment, after attribute mapping, the system marks areas that do not meet planning conditions based on preset judgment rules (such as functional compatibility standards, planning indicator thresholds, etc.). The marking results include conflict type, conflict field, and recommended processing method.

[0083] In one embodiment, there is a discrepancy between the current attributes of the regeneration space unit and the planned attributes of the corresponding urban planning unit, or the regeneration unit violates the requirements of urban planning indicators. The conflict detection content includes the following aspects, including functional conflict, which refers to the original function of the regeneration space unit being different from the functional classification preset by the planning unit, and the functional categories being incompatible. For example, the current use of the historical block is inconsistent with the planned industrial land use. The function field is mapped to [0,1] according to the degree of non-conformity / non-overlap; morphological conflict, which refers to the building height or volume ratio of the regeneration space unit exceeding the limit set by the corresponding urban planning unit. For example, the volume ratio between 2.2-2.5 is set as a minor conflict zone, and the system outputs the conflict membership μ∈[0,1]; layout conflict, which refers to the spatial layout of the regeneration space unit overlapping with the urban planning red line or road system layout, or violating the planning boundary regulations. The degree of overlap of the planning red line is output in a numerical form. Conflict detection follows the following rules: functional compatibility testing, based on a pre-set functional compatibility table; for example, commercial functions can be compatible with historic blocks, but industrial functions cannot; and planning indicator hard constraint testing, which sets hard constraints on building volume ratios, building heights, and public facilities configuration indicators. Any violation exceeding the specified upper limit is considered a conflict and mapped to a scale of [0, 1] based on the degree of non-compliance. The output includes a conflict detection list containing the following content: conflict type (functional conflict, morphological conflict, layout conflict); conflict severity (classified by conflict severity, such as minor conflict, moderate conflict, and severe conflict); and a list of involved spatial units and corresponding attribute information.

[0084] Step S44: collaboratively adjusting the spatial conflict detection data to obtain spatial adjustment data; In one embodiment, the system constructs a structured representation of all conflicting spatial units, integrating their attribute vectors (such as area, building density, and land use history) with spatial adjacency relationships. This is then encoded using a graph neural network or spatial autoencoder network to obtain a low-dimensional representation vector for the conflicting regions. Based on the vector representation of the conflicting regions, unsupervised clustering (such as DBSCAN or Spectral Clustering) is used to automatically identify conflict pattern categories. For example, cluster labels are implicitly categorized as high-density conflict, functional mismatch conflict, or spatial boundary compression conflict. (This is done by determining the ratio of cluster center values generated by clustering, or by thresholding based on pre-set expert knowledge. Alternatively, semantic labels such as high-density conflict, functional mismatch conflict, and spatial boundary compression conflict can be generated through an unsupervised combined analysis of inter-cluster distances, intra-cluster compactness, and attribute means.) The system constructs an optimization function based on spatial accessibility, node connectivity, and attribute distribution balance, or one of these specific metrics, to evaluate and rank candidate adjustment strategies for the conflicting regions. The optimization model uses an evolutionary search algorithm (such as a genetic algorithm or particle swarm optimization) to automatically generate the optimal adjustment strategy vector from the candidate strategy set. Based on the optimization results, the spatial form, functional distribution, and boundary relationships of the conflicting areas are restructured, including adjustments to building volume, functional layout relocation, and boundary coordination. The system records all adjustment operations as an operation log and automatically generates a spatial adjustment dataset, including the original state and adjusted state of each modified area, the adjustment vector parameters, and the impact range.

[0085] Step S45: performing spatial layout fusion on the spatially adjusted data to obtain spatially regenerated fused data.

[0086] In one embodiment, the system constructs a spatial graph model based on the adjusted spatial unit data, where nodes represent individual spatial units and edges represent spatial adjacencies or functional associations. Each node is accompanied by its attribute vector, including building scale, spatial type, connectivity weight, historical preservation identifier, etc., which are uniformly embedded into a vector space representation. Based on the spatial graph, the system calculates the accessibility indicators of each node pair, such as walking connectivity, interaction path length, etc. The shortest path search and graph structure diffusion algorithm are used to obtain the potential accessibility matrix of the global space. The system constructs an evolution function ( ,in For the spatial layout goal, Optimize the loss weight for local connectivity with a value of 0.2. Optimize loss for local connectivity, is the spatial function distribution balance loss weight, which is set to 0.2. The loss of spatial functional distribution balance, is the building density continuity loss weight, which is set to 0.5. is the loss of building density continuity, is the semantic clustering consistency loss weight, with a value of 0.1. The system automatically defines spatial layout objectives based on the semantic clustering consistency loss (i.e., semantic clustering consistency loss). The evolution function parameters include enhancing local connectivity (optimizing adjacency paths), spatial distribution balance (preventing functional polarization), building density continuity, and spatial semantic clustering consistency (derived from self-supervised semantic clustering results). The optimization function is automatically learned through historical spatial sample distribution, self-supervised graph learning, or latent variable modeling. Evolutionary search algorithms (such as genetic algorithms, simulated annealing, and graph embedding perturbation optimization) are used to iteratively update spatial node positions and connectivity. The system dynamically selects the optimal fused layout path based on minimizing the deviation from the objective function. During each round of evolution, nodes can undergo edge movement, region merging, and functional swapping to achieve self-organizing layout optimization. After fusion is complete, the system outputs a spatial fusion graph corresponding to the final layout solution, containing information such as the new spatial positions, functional labels, and connectivity structure of all nodes. The system also automatically generates change annotation data, documenting the structural changes, evolutionary process, and fusion path from the initial layout to the fused layout.

[0087] Preferably, the present application further provides a historical and culturally oriented space classification and regeneration system for executing the above-mentioned historical and culturally oriented space classification and regeneration method, the historical and culturally oriented space classification and regeneration system comprising: The historical and cultural data processing module is used to obtain historical and cultural data, perform multimodal perception on the historical and cultural data, and obtain historical and cultural spatial data; perform deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; The spatial semantic modeling and trajectory classification module is used to construct a spatial semantic map of historical and cultural element data to obtain spatial semantic map data; and to classify the chronological evolution trajectory of the spatial semantic map data to obtain historical and cultural spatial classification data; A space regeneration generation module is used to generate space regeneration based on historical and cultural space classification data to obtain space regeneration data; The spatial regeneration fusion module is used to obtain urban planning demand data, and perform spatial regeneration fusion based on the urban planning demand data and spatial regeneration data to obtain spatial regeneration fusion data.

[0088] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes that fall within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0089] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A historical and culturally oriented spatial classification and regeneration method, characterized by: The following steps are involved: Step S1: Acquire historical and cultural data, perform multimodal perception on the historical and cultural data, and obtain historical and cultural spatial data; perform deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; Step S2: constructing a spatial semantic map of the historical and cultural element data to obtain spatial semantic map data; classifying the spatial semantic map data by chronological evolution trajectory to obtain historical and cultural spatial classification data; Step S3: performing spatial regeneration based on the historical and cultural space classification data to obtain spatial regeneration data; Step S4: Acquire urban planning demand data, and perform spatial regeneration fusion according to the urban planning demand data and the spatial regeneration data to obtain spatial regeneration fusion data.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Access to historical and cultural data; Perform multimodal feature extraction on historical and cultural data to obtain multimodal feature data; Perform feature splicing on the multimodal feature data to obtain feature splicing data; Perform self-attention encoding on the feature splicing data to obtain feature attention data; Perform fusion reasoning on feature attention data to obtain historical and cultural space data; Perform deep feature extraction on historical and cultural spatial data to obtain deep feature data; The cultural attributes of historical and cultural space data are annotated based on the deep feature data to obtain historical and cultural element data.

3. The method according to claim 2, characterized in that The fusion reasoning is specifically as follows: Perform cross-modal correlation processing on the feature attention data to obtain cross-modal correlation data; Perform modal consistency optimization on cross-modal correlation data to obtain characteristic modal optimization data; Perform semantic completion reasoning on the feature modality optimization data to obtain feature modality completion data; The spatial structure consistency of the characteristic modal completion data is fused to obtain historical and cultural spatial data.

4. The method according to claim 1, wherein Step S2 is specifically as follows: Element nodes and spatial relationship edges are constructed for historical and cultural element data to obtain element node data and spatial relationship edge data respectively; Generate a spatial semantic graph based on element node data and spatial relationship edge data to obtain spatial semantic graph data; Extract the chronological evolution trajectory based on the spatial semantic atlas data to obtain the chronological evolution trajectory data; Perform trajectory clustering on the chronological evolution trajectory data to obtain trajectory cluster data; The cultural space classification of the chronological evolution trajectory data is carried out according to the trajectory clustering data to obtain the historical and cultural space classification data.

5. The method according to claim 4, characterized in that The spatial semantic map data includes first spatial semantic map data and second spatial semantic map data. The time parameter corresponding to the first spatial semantic map data is not greater than the time parameter corresponding to the second spatial semantic map data. The chronological evolution trajectory extraction is specifically as follows: Performing time-constrained node association pairing on the first spatial semantic graph data and the second spatial semantic graph data to obtain time-constrained node pair data; Determine the node pair evolution relationship of the time-constrained node pair data to obtain the node evolution data; The spatial connectivity is concatenated according to the node evolution data to obtain the chronological evolution trajectory data.

6. The method according to claim 4, characterized in that The trajectory clustering is specifically as follows: Perform spatial node extraction and cultural semantic node extraction on the chronological evolution trajectory data to obtain spatial node data and cultural semantic node data; Pairing spatial node data and cultural semantic node data to obtain node pair data; Generate evolution trajectory of data according to nodes to obtain evolution trajectory data; Perform local trajectory repair on the evolving trajectory data to obtain trajectory repair data; The spatial path similarity matrix and the time span similarity matrix are constructed for the trajectory repair data to obtain the spatial similarity matrix data and the time span similarity matrix data respectively; The trajectory repair data is hierarchically clustered according to the spatial similarity matrix data and the time span similarity matrix data to obtain trajectory clustering data.

7. The method according to claim 1, characterized in that Step S3 is specifically as follows: Extract cultural evolution characteristics based on historical and cultural space classification data to obtain cultural evolution characteristic data, where the cultural evolution characteristic data includes spatial function evolution characteristic data, architectural form change characteristic data, cultural symbol evolution characteristic data, and time evolution trajectory characteristic data; The preset historical and cultural space evolution generation model is used to generate multi-path natural evolution samples of cultural evolution feature data to obtain candidate spatial regeneration data. The preset historical and cultural space evolution generation model is obtained by generative adversarial training based on pre-stored historical and cultural space classification experience data and spatial regeneration label data. Perform spatial coherence test on the candidate spatial regeneration data to obtain spatial regeneration data.

8. The method according to claim 7, characterized in that The steps for constructing the preset historical and cultural space evolution generation model include the following steps: Perform feature expansion layer processing on the historical and cultural space classification experience data stored locally to obtain feature expansion data; Performing deep feature transformation processing on the feature expansion data to obtain deep feature data; Generating structured features on the deep feature data to obtain structured feature data; Performing feature upsampling processing according to the structured feature data to obtain feature upsampling data; Perform convolution processing on the feature upsampled data to obtain candidate space regeneration sample data; Performing convolutional coding block processing on the candidate spatially regenerated sample data and the spatially regenerated label data to obtain convolutional coding block data; Performing feature compression layer processing on the convolutional coding block data to obtain feature compression layer data; Perform authenticity scoring layer processing based on the feature compression layer data to obtain sample authenticity scoring data; The candidate space regeneration sample data is iteratively trained according to the sample authenticity score data to obtain the historical and cultural space evolution generation model.

9. The method according to claim 1, characterized in that Step S4 is specifically as follows: Obtain urban planning demand data; Perform attribute mapping based on urban planning demand data and spatial regeneration data to obtain spatial planning mapping data; Performing conflict detection on the spatial planning mapping data to obtain spatial conflict detection data; Coordinately adjust the spatial conflict detection data to obtain spatial adjustment data; The spatial layout fusion is performed on the spatial adjustment data to obtain the spatial regeneration fusion data.

10. A historical and culturally oriented space classification and regeneration system, characterized by: For executing the historical and cultural oriented space classification and regeneration method as claimed in claim 1, the historical and cultural oriented space classification and regeneration system comprises: The historical and cultural data processing module is used to obtain historical and cultural data, perform multimodal perception on the historical and cultural data, and obtain historical and cultural spatial data; perform deep feature extraction on the historical and cultural spatial data to obtain historical and cultural element data; The spatial semantic modeling and trajectory classification module is used to construct a spatial semantic map of historical and cultural element data to obtain spatial semantic map data; and to classify the chronological evolution trajectory of the spatial semantic map data to obtain historical and cultural spatial classification data; A space regeneration generation module is used to generate space regeneration based on historical and cultural space classification data to obtain space regeneration data; The spatial regeneration fusion module is used to obtain urban planning demand data, and perform spatial regeneration fusion based on the urban planning demand data and spatial regeneration data to obtain spatial regeneration fusion data.

Images