Urban building group three-dimensional function partitioning and floor population dynamic distribution method and device

Abstract

The invention provides an urban building group three-dimensional function partitioning and floor population dynamic distribution method and device, and relates to the technical field of urban space information processing. The method comprises the steps of constructing semantic feature vectors of multi-source POI data based on a BERT model and domain specific vocabularies to generate fusion data; building a floor position probability distribution model according to a three-layer progressive reasoning structure and a reasoning technology; carrying out three-dimensional function partitioning according to the function coupling strength and an improved spectral clustering algorithm; building-level demographic data are generated, and a differentiation attraction model is constructed; and solving the floor population distribution constraint optimization model according to the improved Haff model to obtain population dynamic distribution of each floor. According to the method, granularity limitation of traditional function division is broken through, floor-level refined three-dimensional function division and population dynamic distribution prediction are realized, and important decision support is provided for applications such as commercial site selection, emergency evacuation path planning, public service facility configuration optimization and the like in smart city construction.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of urban spatial information processing and artificial intelligence technology, in particular to a method and device for three-dimensional functional zoning of urban building groups and dynamic distribution of floor population, which belongs to the technical field of smart city, spatial big data analysis, machine learning and urban planning. BACKGROUND

[0002] Three-dimensional functional zoning of urban building groups and population distribution prediction are core technical problems in smart city construction, which are of great significance to urban management, emergency response, business planning and public service optimization. With the accelerated development of urbanization and the continuous improvement of building density, modern urban building groups show significant characteristics of functional integration, spatial three-dimensionality and dynamic population flow. Traditional urban spatial analysis methods are mainly based on two-dimensional plane data and static statistical information, which are difficult to accurately reflect the three-dimensional functional distribution characteristics within the building group, and even more difficult to achieve fine-grained dynamic population prediction at the floor level. This technical limitation seriously restricts the development level of urban fine-grained management, intelligent decision-making and personalized service. Therefore, the research on the method of three-dimensional functional zoning of urban building groups and dynamic distribution of floor population based on multi-source data fusion and artificial intelligence technology has important theoretical value and practical significance for improving the level of urban governance modernization and optimizing resource allocation efficiency.

[0003] In recent years, some progress has been made in the research on urban spatial function identification and population distribution prediction. In terms of urban function identification, scholars have conducted a large number of research work based on POI (Point of Interest) data, remote sensing images, mobile communication data and other multi-source information (Liu Y, Xiao Y, Gao S, et al. A review of human mobility based on location-aware devices[J]. Acta Geographica Sinica, 2011, 66(10): 1-12; Xue B, Xiao X, Li J, et al. Multi-source data-driven identification of urban functional areas: A case of Shenyang, China[J]. Chinese Geographical Science, 2023, 32: 1-18; Wang J, Gao C, Wang M, et al. Identification of urban functional areas and urban spatial structure analysis by fusing multi-source data features: A case study of Zhengzhou, China[J]. Sustainability, 2023, 15(8): 6505; Wang Y, Li C, Zhang H, et al. Research on multi-source data fusion urban functional area identification method based on random forest model[J]. Sustainability, 2025, 17(2): 515).

[0004] In terms of population distribution prediction, researchers have explored population spatial distribution modeling techniques based on machine learning, deep learning, and other methods (Zhao S, Liu Y, Zhang R, et al. China's population spatialization based on three machine learning models[J]. Journal of Cleaner Production, 2020, 256: 120910; Robinson C, Hohman F, Dilkina B. A deep learning approach for population estimation from satellite imagery[C]. Proceedings of the 1st ACM SIGCAS Conference on Computing and Sustainable Societies, 2017: 1-9; Zhang Y, Zhang Y, Huang B, et al. A hybrid model for high spatial and temporal resolution population distribution prediction[J]. International Journal of Digital Earth, 2022, 15(1): 2155718; Botella C, Joly A, Bonnet P, et al. A deep learning approach to species distribution modelling[M]. Multimedia tools and applications for environmental & biodiversity informatics. Springer, 2018: 169-199).

[0005] However, these studies mainly focus on functional recognition in two-dimensional planes and macro-scale population prediction, and lack analysis of three-dimensional functional zoning and floor-level population distribution characteristics within building complexes. In addition, existing methods rely on a single data source or simple data stacking, lacking deep semantic understanding and intelligent reasoning capabilities. For the problem of missing floor location information in POI data, as well as the spatial correlation and synergistic effects between different functions, existing research has not yet proposed effective solutions, making it difficult to meet the practical needs of urban fine management. SUMMARY

[0006] To solve the above technical problems existing in the prior art, the embodiments of the present application provide a method and device for three-dimensional functional zoning and dynamic floor population distribution of urban building groups. The technical solution is as follows:

[0007] On the one hand, a method for three-dimensional functional zoning and dynamic floor population distribution of urban building groups is provided, which is realized by a device for three-dimensional functional zoning and dynamic floor population distribution of urban building groups. The method comprises the following steps:

[0008] S1. Obtain multi-source POI data of the urban building group, construct a semantic feature vector of the multi-source POI data based on a pre-trained BERT model combined with field-specific vocabulary, perform multi-level similarity calculation on the semantic feature vector, and generate a fused POI data set according to the similarity calculation result.

[0009] S2. Construct a three-layer progressive reasoning structure based on a large language model, construct a floor position probability distribution model according to the fused POI data set, the three-layer progressive reasoning structure, and Chain-of-Thought reasoning technology, and obtain POI floor positions.

[0010] S3. Perform three-dimensional functional zoning on the interior of the urban building group according to the fused POI data set, a functional coupling strength index, and an improved spectral clustering algorithm, and obtain a functional zoning result of the interior of the urban building group; optimize the functional zoning result based on a simulated annealing algorithm, and obtain an optimized functional zoning result.

[0011] S4. Generate building-level population statistics data through a mobile phone signaling data processing algorithm, construct a differentiated attraction model according to the population statistics data and the optimized functional zoning result, and calculate the attraction weight coefficients of the population for each functional zoning.

[0012] S5. Establish a floor population distribution constraint optimization model, solve the floor population distribution constraint optimization model according to the attraction weight coefficients of the population for each floor functional zoning and an improved Huff model, and obtain the dynamic distribution of the population on each floor; establish a dynamic adjustment mechanism to adjust the dynamic distribution of the population on each floor in real time.

[0013] S6. Construct a building three-dimensional population distribution model according to the dynamic distribution of the population on each floor and the optimized functional zoning result, and generate a hierarchical population heat map and a function-population correlation graph of the urban building group.

[0014] Optionally, the step of constructing a semantic feature vector of the multi-source POI data based on a pre-trained BERT model combined with field-specific vocabulary, performing multi-level similarity calculation on the semantic feature vector, and generating a fused POI data set according to the similarity calculation result in S1 comprises:

[0015] S11, constructing a semantic feature vector of the multi-source POI data according to the multi-source POI data, a pre-trained BERT model and a domain-specific vocabulary.

[0016] S12, performing multi-level similarity calculation on the semantic feature vector according to semantic similarity, geographical similarity and attribute similarity to obtain a similarity calculation result.

[0017] S13, establishing a dynamic adjustment mechanism of a POI adaptive fusion threshold based on data source credibility and historical accuracy of the multi-source POI data.

[0018] wherein, the calculation formula of the POI adaptive fusion threshold is shown as formula (1) below:

[0019] (1)

[0020] in the formula, POI adaptive fusion threshold, base threshold, adjustment coefficient, data source quality evaluation index.

[0021] S14, performing multi-source POI data fusion according to the similarity calculation result and the POI adaptive fusion threshold to generate a fused POI data set.

[0022] Optionally, the constructing of the semantic feature vector of the multi-source POI data according to the multi-source POI data, the pre-trained BERT model and the domain-specific vocabulary in S11 comprises:

[0023] S111, establishing a POI domain-specific dictionary containing commercial terms, geographical terms and domain-specific vocabulary.

[0024] S112, performing further pre-training on the multi-source POI data by using the BERT model according to the POI domain-specific dictionary to obtain the pre-trained BERT model.

[0025] S113, calculating the semantic feature vector of the multi-source POI data according to the pre-trained BERT model and a multi-granularity feature fusion strategy.

[0026] Optionally, the constructing of the three-layer progressive reasoning structure based on the large language model in S2, the constructing of the floor position probability distribution model according to the fused POI data set, the three-layer progressive reasoning structure and the Chain-of-Thought reasoning technology to obtain the POI floor position comprises:

[0027] S21, a three-layer progressive reasoning structure based on a large language model is constructed; wherein the three-layer progressive reasoning structure comprises a basic information reasoning layer, an associated information reasoning layer, and a logical verification reasoning layer.

[0028] S22, explicit information analysis is performed on the fused POI data set by the basic information reasoning layer to generate a preliminary floor judgment; implicit information analysis is performed on the fused POI data set by the associated information reasoning layer to correct the floor judgment; logical consistency and real reasonableness of the reasoning result are verified by the logical verification reasoning layer, and a floor position probability is output.

[0029] S23, context information is analyzed by a Chain-of-Thought reasoning technology according to the preliminary floor judgment, the corrected floor judgment, and the floor position probability.

[0030] S24, a floor position probability distribution model is constructed based on the context information, and a floor position reasoning result is obtained according to the floor position probability distribution model.

[0031] S25, a confidence evaluation mechanism is established to score the credibility of the floor position reasoning result, and a POI floor position and a confidence are obtained.

[0032] Optionally, in S3, the interior of the urban building group is functionally zoned according to the fused POI data set, the functional coupling strength index, and the improved spectral clustering algorithm, and a functional zoning result of the interior of the urban building group is obtained; the functional zoning result is optimized based on a simulated annealing algorithm, and an optimized functional zoning result is obtained, including:

[0033] S31, a functional coupling strength index is calculated based on the spatial proximity, the business association, and the human flow interaction of the fused POI data set.

[0034] S32, a floor adjacency matrix is constructed based on the fused POI data set, and a Laplacian matrix is obtained according to the floor adjacency matrix.

[0035] S33, according to the functional coupling strength index and the Laplacian matrix, an improved spectral clustering algorithm is used to perform three-dimensional functional zoning of the interior of the urban building group, and a functional zoning result of the interior of the urban building group is obtained; wherein the improved spectral clustering algorithm introduces floor connectivity constraints and functional similarity weights.

[0036] S34, the partition boundary in the functional zoning result is optimized based on a simulated annealing algorithm, and an optimized functional zoning result is obtained.

[0037] Optionally, in S4, the building-level demographic data is generated by a mobile phone signaling data processing algorithm, a differentiated attraction model is constructed according to the demographic data and the optimized functional zoning result, and the attraction weight coefficient of each functional zoning to the population is calculated, including:

[0038] S41, obtain mobile phone signaling data, extract the number of mobile phone users within the boundary of each building in the urban building group by a mobile phone signaling data processing algorithm, and generate building-level demographic data according to the number of mobile phone users.

[0039] S42, construct a differentiated attraction model according to the demographic data and the optimized functional zoning result; wherein the differentiated attraction model includes multiple time dynamic models, each time dynamic model includes a basic attraction and a time attraction; the multiple time dynamic models include a commercial function attraction model, a residential function attraction model, an education function attraction model, a medical function attraction model, an office function attraction model, and a comprehensive function attraction model.

[0040] S43, calculate the attraction weight coefficient of each floor functional zoning to the population according to the differentiated attraction model.

[0041] Optionally, in S5, a floor population distribution constraint optimization model is established, the floor population distribution constraint optimization model is solved according to the attraction weight coefficient of each floor functional zoning to the population and the improved Haaf model, and the dynamic distribution of the population of each floor is obtained; a dynamic adjustment mechanism is established to adjust the dynamic distribution of the population of each floor in real time, including:

[0042] S51, establish a floor population distribution constraint optimization model.

[0043] S52, solve the floor population distribution constraint optimization model by using the improved Haaf model according to the attraction weight coefficient of each floor functional zoning to the population, and obtain the dynamic distribution of the population of each floor; wherein the improved Haaf model introduces floor capacity constraints and function matching degree constraints.

[0044] S53, establish a dynamic adjustment mechanism, and adjust the dynamic distribution of the population of each floor in real time based on the real-time acquired mobile phone signaling data and the dynamic adjustment mechanism; wherein the dynamic adjustment mechanism includes designing a sliding window data updating strategy and updating the attraction weight coefficient by using a weighted average method; the length of the sliding window is adaptively adjusted according to the data change frequency.

[0045] On the other hand, a device for three-dimensional functional zoning of urban building groups and dynamic distribution of floor population is provided, which is applied to the method for three-dimensional functional zoning of urban building groups and dynamic distribution of floor population, and the device includes:

[0046] The data fusion module is configured to obtain multi-source POI data of the urban building group, construct a semantic feature vector of the multi-source POI data based on a pre-trained BERT model and a field-specific vocabulary, perform multi-level similarity calculation on the semantic feature vector, and generate a fused POI data set according to a similarity calculation result.

[0047] The AI inference module is configured to construct a three-layer progressive inference structure based on a large language model, construct a floor position probability distribution model according to the fused POI data set, the three-layer progressive inference structure, and a Chain-of-Thought reasoning technique, and obtain a POI floor position.

[0048] The functional partition module is configured to perform three-dimensional functional partition on an interior of the urban building group according to the fused POI data set, a functional coupling strength index, and an improved spectral clustering algorithm, obtain a functional partition result of the interior of the urban building group, and optimize the functional partition result based on a simulated annealing algorithm to obtain an optimized functional partition result.

[0049] The attraction modeling module is configured to generate building-level population statistics by a mobile phone signaling data processing algorithm, construct a differential attraction model according to the population statistics and the optimized functional partition result, and calculate attraction weight coefficients of the population for each functional partition.

[0050] The population distribution optimization module is configured to establish a floor population distribution constraint optimization model, solve the floor population distribution constraint optimization model according to the attraction weight coefficients of the population for each floor functional partition and an improved Huff model, and obtain a dynamic population distribution of each floor; and establish a dynamic adjustment mechanism to adjust the dynamic population distribution of each floor in real time.

[0051] The visualization display module is configured to construct a building three-dimensional population distribution model according to the dynamic population distribution of each floor and the optimized functional partition result, and generate a hierarchical population heat map and a function-population correlation graph of the urban building group.

[0052] Optionally, the data fusion module is further configured to:

[0053] S11. Construct a semantic feature vector of the multi-source POI data according to the multi-source POI data, the pre-trained BERT model, and the field-specific vocabulary.

[0054] S12. Perform multi-level similarity calculation on the semantic feature vector according to semantic similarity, geographical similarity, and attribute similarity to obtain a similarity calculation result.

[0055] S13. Establish a dynamic adjustment mechanism of a POI adaptive fusion threshold based on data source credibility and historical accuracy of the multi-source POI data.

[0056] The calculation formula of the POI adaptive fusion threshold is shown in the following formula (1):

[0057] (1)

[0058] In the formula, POI adaptive fusion threshold, base threshold, adjustment coefficient, data source quality evaluation index.

[0059] S14, performing multi-source POI data fusion according to the similarity calculation result and the POI adaptive fusion threshold to generate a fused POI data set.

[0060] Optionally, the data fusion module is further configured to:

[0061] S111, establishing a POI field-specific dictionary containing commercial terms, geographical terms and field-specific vocabulary.

[0062] S112, using the BERT model to continue pre-training on the multi-source POI data according to the POI field-specific dictionary to obtain a pre-trained BERT model.

[0063] S113, calculating the semantic feature vector of the multi-source POI data according to the pre-trained BERT model and the multi-granularity feature fusion strategy.

[0064] Optionally, the AI inference module is further configured to:

[0065] S21, constructing a three-layer progressive inference structure based on a large language model; wherein the three-layer progressive inference structure includes a basic information inference layer, an associated information inference layer and a logical verification inference layer.

[0066] S22, performing explicit information analysis on the fused POI data set through the basic information inference layer to generate a preliminary floor judgment; performing implicit information analysis on the fused POI data set through the associated information inference layer to correct the floor judgment; and verifying the logical consistency and real-world rationality of the inference result through the logical verification inference layer to output the floor position probability.

[0067] S23, obtaining context information by analyzing the preliminary floor judgment, the corrected floor judgment and the floor position probability through Chain-of-Thought reasoning technology.

[0068] S24, constructing a floor position probability distribution model based on the context information, and obtaining a floor position inference result according to the floor position probability distribution model.

[0069] S25, a confidence evaluation mechanism is established to score the confidence of the floor position reasoning result, and the POI floor position and confidence are obtained.

[0070] Optionally, the function partition module is further used for:

[0071] S31, based on the spatial proximity, business correlation and people flow interaction of the fused POI data set, a function coupling strength index is calculated.

[0072] S32, a floor adjacency matrix is constructed based on the fused POI data set, and a Laplacian matrix is obtained according to the floor adjacency matrix.

[0073] S33, according to the function coupling strength index and the Laplacian matrix, an improved spectral clustering algorithm is used for three-dimensional function partition of the urban building group, and a function partition result of the urban building group is obtained; wherein the improved spectral clustering algorithm introduces floor connectivity constraints and function similarity weights.

[0074] S34, the partition boundary in the function partition result is optimized based on the simulated annealing algorithm, and an optimized function partition result is obtained.

[0075] Optionally, the attraction modeling module is further used for:

[0076] S41, mobile signaling data is obtained, the number of mobile phone users in each building boundary in the urban building group is extracted through a mobile signaling data processing algorithm, and building-level population statistics data is generated according to the number of mobile phone users.

[0077] S42, a differential attraction model is constructed according to the population statistics data and the optimized function partition result; wherein the differential attraction model includes multiple time dynamic models, each time dynamic model includes a basic attraction and a time attraction; the multiple time dynamic models include a commercial function attraction model, a residential function attraction model, an education function attraction model, a medical function attraction model, an office function attraction model and a comprehensive function attraction model.

[0078] S43, according to the differential attraction model, the attraction weight coefficient of each floor function partition to the population is calculated.

[0079] Optionally, the population distribution optimization module is further used for:

[0080] S51, a floor population distribution constraint optimization model is established.

[0081] S52, according to the function partition of each floor to the attraction weight coefficient of population, the improved Haaf model is used to solve the floor population distribution constraint optimization model, and the dynamic distribution of population of each floor is obtained; wherein the improved Haaf model introduces floor capacity constraint and function matching degree constraint.

[0082] S53, a dynamic adjustment mechanism is established, and the dynamic distribution of population of each floor is adjusted in real time based on the real-time acquisition of mobile phone signaling data and the dynamic adjustment mechanism; wherein the dynamic adjustment mechanism includes designing a sliding window data updating strategy and updating the attraction weight coefficient by using a weighted average method; the length of the sliding window is adaptively adjusted according to the data change frequency.

[0083] On the other hand, a kind of urban building group three-dimensional function partition and floor population dynamic distribution equipment is provided, and the urban building group three-dimensional function partition and floor population dynamic distribution equipment include: processor;Memory, at least one instruction is stored on the memory, the at least one instruction is loaded and executed by processor to realize any one method in the above urban building group three-dimensional function partition and floor population dynamic distribution method.

[0084] In another aspect, a computer-readable storage medium is provided, and the storage medium stores at least one instruction, which is loaded and executed by a processor to implement any one of the above urban building group three-dimensional function partition and floor population dynamic distribution methods.

[0085] The technical scheme provided by the embodiment of the application has at least the following beneficial effects:

[0086] In the present application, a kind of urban building group three-dimensional function partition and floor population dynamic distribution method and system are provided, which can intelligently integrate multi-source POI data and complete the missing floor position information, break through the granularity limit of traditional function partition, realize the fine three-dimensional function partition of floor level, scientifically calculate the attraction of each function partition to population, realize the optimal stratified distribution of total population of building, and provide real-time dynamic adjustment based on sliding window data updating strategy and abnormal detection mechanism, to provide important decision support for key applications such as commercial site selection, emergency evacuation and facility planning in smart city construction, effectively improve the intelligent and fine level of urban space analysis. BRIEF DESCRIPTION OF DRAWINGS

[0087] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0088] Figure 1is a method flow chart of the urban building group three-dimensional function zoning and floor population dynamic distribution provided by the embodiment of the present application;

[0089] Figure 2 is a POI floor position inference flow chart based on a large language model provided by the embodiment of the present application;

[0090] Figure 3 is a city building group hierarchical population heat map provided by the embodiment of the present application;

[0091] Figure 4 is a schematic diagram of a device for urban building group three-dimensional function zoning and floor population dynamic distribution based on AI semantic analysis and multi-source data fusion provided by the embodiment of the present application;

[0092] Figure 5 is a structural schematic diagram of a device for urban building group three-dimensional function zoning and floor population dynamic distribution provided by the embodiment of the present application. DETAILED DESCRIPTION

[0093] The technical solutions in the present application will be described below with reference to the drawings.

[0094] In the embodiments of the present application, the words such as "example", "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0095] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0096] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0097] In order to make the technical problems, technical solutions and advantages to be solved by the present application clearer, specific embodiments will be described in detail below with reference to the drawings.

[0098] The embodiment of the application provides a kind of urban building group three-dimensional function zoning and floor population dynamic distribution method, this method can be realized by urban building group three-dimensional function zoning and floor population dynamic distribution equipment, the urban building group three-dimensional function zoning and floor population dynamic distribution equipment can be terminal or server.As Figure 1 The processing flow of the method can include the following steps as shown in the flow chart of the urban building group three-dimensional function zoning and floor population dynamic distribution method.

[0099] S1, obtain the multi-source POI data of the urban building group, construct the semantic feature vector of the multi-source POI data based on the pre-trained BERT model combined with the field-specific vocabulary, perform multi-level similarity calculation on the semantic feature vector, and generate the fused POI data set according to the similarity calculation result.

[0100] Optionally, the step S1 of constructing the semantic feature vector of the multi-source POI data based on the pre-trained BERT model combined with the field-specific vocabulary, performing multi-level similarity calculation on the semantic feature vector, and generating the fused POI data set according to the similarity calculation result can include the following steps S11-S14:

[0101] S11, construct the semantic feature vector of the multi-source POI data according to the multi-source POI data, the pre-trained BERT model and the field-specific vocabulary, and generate a 768-dimensional semantic embedding representation.

[0102] Optionally, the step S11 can include the following steps S111-S113:

[0103] S111, establish a POI field-specific dictionary containing commercial terms, geographical nouns and field-specific vocabulary.

[0104] The field-specific vocabulary is an industry-specific vocabulary, which is a self-defined standardized vocabulary library. The POI field-specific dictionary constructed eliminates the ambiguity caused by polysemous words by standardizing the same type of vocabulary from different data sources, thereby enhancing semantic understanding and assisting feature extraction.

[0105] S112, according to the POI field-specific dictionary, continue pre-training on the multi-source POI data using a mask language model, improve the field adaptability of the model, and obtain a trained mask language model.

[0106] The BERT model belongs to a kind of mask language model, and the mask language model learns to predict the masked word through context text by masking part of the input text, so as to generate a vector containing semantic information.

[0107] S113, calculate the semantic feature vector of the multi-source POI data according to the trained mask language model and the multi-granularity feature fusion strategy.

[0108] wherein the calculation formula of the multi-granularity feature fusion is: .

[0109] S12, multi-level similarity calculation is performed on the semantic feature vector according to the semantic similarity, geographical similarity and attribute similarity, and a similarity calculation result is obtained.

[0110] wherein the calculation formula of the multi-level similarity is: .

[0111] S13, a dynamic threshold adjustment mechanism is established based on the data source credibility and historical accuracy of the multi-source POI data, and the POI fusion threshold is adaptively adjusted.

[0112] wherein the calculation formula of the POI adaptive fusion threshold is shown in the following formula (1):

[0113] (1)

[0114] In the formula, POI adaptive fusion threshold, base threshold, adjustment coefficient, data source quality evaluation index.

[0115] S14, multi-source POI data fusion is performed according to the similarity calculation result and the POI adaptive fusion threshold, and a fused high-quality POI data set is generated.

[0116] S2, a three-layer progressive reasoning structure based on a large language model is constructed, a floor position probability distribution model is constructed according to the fused POI data set, the three-layer progressive reasoning structure and the Chain-of-Thought reasoning technology, and a POI floor position is obtained.

[0117] Optionally, as shown in Figure 2 the above step S2 can include the following steps S21-S25:

[0118] S21, a three-layer progressive reasoning structure based on a large language model is constructed; wherein the three-layer progressive reasoning structure includes a basic information reasoning layer, an associated information reasoning layer and a logical verification reasoning layer.

[0119] S22, explicit information analysis is performed on the fused POI data set through the basic information reasoning layer to generate a preliminary floor judgment; implicit information analysis is performed on the fused POI data set through the associated information reasoning layer to correct the floor judgment; logical consistency and reality rationality of the reasoning result are verified through the logical verification reasoning layer, and a floor position probability is output.

[0120] In a feasible implementation, the first layer of basic information reasoning: analyze the explicit information such as POI name, category, business hours, etc., to generate a preliminary floor judgment.

[0121] The second layer of associated information reasoning: combine implicit information such as surrounding POI distribution, building type, regional characteristics, etc., to correct the floor judgment.

[0122] The third layer of logical verification reasoning: verify the logical consistency and realistic rationality of the reasoning result, and output the final floor position probability.

[0123] S23, according to the preliminary floor judgment, the corrected floor judgment and the floor position probability, the context information is analyzed by the Chain-of-Thought reasoning technology.

[0124] In a feasible implementation, based on Chain-of-Thought reasoning technology, analyze POI format, surrounding environment, user comments and other context information.

[0125] S24, based on the context information, a floor position probability distribution model is constructed, and a floor position reasoning result is obtained according to the floor position probability distribution model.

[0126] The calculation formula of the probability distribution model is: (2)

[0127] Wherein, represents the probability distribution of which floor under the given context condition, represents the weight matrix, represents the semantic feature vector of the context, represents the bias vector.

[0128] S25, a confidence evaluation mechanism is established, the floor position reasoning result is scored for credibility, and the POI floor position and confidence are obtained.

[0129] The calculation formula of the confidence degree includes:

[0130] (3)

[0131] Wherein, is the confidence of the first floor, is the weight of the first reasoning factor, is the support probability of the first reasoning factor to the first floor, is the total number of reasoning factors.

[0132] The application is based on a context reasoning mechanism of a large language model, extracts implicit floor position information of a POI, and generates a floor position probability distribution.

[0133] S3, according to the fused POI data set, the function coupling strength index and the improved spectral clustering algorithm, the internal of the urban building group is stereoscopic function partitioned, and the function partition result of the internal of the urban building group is obtained; the function partition result is optimized based on the simulated annealing algorithm, and the optimized function partition result is obtained.

[0134] Optionally, the above step S3 can include the following steps S31-S34:

[0135] S31, based on the spatial proximity, the business correlation and the people flow interactivity of the fused POI data set, the function coupling strength index is calculated.

[0136] The calculation formula of the function coupling strength is: The function partition is divided based on the function coupling strength index.

[0137] S32, a floor adjacency matrix is constructed based on the fused POI data set, and a Laplacian matrix is obtained according to the floor adjacency matrix.

[0138] In a feasible implementation manner, the floor adjacency matrix is constructed , the Laplacian matrix is modified; the eigenvector of the Laplacian matrix is calculated, and the calculation formula of the Laplacian matrix includes:

[0139] (4)

[0140] Wherein, represents the modified Laplacian matrix, represents the Laplacian matrix, represents the topological constraint weight. Based on the Laplacian matrix, the spatial relationship of each functional unit is expressed in a mathematical form, which provides a basis for subsequent automatic partition.

[0141] S33, according to the function coupling strength index and the Laplacian matrix, the internal of the urban building group is stereoscopic function partitioned by using the improved spectral clustering algorithm, and the function partition result of the internal of the urban building group is obtained; wherein the improved spectral clustering algorithm introduces floor connectivity constraint and function similarity weight.

[0142] Compared with the existing spectral clustering algorithm, the improved spectral clustering algorithm introduces a functional coupling strength index, a spatial connectivity constraint and a functional similarity weight, realizes fusion modeling and constraint optimization of multi-source heterogeneous data, and effectively solves the problems of the existing method, such as only relying on spatial adjacency, ignoring spatial topology and functional heterogeneity, and the like. The improved algorithm significantly improves the scientificity and automation level of spatial functional partitioning.

[0143] S34, based on the simulated annealing algorithm, the partition boundary in the functional partitioning result is optimized, and the functional conflict degree in the partition is minimized, to obtain an optimized functional partitioning result.

[0144] S4, generating building-level population statistics data through a mobile signaling data processing algorithm, constructing a differentiated attractiveness model according to the population statistics data and the optimized functional partitioning result, and calculating the attraction weight coefficient of each functional partition to the population.

[0145] Optionally, the above step S4 can include the following steps S41-S43:

[0146] S41, obtaining mobile signaling data, extracting the number of mobile phone users within the boundary of each building in the urban building group through a mobile signaling data processing algorithm, and generating building-level population statistics data according to the number of mobile phone users.

[0147] S42, constructing a differentiated attractiveness model according to the population statistics data and the optimized functional partitioning result; wherein the differentiated attractiveness model includes a plurality of time dynamic models, each time dynamic model includes a basic attractiveness and a time attractiveness; the plurality of time dynamic models include a commercial function attractiveness model, a residential function attractiveness model, an education function attractiveness model, a medical function attractiveness model, an office function attractiveness model and a comprehensive function attractiveness model.

[0148] In a feasible implementation manner, the differentiated attractiveness model includes a commercial function attractiveness model, a residential function attractiveness model, an education function attractiveness model, a medical function attractiveness model, an office function attractiveness model and a comprehensive function attractiveness model, each model is a time dynamic model, and includes two components of a basic attractiveness and a time attractiveness.

[0149] Specifically, the commercial function attractiveness model is: ;

[0150] Wherein, ;

[0151] .

[0152] The residential function attractiveness model is: ;

[0153] wherein, ;

[0154] .

[0155] The education function attraction model is: ;

[0156] wherein, ;

[0157] .

[0158] The medical function attraction model is: ;

[0159] wherein, ;

[0160] .

[0161] The office function attraction model is: ;

[0162] wherein, ;

[0163] .

[0164] The comprehensive function attraction model is: ; wherein, .

[0165] S43, according to the differentiated attraction model, calculate the attraction weight coefficient of each floor function partition to the population. The attraction weight coefficient is taken as an input parameter of the Haf model, that is, the attraction factor.

[0166] S5, establish a floor population distribution constraint optimization model, according to the attraction weight coefficient of each floor function partition to the population and the improved Haf model, solve the floor population distribution constraint optimization model, and obtain the dynamic distribution of the population of each floor; a dynamic adjustment mechanism is established to adjust the dynamic distribution of the population of each floor in real time.

[0167] Optionally, the above step S5 can include the following steps S51-S53:

[0168] S51, establish a floor population distribution constraint optimization model.

[0169] S52, according to the attraction weight coefficient of each floor function partition to the population, solve the floor population distribution constraint optimization model by using the improved Haf model, and obtain the dynamic distribution of the population of each floor; wherein, the improved Haf model introduces floor capacity constraints and function matching degree constraints.

[0170] In a feasible implementation, the total population of the building is distributed according to the attraction weight of the function zoning of each floor, and the dynamic population distribution of each floor is generated. The improved Haf model is used to solve the personnel floor distribution scheme.

[0171] The floor population distribution constraint optimization model is established as follows: The constraint condition is as follows: .

[0172] Further, a floor capacity constraint processing mechanism is designed to ensure that the distributed population does not exceed the maximum carrying capacity of each floor, and the maximum carrying capacity is calculated based on the floor area, function type and safety specification;

[0173] The smoothness constraint between floors is introduced to minimize the mutation of the population density of adjacent floors, and the constraint formula is as follows: .

[0174] The traditional model mostly only considers static attraction and ignores the time-varying nature of population flow, and lacks constraints on floor-level capacity, which may lead to floor overload. The present application introduces a differentiated attraction model to provide attraction factors for the improved Haf model, solves the problem that the static and single attraction of the traditional model cannot reflect the actual population flow rule, and introduces the floor capacity constraint to solve the problem that the traditional model lacks constraints and is prone to overload.

[0175] S53, a dynamic adjustment mechanism is established, and the dynamic population distribution of each floor is adjusted in real time based on the real-time acquisition of mobile signaling data and the dynamic adjustment mechanism; wherein the dynamic adjustment mechanism includes designing a sliding window data update strategy and updating the attraction weight coefficient by using a weighted average method; the length of the sliding window is adaptively adjusted according to the data change frequency.

[0176] In a feasible implementation, a sliding window data update strategy is designed to maintain the mobile signaling observation data of the last time period, and the length of the sliding window is adaptively adjusted according to the data change frequency.

[0177] Further, the attraction weight parameter is updated by using a weighted average method, and the parameter update formula is as follows: , wherein is a learning rate parameter;

[0178] Further, an anomaly detection mechanism is established, and when the deviation between the real-time observation data and the prediction result exceeds a set threshold, the weight parameter is automatically recalibrated.

[0179] S6, according to the dynamic population distribution of each floor and the optimized function zoning result, a three-dimensional population distribution model of the building is constructed, and a hierarchical population heat map and a function-population correlation graph of the urban building group are generated.

[0180] In a feasible implementation, based on the floor population distribution result and the function partition information, a building three-dimensional population distribution model is constructed, and a hierarchical population heat map and a function-population correlation graph of the urban building group are generated.

[0181] Specifically, based on the prediction result, a three-dimensional function distribution graph and a floor population density heat map of the urban building group are generated.

[0182] A hierarchical population heat map is constructed to intuitively display the population density distribution of each floor, such as Figure 3 as shown in the figure.

[0183] A function-population correlation graph is generated to reveal the correlation between different function partitions and population distribution.

[0184] Decision support is provided for different application scenarios, including business site selection recommendation, emergency evacuation path planning, and public service facility configuration optimization.

[0185] In the embodiment of the application, a POI semantic feature vector is constructed based on a pre-trained BERT model combined with domain-specific vocabulary, a high-quality POI data set is generated by multi-level similarity calculation, a three-layer progressive reasoning structure based on a large language model is adopted, Chain-of-Thought reasoning technology is used to extract POI implicit floor location information, and a floor location probability distribution is generated, a three-dimensional function partition of the building group is performed based on function coupling degree and an improved spectral clustering algorithm, a total population of the building is obtained based on mobile signaling data, six types of differentiated floor attraction models are constructed, each model is a time dynamic model, a floor population distribution optimization algorithm based on an improved Haaf model is introduced, floor capacity constraints and smoothness constraints are introduced, and the total population of the building is distributed according to the attraction weight of each floor function partition; a dynamic adjustment mechanism based on a sliding window data update strategy and an abnormality detection mechanism is established.

[0186] Figure 4 is a device block diagram of a three-dimensional function partition and floor population dynamic distribution of an urban building group according to an exemplary embodiment, which is used for a three-dimensional function partition and floor population dynamic distribution method of an urban building group. Referring to Figure 4 , the device includes a data fusion module 10, an AI reasoning module 20, a function partition module 30, an attraction modeling module 40, a population distribution optimization module 50, and a visualization display module 60. Among them:

[0187] The data fusion module 10 is used to obtain multi-source POI data of the urban building group, construct semantic feature vectors of the multi-source POI data based on a pre-trained BERT model combined with domain-specific vocabulary, perform multi-level similarity calculation on the semantic feature vectors, and generate a fused POI data set according to the similarity calculation result.

[0188] An AI inference module 20 is configured to construct a three-layer progressive inference structure based on a large language model, construct a floor position probability distribution model according to the fused POI data set, the three-layer progressive inference structure and a Chain-of-Thought inference technology, and obtain a POI floor position.

[0189] A functional partition module 30 is configured to perform three-dimensional functional partitioning on an internal city building group according to the fused POI data set, a functional coupling strength index and an improved spectral clustering algorithm, to obtain a functional partition result of the internal city building group; and perform optimization on the functional partition result based on a simulated annealing algorithm, to obtain an optimized functional partition result.

[0190] An attraction modeling module 40 is configured to generate building-level population statistics by a mobile phone signaling data processing algorithm, construct a differential attraction model according to the population statistics and the optimized functional partition result, and calculate attraction weight coefficients of each functional partition to the population.

[0191] A population distribution optimization module 50 is configured to establish a floor population distribution constraint optimization model, solve the floor population distribution constraint optimization model according to the attraction weight coefficients of each floor functional partition to the population and an improved Huff model, to obtain a population dynamic distribution of each floor; and establish a dynamic adjustment mechanism to adjust the population dynamic distribution of each floor in real time.

[0192] A visual display module 60 is configured to construct a building three-dimensional population distribution model according to the population dynamic distribution of each floor and the optimized functional partition result, and generate a hierarchical population heat map and a function-population correlation graph of the city building group.

[0193] The data fusion module of the present application is configured to perform intelligent fusion processing of multi-source POI data; the AI inference module is configured to perform floor position inference based on a large language model; the functional partition module is configured to perform three-dimensional functional partitioning based on an improved spectral clustering; the attraction modeling module is configured to perform differential floor attraction model construction and calculation based on the functional partition type; the population distribution optimization module is configured to perform floor population dynamic distribution calculation and result generation based on an optimization algorithm; and the visual display platform is configured to provide interactive display functions of the building three-dimensional population distribution model, the hierarchical population heat map and the function-population correlation graph based on Web services.

[0194] Specifically, the data fusion module 10 includes a data acquisition unit, a preprocessing unit, a semantic feature extraction unit and a fusion processing unit, and is configured to realize high-quality fusion of multi-source POI data.

[0195] The AI inference module 20 comprises three-layer inference units, Chain-of-Thought inference units, probability distribution calculation units and confidence evaluation units, and is used for accurately inferring POI floor position information.

[0196] The functional partition module 30 comprises coupling degree calculation units, spectral clustering units, constraint optimization units and partition result generation units, and is used for realizing three-dimensional functional partition of the building group.

[0197] The attraction modeling module 40 comprises six types of attraction model construction units, corresponding to business, residence, education, medical treatment, office and comprehensive functions respectively, and each unit supports time dynamic modeling.

[0198] The population distribution optimization module 50 comprises constraint optimization units, Haaf model solving units, capacity constraint processing units and smoothness constraint units, and is used for realizing optimal floor population distribution.

[0199] The visualization display module 60 comprises three-dimensional rendering units, heat map generation units, correlation diagram generation units and interactive query units, and provides comprehensive result display and analysis functions.

[0200] Further, the device further comprises:

[0201] A distributed data processing engine supports parallel processing and real-time updating of large-scale POI data and mobile signaling data;

[0202] A cache system realizes efficient storage and rapid access of model parameters, intermediate calculation results and historical data;

[0203] An automated operation and maintenance mechanism dynamically adjusts resource allocation according to the calculation load, and ensures stable operation of the system.

[0204] In the embodiment of the present application, a method and system for three-dimensional functional partition and dynamic floor population distribution of urban building groups are provided. A POI semantic feature vector is constructed by pre-training a BERT model combined with field-specific vocabulary, and high-quality data fusion is realized based on multi-level similarity calculation. A three-layer progressive inference structure of a large language model is used to accurately extract floor position information. Three-dimensional functional partition is realized based on functional coupling degree and an improved spectral clustering algorithm. Six types of differentiated time dynamic attraction models are constructed. Reasonable allocation of floor population is realized through an improved Haaf model and constraint optimization. Dynamic adjustment functions are provided based on a sliding window and an anomaly detection mechanism. The method realizes intelligent, refined and dynamic analysis of three-dimensional functional partition and dynamic floor population distribution of urban building groups, and provides important technical support for smart city construction.

[0205] Figure 5 is a structural schematic diagram of a kind of equipment for three-dimensional functional partition and dynamic floor population distribution of urban building group provided in the embodiment of the present application, asFigure 5 The urban building group three-dimensional functional zoning and floor population dynamic distribution device can include the above-mentioned Figure 4 The urban building group three-dimensional functional zoning and floor population dynamic distribution device. Optionally, the urban building group three-dimensional functional zoning and floor population dynamic distribution device 410 can include a first processor 2001.

[0206] Optionally, the urban building group three-dimensional functional zoning and floor population dynamic distribution device 410 can further include a memory 2002 and a transceiver 2003.

[0207] The first processor 2001 is connected with the memory 2002 and the transceiver 2003, for example, through a communication bus.

[0208] The following will be described in detail Figure 5 The various constituent components of the urban building group three-dimensional functional zoning and floor population dynamic distribution device 410 will be described in detail:

[0209] The first processor 2001 is the control center of the urban building group three-dimensional functional zoning and floor population dynamic distribution device 410, and can be one processor or a collective term of multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or can be an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application, such as one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0210] Optionally, the first processor 2001 can execute various functions of the urban building group three-dimensional functional zoning and floor population dynamic distribution device 410 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0211] In a specific implementation, as an embodiment, the first processor 2001 can include one or more CPUs, such as the CPU0 and CPU1 shown in Figure 5

[0212] In a specific implementation, as an embodiment, the urban building group three-dimensional functional zoning and floor population dynamic distribution device 410 can also include multiple processors, such as the CPU0 and CPU1 shown in Figure 5 ​1 and 2. The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0213] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled by the first processor 2001 for execution. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0214] Alternatively, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or exist independently and access the memory 2002 through the interface circuit ( Figure 5 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0215] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0216] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 5 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0217] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or can exist independently and can be connected to the first processor 2001 through the interface circuit of the urban building complex three-dimensional functional zoning and floor population dynamic distribution device 410 ( Figure 5The first processor 2001 is coupled with the city building group three-dimensional function partition and floor population dynamic distribution device 410 (not shown in the figure), and embodiments of the present application do not make specific limitation thereto.

[0218] It should be noted that, Figure 5 The structure of the city building group three-dimensional function partition and floor population dynamic distribution device 410 shown in the figure does not constitute a limitation on the router, and the actual knowledge structure recognition device can include more or fewer components than those shown in the figure, or combine certain components, or different component arrangements.

[0219] In addition, the technical effects of the city building group three-dimensional function partition and floor population dynamic distribution device 410 can refer to the technical effects of the city building group three-dimensional function partition and floor population dynamic distribution method described in the above method embodiments, which will not be repeated here.

[0220] It should be understood that the first processor 2001 in the embodiments of the present application can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), ready programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0221] It should also be understood that the memory in the embodiments of the present application can be volatile or nonvolatile memory, or can include both volatile and nonvolatile memory. Nonvolatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically EPROM (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as external cache. By way of example, and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0222] The above-described embodiments can be implemented in whole or in part by software, hardware (such as a circuit), firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center through a wired (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0223] It should be understood that the term "and / or" herein merely describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after it are in an "or" relationship, but it can also represent an "and / or" relationship, which can be understood according to the context before and after it.

[0224] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0225] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0226] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0227] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the devices, apparatuses and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0228] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0229] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0230] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0231] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of software products. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0232] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for three-dimensional functional zoning of urban buildings and dynamic distribution of floor population, characterized in that: The method comprises: S1. Obtain multi-source POI data of urban building complexes, construct semantic feature vectors of the multi-source POI data based on the pre-trained BERT model combined with domain-specific vocabulary, perform multi-level similarity calculation on the semantic feature vectors, and generate a fused POI dataset based on the similarity calculation results; S2. Build a three-layer progressive reasoning structure based on a large language model. Based on the fused POI dataset, the three-layer progressive reasoning structure, and the Chain-of-Thought reasoning technique, a floor location probability distribution model is constructed to obtain the POI floor location. S3. Perform three-dimensional functional zoning of the urban building complex based on the fused POI dataset, the functional coupling strength index, and the improved spectral clustering algorithm to obtain the functional zoning results within the urban building complex; optimize the functional zoning results based on the simulated annealing algorithm to obtain the optimized functional zoning results; S4. Generate building-level demographic data through a mobile phone signaling data processing algorithm, construct a differentiated attraction model based on the demographic data and the optimized functional zoning results, and calculate the attraction weight coefficient of each functional zoning for the population; S5. Establish a floor population allocation constraint optimization model. Based on the attraction weight coefficient of the functional zoning of each floor and the improved Hough model, solve the floor population allocation constraint optimization model to obtain the dynamic population distribution of each floor; establish a dynamic adjustment mechanism to adjust the dynamic population distribution of each floor in real time; S6. Based on the dynamic population distribution on each floor and the optimized functional zoning results, a three-dimensional population distribution model of the building is constructed to generate a stratified population heat map and a function-population correlation map of the urban building complex.

2. The method for three-dimensional functional zoning of urban buildings and dynamic distribution of floor population according to claim 1 is characterized in that: The S1 constructs semantic feature vectors of multi-source POI data based on the pre-trained BERT model combined with domain-specific vocabulary, performs multi-level similarity calculation on the semantic feature vectors, and generates a fused POI dataset based on the similarity calculation results, including: S11. Construct semantic feature vectors of multi-source POI data based on multi-source POI data, pre-trained BERT model, and domain-specific vocabulary. S12, performing multi-level similarity calculation on the semantic feature vector according to semantic similarity, geographical similarity, and attribute similarity to obtain a similarity calculation result; S13. Based on the data source credibility and historical accuracy of multi-source POI data, a dynamic adjustment mechanism for the POI adaptive fusion threshold is established; The calculation formula of the POI adaptive fusion threshold is as shown in the following formula (1): （1） Where, represents the POI adaptive fusion threshold, represents the basic threshold, represents the adjustment coefficient, Represents the data source quality assessment indicator; S14. Perform multi-source POI data fusion based on the similarity calculation result and the POI adaptive fusion threshold to generate a fused POI dataset.

3. The method for three-dimensional functional zoning of urban buildings and dynamic population distribution on each floor according to claim 2 is characterized in that: The step S11 of constructing a semantic feature vector of the multi-source POI data based on the multi-source POI data, the pre-trained BERT model, and the domain-specific vocabulary includes: S111. Establish a POI domain-specific dictionary containing business terms, geographical terms, and domain-specific vocabulary; S112. Based on the POI domain-specific dictionary, the BERT model is used to continue pre-training on the multi-source POI data to obtain a pre-trained BERT model; S113. Calculate the semantic feature vector of multi-source POI data based on the pre-trained BERT model and the multi-granularity feature fusion strategy.

4. The method for three-dimensional functional zoning of urban buildings and dynamic population distribution on each floor according to claim 1 is characterized in that: The construction in S2 is based on a three-layer progressive reasoning structure of a large language model. The floor location probability distribution model is constructed based on the fused POI dataset, the three-layer progressive reasoning structure, and the Chain-of-Thought reasoning technology to obtain the POI floor location, including: S21. Construct a three-layer progressive reasoning structure based on a large language model; wherein the three-layer progressive reasoning structure includes a basic information reasoning layer, an associated information reasoning layer, and a logical verification reasoning layer; S22. Perform explicit information analysis on the fused POI dataset through the basic information reasoning layer to generate a preliminary floor judgment; perform implicit information analysis on the fused POI dataset through the associated information reasoning layer to revise the floor judgment; verify the logical consistency and practical rationality of the reasoning result through the logical verification reasoning layer, and output the floor location probability; S23, based on the preliminary floor judgment, the revised floor judgment and the floor location probability, obtain context information through Chain-of-Thought reasoning technology analysis; S24. Construct a floor position probability distribution model based on the context information, and obtain a floor position inference result according to the floor position probability distribution model; S25. Establish a confidence evaluation mechanism to perform a credibility score on the floor position inference result to obtain the POI floor position and confidence.

5. The method for three-dimensional functional zoning of urban buildings and dynamic population distribution on each floor according to claim 1 is characterized in that: In said S3, a three-dimensional functional zoning is performed on the interior of the urban building complex according to the fused POI dataset, the functional coupling strength index and the improved spectral clustering algorithm to obtain a functional zoning result within the urban building complex; The functional partitioning results are optimized based on the simulated annealing algorithm to obtain the optimized functional partitioning results, including: S31. Calculate the functional coupling strength index based on the spatial proximity, business relevance, and human interaction of the fused POI dataset; S32, constructing a floor adjacency matrix based on the fused POI dataset, and obtaining a Laplace matrix according to the floor adjacency matrix; S33. Based on the functional coupling strength index and the Laplace matrix, an improved spectral clustering algorithm is used to perform three-dimensional functional zoning within the urban building complex, thereby obtaining the functional zoning results within the urban building complex; wherein the improved spectral clustering algorithm introduces floor connectivity constraints and functional similarity weights; S34. Optimize the partition boundaries in the functional partitioning result based on a simulated annealing algorithm to obtain an optimized functional partitioning result.

6. The method for three-dimensional functional zoning of urban buildings and dynamic population distribution on each floor according to claim 1 is characterized in that: The S4 generates building-level demographic data through a mobile phone signaling data processing algorithm, constructs a differentiated attraction model based on the demographic data and the optimized functional zoning results, and calculates the attraction weight coefficient of each functional zoning to the population, including: S41. Obtain mobile phone signaling data, extract the number of mobile phone users within the boundaries of each building in the urban building complex through a mobile phone signaling data processing algorithm, and generate building-level demographic data based on the number of mobile phone users; S42. Construct a differentiated attraction model based on demographic data and the optimized functional zoning results; wherein the differentiated attraction model includes multiple time-dynamic models, each of which includes basic attraction and time attraction; the multiple time-dynamic models include a commercial function attraction model, a residential function attraction model, an educational function attraction model, a medical function attraction model, an office function attraction model, and a comprehensive function attraction model; S43. Based on the differentiated attraction model, calculate the attraction weight coefficient of the functional zoning on each floor to the population.

7. The method for three-dimensional functional zoning of urban buildings and dynamic population distribution on each floor according to claim 1 is characterized in that: The above-mentioned S5 establishes a floor population allocation constraint optimization model, solves the floor population allocation constraint optimization model based on the attraction weight coefficient of the functional zoning of each floor to the population and the improved Hough model, and obtains the dynamic distribution of the population on each floor; Establish a dynamic adjustment mechanism to adjust the dynamic distribution of population on each floor in real time, including: S51. Establish a constrained optimization model for floor population distribution; S52. Based on the weight coefficient of population attraction of each floor functional zoning, an improved Hough model is used to solve the floor population allocation constraint optimization model to obtain the dynamic population distribution of each floor; wherein the improved Hough model introduces floor capacity constraints and functional matching constraints; S53. Establish a dynamic adjustment mechanism to adjust the dynamic distribution of the population on each floor in real time based on the real-time acquired mobile phone signaling data and the dynamic adjustment mechanism; the dynamic adjustment mechanism includes designing a sliding window data update strategy and adopting a weighted average method to update the attraction weight coefficient; the length of the sliding window is adaptively adjusted according to the frequency of data changes.

8. A device for three-dimensional functional zoning of urban building complexes and dynamic distribution of population on each floor, the device being used to implement the method for three-dimensional functional zoning of urban building complexes and dynamic distribution of population on each floor as claimed in any one of claims 1 to 7, characterized in that: The device comprises: The data fusion module is used to obtain multi-source POI data of urban building complexes, construct semantic feature vectors of multi-source POI data based on the pre-trained BERT model and domain-specific vocabulary, perform multi-level similarity calculations on the semantic feature vectors, and generate a fused POI dataset based on the similarity calculation results; The AI ​​reasoning module is used to build a three-layer progressive reasoning structure based on a large language model. It constructs a floor location probability distribution model based on the fused POI dataset, the three-layer progressive reasoning structure, and the Chain-of-Thought reasoning technology to obtain the POI floor location. The functional zoning module is used to perform three-dimensional functional zoning within the urban building complex based on the fused POI dataset, functional coupling intensity index, and improved spectral clustering algorithm to obtain the functional zoning results within the urban building complex; the functional zoning results are optimized based on the simulated annealing algorithm to obtain the optimized functional zoning results; The attraction modeling module is used to generate building-level demographic data through mobile phone signaling data processing algorithms, build a differentiated attraction model based on the demographic data and the optimized functional zoning results, and calculate the attraction weight coefficient of each functional zoning for the population; The population distribution optimization module is used to establish a floor population distribution constraint optimization model. Based on the attraction weight coefficient of the functional zoning of each floor and the improved Hough model, the floor population distribution constraint optimization model is solved to obtain the dynamic population distribution of each floor; a dynamic adjustment mechanism is established to adjust the dynamic population distribution of each floor in real time; The visualization display module is used to construct a three-dimensional population distribution model of buildings based on the dynamic population distribution on each floor and the optimized functional zoning results, and to generate a layered population heat map and function-population correlation map of the urban building complex.

9. A device for three-dimensional functional zoning of urban buildings and dynamic distribution of floor population, characterized in that: The three-dimensional functional zoning and floor population dynamic distribution equipment of the urban building complex includes: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which can be called by a processor to execute the method according to any one of claims 1 to 7.

Images