A method and system for urban spatial intensity zoning based on Bayesian networks

By using Bayesian networks and 3D holographic projection technology, the scientific rigor and dynamic optimization issues of traditional urban spatial intensity zoning methods have been addressed, enabling the generation of more accurate and interactive urban spatial intensity zoning schemes.

CN117668688BActive Publication Date: 2026-11-17SOUTHEAST UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202311437132.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-31
Publication Date
2026-11-17
Estimated Expiration
2043-10-31

AI Technical Summary

Technical Problem

Traditional urban spatial intensity zoning methods lack comprehensive consideration of multiple factors, are not scientifically sound, and cannot be dynamically and intuitively adjusted and optimized.

Method used

A Bayesian network-based urban spatial intensity zoning method is adopted. Through data processing, correlation analysis, clustering algorithms and model training, a street spatial intensity prediction model is constructed, and a 3D holographic projection device is used for scheme display and interactive optimization.

Benefits of technology

It improves the accuracy and scientific nature of urban spatial intensity zoning, enhances the displayability and interactivity of the schemes, and enables real-time selection and optimization of multiple schemes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117668688B_ABST
    Figure CN117668688B_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for urban spatial intensity zoning based on Bayesian networks. The method includes inputting target block and case data and calculating indicators; determining the Bayesian network training set by filtering training set indicators and identifying all categories of the case block; training the Bayesian network based on the classification to obtain a prediction model; determining the category of the target block according to the classification of the case block dataset, inputting the target block dataset into the prediction model of the corresponding category to obtain the spatial intensity of the target block; and outputting and interactively displaying the predicted scheme. This invention can intelligently predict urban spatial intensity zoning in the field of urban planning and design, realize urban spatial intensity zoning based on Bayesian networks to optimize urban spatial patterns, and make urban spatial intensity zoning more reasonable and of higher quality through indicator quantification.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of urban planning technology, specifically relating to a method and system for urban spatial intensity zoning based on Bayesian networks. Background Technology

[0002] Urban spatial intensity zoning is a crucial aspect of urban planning, serving as an important method for controlling and guiding urban development, and a vital means for shaping spatial form and enriching spatial hierarchy in urban design. Rational urban spatial intensity zoning plays a significant role in the rational use of land resources, controlling uncontrolled urban sprawl, protecting the ecological environment, and improving urban quality. Traditional urban spatial intensity zoning methods typically rely on designers' practical experience and professional knowledge to arrive at zoning schemes. However, these methods suffer from excessive subjectivity and insufficient scientific rigor, failing to allow for dynamic and intuitive adjustments and optimization of the zoning plan. Summary of the Invention

[0003] Purpose of the invention: To address the shortcomings of existing technologies, this invention provides a method and system for urban spatial intensity zoning based on Bayesian networks. This solves the problems of traditional urban spatial intensity zoning lacking comprehensive consideration of multiple factors, insufficient scientific rationality and data quantification support, and the inability to dynamically, intuitively and effectively adjust and optimize urban spatial intensity zoning schemes.

[0004] Technical solution: The urban spatial intensity zoning method based on Bayesian networks described in this invention specifically includes the following steps:

[0005] (1) The target street and the street vector data of the city where the target street are located are processed in a unified format to construct a digital sand table of the target street, forming a target street dataset and a case street dataset respectively. The data is processed into street external feature indicators, street internal feature indicators, and spatial intensity indicators, and then summarized to form an indicator library.

[0006] (2) By performing correlation analysis on the indicator library, indicators with high correlation were screened out and removed; and by using clustering algorithms, all sub-datasets of the case street dataset were determined.

[0007] (3) The Bayesian network is trained using the aforementioned sub-datasets, and after structure learning and parameter learning, different classification street spatial intensity prediction models are obtained.

[0008] (4) Based on the classification of the case street dataset, the dataset of the street to be predicted is classified using a classification algorithm to determine the category to which the target street belongs; based on the category to which the target street belongs, the target street dataset is fed into the prediction model of the category to obtain the spatial intensity of the target street.

[0009] (5) Output the spatial intensity zoning scheme of the target block to the digital sand table of the target block, integrate it into a 3D urban spatial intensity zoning holographic sand table, carry out multi-scheme display and interactive optimization of the spatial intensity of the block, and finally output the land use spatial intensity map and spatial intensity index table.

[0010] Furthermore, the implementation process of step (1) is as follows:

[0011] The target street dataset and the case street dataset are formed respectively, and the data are processed into street external feature indicators, street internal feature indicators, and spatial intensity indicators.

[0012] External characteristics of a block include its urban location, surrounding land use functions, and boundary road attributes; internal characteristics include its area, perimeter, shape, and urban land use functions; spatial intensity characteristics include building density, floor area ratio, and average building height.

[0013] Furthermore, the process of performing correlation analysis on the indicator library and removing highly correlated indicators in step (2) is as follows:

[0014] Perform correlation analysis on the indicator library, calculate the Pearson correlation coefficient, and remove indicators with a Pearson correlation coefficient greater than 0.7;

[0015] The selected case street dataset was standardized using the range standardization method, and the continuous variables were discretized using the one-hot discretization method.

[0016] 80% of the case study street datasets were randomly selected as the Bayesian network training set, and the remaining 20% ​​of the case study street datasets were selected as the Bayesian network test set.

[0017] Furthermore, the process of determining all categorical subsets of the case street dataset using a clustering algorithm in step (2) is as follows:

[0018] Based on the strength of external and internal characteristic indicators of a neighborhood, neighborhoods are classified according to the ratio of external to internal parameter indicators. These classifications can be divided into four types: strong-strong neighborhoods, strong-weak neighborhoods, weak-strong neighborhoods, and weak-weak neighborhoods.

[0019] The k-means clustering algorithm uses a value of 4 for k to divide the case street dataset into four sub-datasets corresponding to the four street classifications.

[0020] Furthermore, the implementation process of step (3) is as follows:

[0021] Using the external feature indicators, internal feature indicators, and spatial intensity indicators of each of the four types of case street subsets as node variables, a preliminary Bayesian network model structure for predicting the spatial intensity of each street is constructed. The optimal Bayesian network model structure for each street is obtained through structure learning. Based on the determined Bayesian network model structure, the conditional probability table of the spatial intensity indicators for each street is obtained through parameter learning of the Bayesian network model, which serves as the reasoning basis for the Bayesian network model for predicting the spatial intensity of each type of street, and finally, the prediction model for the spatial intensity of the four types of streets is obtained.

[0022] Further, step (4), which involves classifying the target street dataset based on the classification of the case street dataset using a classification algorithm to determine the category to which the target street belongs, specifically includes:

[0023] Using the k-nearest neighbor classification algorithm, the case street dataset is used as the training dataset for the k-nearest neighbor algorithm. The external and internal feature indicators of the target street dataset are fed into the k-nearest neighbor classification algorithm. The existing classification results of the case street are used as the classification labels, and the output is the classification of the target street.

[0024] Furthermore, the multi-scheme display and interactive optimization of street space intensity described in step (5) specifically includes:

[0025] The spatial intensity distribution scheme of the target block is output to the digital sand table of the target block and integrated into a 3D urban spatial intensity distribution holographic sand table for three-dimensional display. Using a 3D holographic projection device, the spatial intensity scheme of the block is selected, called, and modified through gesture commands. After adjusting and optimizing the spatial intensity distribution of the target block, it is output, and finally outputting a land use spatial intensity map in DWG format containing a land use spatial intensity index table. The 3D holographic projection device includes a VR panoramic display stand, VR glasses, virtual reality data gloves, and a high-performance graphics computer workstation.

[0026] Furthermore, the structure learning of the Bayesian network model is as follows:

[0027] Based on the existing case street subset dataset, the PC algorithm is used to analyze the dependencies between the variables of each node, thereby determining the dependency direction and the optimal Bayesian network model structure.

[0028] Furthermore, the parameters of the Bayesian network model are learned as follows:

[0029] Based on the determined Bayesian network model structure, the maximum likelihood estimation algorithm is used to estimate the conditional probabilities between the variables of each node in the Bayesian network model structure, and the probability distribution of each node variable is obtained.

[0030] Based on the same inventive concept, the present invention provides an urban spatial intensity zoning system based on Bayesian networks, comprising:

[0031] Data Input and Indicator Calculation Module: Acquires and processes street vector data and case data of the city where the target street is located, and calculates and processes the data to form an indicator library;

[0032] Bayesian network training set determination module: Based on the processed data, the training set indicators are screened through correlation analysis, and all classification sub-datasets of the case street dataset are determined through clustering algorithm;

[0033] Prediction model acquisition module: The Bayesian network is trained based on the classification subset dataset. After structure learning and parameter learning, different classification block spatial intensity prediction models are obtained.

[0034] Bayesian network module for predicting street spatial intensity: Based on the classification of the case street dataset, the k-nearest neighbor classification algorithm is used to determine the classification of the target street, and the target street dataset is fed into the prediction model of the classification to obtain the spatial intensity of the target street.

[0035] Data output and interactive display module: Outputs the spatial intensity distribution scheme of the target block to the digital sand table of the target block, integrates it into a 3D urban spatial intensity block-level distribution holographic sand table, displays and optimizes multiple schemes of block spatial intensity, and finally outputs land use spatial intensity map and spatial intensity index table.

[0036] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] 1. This invention unifies the format of the collected target street and the street vector data of the city where the target street is located, constructs a digital sand table of the target street, and forms a target street dataset and a case street dataset respectively. The data is processed to calculate the external characteristic indicators, internal characteristic indicators and spatial intensity indicators of the street, and summarized to form an indicator library, which ensures the standardization and accuracy of the data and maximizes the accuracy and efficiency of urban spatial intensity zoning.

[0038] 2. This invention filters out highly correlated indicators by performing correlation analysis on the indicator library; and determines all classification subsets of the case street dataset using the k-means clustering algorithm; the Bayesian network is trained on the classification subsets, and after structure learning and parameter learning, a prediction model is obtained. This model can comprehensively consider multiple factors, enabling the Bayesian network model for urban spatial intensity zoning to more accurately predict the spatial intensity of different types of streets, greatly improving the accuracy of the model prediction and the scientific nature of spatial intensity zoning.

[0039] 3. This invention uses the k-nearest neighbor classification algorithm to classify the dataset of the target street to determine the category to which the target street belongs; based on the category to which the target street belongs, the dataset of the target street is put into the prediction model of the category to obtain the spatial intensity of the target street, which can more scientifically and accurately predict the category to which the target street belongs, thus making the predicted spatial intensity of the target street more scientific and accurate.

[0040] 4. This invention outputs the spatial intensity zoning scheme of the target block to the digital sand table of the target block, and integrates it into a 3D urban spatial intensity zoning holographic sand table. It enables the display and interactive optimization of multiple spatial intensity schemes for the block, making the generated spatial intensity zoning scheme of the target block displayable, interactive, and optimizable. Planners can use 3D holographic projection equipment to select, call, modify, and optimize the spatial intensity scheme of the block in real time, which greatly improves the displayability and interactivity of the scheme. Attached Figure Description

[0041] Figure 1 This is a flowchart of a Bayesian network-based urban spatial intensity zoning method.

[0042] Figure 2 This is a map showing the current status of the land parcel before the implementation of this invention;

[0043] Figure 3 This is a diagram of the intensity zoning scheme generated by the present invention. Detailed Implementation

[0044] The present invention will now be described in further detail with reference to the accompanying drawings:

[0045] like Figure 1 As shown, this invention provides a method for urban spatial intensity zoning based on Bayesian networks. This method is intended for the intelligent generation of spatial intensity zoning schemes in the field of urban planning and design, and specifically includes the following steps:

[0046] S1. Standardize the format of the collected target street and the city where the target street is located, construct a digital sandbox of the target street, and form a target street dataset and a case street dataset. Calculate and process the data into external characteristic indicators, internal characteristic indicators, and spatial intensity indicators of the street, summarize them to form an indicator library, and generate a block status information map, such as... Figure 2 As shown.

[0047] The collected vector data of target blocks and the cities where the target blocks are located were standardized in format. The data was divided into target block datasets and case block datasets. The data was then processed into external feature indicators, internal feature indicators, and spatial intensity indicators. The external feature indicators include the urban location of the block, the land use function of the surrounding area, and the road attributes of the block boundary. The internal feature indicators include the block area, the block perimeter, the block shape, and the urban land use function attributes. The spatial intensity indicator includes building density, block floor area ratio, and average building height. The specific parameter meanings and calculation methods are shown in Table 1.

[0048] Table 1 Indicator Library

[0049]

[0050]

[0051] S2. By performing correlation analysis on the S1 indicator library, indicators with high correlation were screened out and removed; and by using a clustering algorithm, all sub-datasets of the case street dataset were determined.

[0052] Correlation analysis was performed on the indicator database, and Pearson correlation coefficients were calculated. Indicators with Pearson correlation coefficients greater than 0.7 were removed. The selected case study street datasets were standardized using range standardization, and continuous variables were discretized using one-hot discretization. 80% of the case study street datasets were randomly selected as the Bayesian network training set, and the remaining 20% ​​were used as the Bayesian network test set. Based on the strength of external and internal characteristic indicators of the streets, a street classification system based on external and internal parameter indicators was established, resulting in four street classifications: strong-strong, strong-weak, weak-strong, and weak-weak, as shown in Table 2.

[0053] Table 2 Street Classification Table

[0054]

[0055]

[0056] The k-means clustering algorithm was used, with k equal to 4, to divide the case street dataset into four sub-datasets corresponding to the four street classifications.

[0057] S3. The Bayesian network is trained using the classification subset dataset described in S2. After structure learning and parameter learning, different classification street spatial intensity prediction models are obtained.

[0058] Using the external feature indicators, internal feature indicators, and spatial intensity indicators of each of the four identified case street subsets as node variables, a preliminary Bayesian network model structure for predicting street spatial intensity is constructed. The optimal Bayesian network model structure for each type is obtained through structure learning. Based on the determined Bayesian network model structure, the conditional probability tables for street spatial intensity indicators are obtained through parameter learning, serving as the inference basis for the Bayesian network model for predicting the spatial intensity of each type of street, ultimately yielding prediction models for the spatial intensity of the four types of streets.

[0059] The structure learning of the Bayesian network model specifically includes: based on the existing case street sub-dataset, using the PC algorithm to analyze the dependencies between the variables of each node, thereby determining the dependency direction and determining the optimal Bayesian network model structure.

[0060] The parameter learning of the Bayesian network model specifically includes: based on the determined Bayesian network model structure, using the maximum likelihood estimation algorithm to estimate the conditional probabilities between the variables of each node in the Bayesian network model structure, and obtaining the probability distribution of each node variable.

[0061] S4. Predict the spatial intensity of the target block using the Bayesian network model trained in S3;

[0062] Using the k-nearest neighbor (kNN) classification algorithm, the case study street dataset is used as the training dataset for the kNN algorithm. The external and internal feature indices of the target street dataset are then fed into the kNN classification algorithm. The existing classification results of the case study streets are used as classification labels, and the output is the classification of the target street. Based on the classification of the target street, the target street dataset is fed into the prediction model of that classification to obtain the spatial intensity of the target street.

[0063] S5. Output the spatial intensity zoning scheme of the target block to the digital sand table of the target block, integrate it into a 3D urban spatial intensity zoning holographic sand table, display and optimize multiple schemes of block spatial intensity, and finally output the land use spatial intensity map and spatial intensity index table.

[0064] The spatial intensity distribution scheme of the target block is output to the target block's digital sand table and integrated into a 3D urban spatial intensity distribution holographic sand table for three-dimensional display. Using 3D holographic projection equipment, users can select, recall, and modify the block's spatial intensity scheme via gesture commands. After adjusting and optimizing the spatial intensity distribution of the target block, the scheme is output, ultimately producing a DWG format map of the intensity zoning scheme containing a land use spatial intensity index table, as shown below. Figure 3 As shown. The 3D holographic projection equipment includes a VR panoramic display stand, VR glasses, virtual reality data gloves, and a high-performance graphics computer workstation.

[0065] This invention also provides a Bayesian network-based urban spatial intensity zoning system, comprising:

[0066] Data Input and Indicator Calculation Module: Acquires and processes street vector data and case data of the city where the target street is located, and calculates and processes the data to form an indicator library;

[0067] Bayesian network training set determination module: Based on the processed data, the training set indicators are screened through correlation analysis, and all classification sub-datasets of the case street dataset are determined through clustering algorithm;

[0068] Prediction model acquisition module: The Bayesian network is trained based on the classification subset dataset. After structure learning and parameter learning, different classification block spatial intensity prediction models are obtained.

[0069] Bayesian network module for predicting street spatial intensity: Based on the classification of the case street dataset, the k-nearest neighbor classification algorithm is used to determine the classification of the target street, and the target street dataset is fed into the prediction model of the classification to obtain the spatial intensity of the target street.

[0070] Data output and interactive display module: Outputs the spatial intensity distribution scheme of the target block to the digital sand table of the target block, integrates it into a 3D urban spatial intensity block-level distribution holographic sand table, displays and optimizes multiple schemes of block spatial intensity, and finally outputs land use spatial intensity map and spatial intensity index table.

Claims

1. A method for urban spatial intensity zoning based on Bayesian networks, characterized in that, Includes the following steps: (1) The target street and the street vector data of the city where the target street are located are processed in a unified format to construct a digital sand table of the target street, forming a target street dataset and a case street dataset respectively. The data are processed into external characteristic indicators, internal characteristic indicators and spatial intensity indicators of the street, and summarized to form an indicator library. (2) By performing correlation analysis on the indicator library, indicators with high correlation were screened out and removed; and by using clustering algorithms, all sub-datasets of the case street dataset were determined; (3) The Bayesian network is trained using the aforementioned sub-datasets, and after structure learning and parameter learning, different classification street spatial intensity prediction models are obtained. (4) Based on the classification of the case street dataset, the dataset of the street to be predicted is classified using a classification algorithm to determine the category to which the target street belongs; Based on the category to which the target street belongs, the target street dataset is fed into the prediction model of the category to obtain the spatial intensity of the target street. (5) Output the spatial intensity zoning scheme of the target block to the digital sand table of the target block, integrate it into a 3D urban spatial intensity zoning holographic sand table, display and optimize multiple schemes of block spatial intensity, and finally output the land use spatial intensity map and spatial intensity index table. The process of determining all categorical subsets of the case street dataset using clustering algorithms in step (2) is as follows: Based on the strength of external and internal characteristic indicators of a neighborhood, neighborhoods are classified according to the ratio of external to internal parameter indicators. These classifications can be divided into four types: strong-strong neighborhoods, strong-weak neighborhoods, weak-strong neighborhoods, and weak-weak neighborhoods. The k-means clustering algorithm takes a value of 4 for k and divides the case neighborhood dataset into four sub-datasets corresponding to the four neighborhood categories. The implementation process of step (3) is as follows: Using the external feature indicators, internal feature indicators, and spatial intensity indicators of the four types of case street subsets as node variables, a preliminary Bayesian network model structure for predicting the spatial intensity of each street is constructed. The optimal Bayesian network model structure for each type is obtained through structure learning. Based on the determined Bayesian network model structure, the conditional probability table of the spatial intensity indicators for each type of street is obtained through parameter learning, which serves as the reasoning basis for the Bayesian network model for predicting the spatial intensity of each type of street. Finally, the prediction models for the spatial intensity of the four types of streets are obtained.

2. The urban spatial intensity zoning method based on Bayesian networks according to claim 1, characterized in that, The implementation process of step (1) is as follows: The target street dataset and the case street dataset are formed respectively, and the data are processed into street external feature indicators, street internal feature indicators, and spatial intensity indicators. External characteristics of a block include its urban location, surrounding land use functions, and boundary road attributes; internal characteristics include its area, perimeter, shape, and urban land use functions; spatial intensity characteristics include building density, floor area ratio, and average building height.

3. The urban spatial intensity zoning method based on Bayesian networks according to claim 1, characterized in that, The process of performing correlation analysis on the indicator library in step (2) to filter out indicators with high correlation and then removing them is as follows: Perform correlation analysis on the indicator library, calculate the Pearson correlation coefficient, and remove indicators with a Pearson correlation coefficient greater than 0.7; The selected case street dataset was standardized using the range standardization method, and the continuous variables were discretized using the one-hot discretization method. 80% of the case study street datasets were randomly selected as the Bayesian network training set, and the remaining 20% ​​of the case study street datasets were selected as the Bayesian network test set.

4. The urban spatial intensity zoning method based on Bayesian networks according to claim 1, characterized in that, Step (4) involves classifying the target street dataset based on the classification of the case street dataset using a classification algorithm to determine the category to which the target street belongs. This specifically includes: use The nearest neighbor classification algorithm uses the case block dataset as... The training dataset for the nearest neighbor algorithm consists of the external and internal feature metrics of the target street dataset. In the nearest neighbor classification algorithm, the classification results of existing case blocks are used as classification labels, and the output is the classification of the target block.

5. The urban spatial intensity zoning method based on Bayesian networks according to claim 1, characterized in that, Step (5) of the multi-scheme display and interactive optimization of street space intensity specifically includes: The spatial intensity distribution scheme of the target block is output to the digital sand table of the target block and integrated into a 3D urban spatial intensity distribution holographic sand table for three-dimensional display. Using a 3D holographic projection device, the spatial intensity scheme of the block is selected, called, and modified through gesture commands. After adjusting and optimizing the spatial intensity distribution of the target block, it is output, and finally outputting a land use spatial intensity map in DWG format containing a land use spatial intensity index table. The 3D holographic projection device includes a VR panoramic display stand, VR glasses, virtual reality data gloves, and a high-performance graphics computer workstation.

6. The urban spatial intensity zoning method based on Bayesian networks according to claim 1, characterized in that, The structure learning of the Bayesian network model is as follows: Based on the existing case street subset dataset, the PC algorithm is used to analyze the dependencies between the variables of each node, thereby determining the dependency direction and the optimal Bayesian network model structure.

7. The urban spatial intensity zoning method based on Bayesian networks according to claim 1, characterized in that, The parameters of the Bayesian network model are learned as follows: Based on the determined Bayesian network model structure, the maximum likelihood estimation algorithm is used to estimate the conditional probabilities between the variables of each node in the Bayesian network model structure, and the probability distribution of each node variable is obtained.

8. A Bayesian network-based urban spatial intensity zoning system employing the method described in any one of claims 1 to 7, characterized in that, The system includes: Data Input and Indicator Calculation Module: Acquires and processes street vector data and case data of the city where the target street is located, and calculates and processes the data to form an indicator library; Bayesian Network Training Set Determination Module: Based on the processed data, training set indicators are selected through correlation analysis, and all classification sub-datasets of the case street dataset are determined through clustering algorithms; Prediction model acquisition module: The Bayesian network is trained based on the classification subset dataset. After structure learning and parameter learning, different classification block spatial intensity prediction models are obtained. Bayesian network module for predicting street spatial intensity: Based on the classification of the case street dataset, through... The nearest neighbor classification algorithm determines the category to which the target block belongs, and the target block dataset is fed into the prediction model of the category to obtain the spatial intensity of the target block; Data output and interactive display module: Outputs the spatial intensity distribution scheme of the target block to the digital sand table of the target block, integrates it into a 3D urban spatial intensity block-level distribution holographic sand table, displays and optimizes multiple schemes of block spatial intensity, and finally outputs land use spatial intensity map and spatial intensity index table.

Citation Information

Patent Citations

  • Urban block vitality prediction method, storage medium and terminal

    CN114662774A

  • Urban vitality quantitative evaluation method integrating multi-source geographic big data

    CN115146990A