Land space element-based traffic planning space data processing method and system

Abstract

The application provides a traffic planning space data processing method and system based on land space elements, wherein the method comprises the following steps: establishing a three-dimensional correlation index table of plots-road sections-time; acquiring the land use attribute of the plot associated with each road section through the three-dimensional correlation index table, and desensitizing real-time dynamic traffic flow data based on the land use attribute; performing steady-state filtering prediction on the desensitized real-time dynamic traffic flow data according to the characteristics associated in the three-dimensional correlation index table; periodically scanning the three-dimensional correlation index table, and outputting adjustment suggestions for plots and road sections contained in the plots based on the predicted traffic flow data. Through the method and the corresponding system, the prediction of future traffic flow is realized, and is reversely associated with plot development to form a dynamic feedback closed loop.

Description

Technical Field

[0001] This invention proposes a method and system for processing spatial data of transportation planning based on land spatial elements, which relates to the field of data processing technology. Background Technology

[0002] With the deep integration of smart cities and land spatial planning, utilizing dynamic traffic big data to assist planning decisions has become a mainstream trend. However, existing spatial data processing methods for traffic planning still have the following limitations in practical applications: 1. Traditional traffic flow monitoring is isolated from land ownership and land use attributes in national land space, lacking a unified spatiotemporal correlation index, which makes it impossible to analyze the generation mechanism and evolution pattern of traffic flow from the source (land parcel attributes).

[0003] 2. Existing desensitization techniques often destroy the original statistical distribution characteristics of traffic flow when protecting privacy, resulting in a significant decrease in the accuracy of the desensitized data in subsequent modeling, making it difficult to balance privacy security with the fidelity requirements of planning research.

[0004] 3. Current deep learning prediction models are mostly data-driven, ignoring rigid physical constraints such as building scale limitations in land spatial planning. This can lead to prediction results that, in extreme cases, exceed the actual carrying capacity of land parcels and roads, lacking physical interpretability.

[0005] 4. Existing planning adjustments are often passive repairs based on current congestion, lacking a dynamic feedback loop based on future traffic forecasts and retrospective analysis of land development intensity, making it difficult to achieve forward-looking management and resource balancing for land to be developed.

[0006] Therefore, how to construct a dynamic transportation planning data processing solution that deeply integrates land and space elements and balances fidelity, anonymization, and physical constraints is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] This invention provides a method and system for processing spatial data of transportation planning based on land spatial elements, in order to solve the problems mentioned above: This invention proposes a method for processing spatial data of transportation planning based on land spatial elements, the method comprising: Establish a three-dimensional correlation index table of land parcels, road sections, and time; The land use attributes of the plots associated with each road segment are obtained through the three-dimensional association index table, and the real-time dynamic traffic flow data is anonymized based on the land use attributes. Steady-state filtering prediction is performed on the anonymized real-time dynamic traffic flow data based on the associated features in the three-dimensional association index table. The three-dimensional correlation index table is scanned periodically, and adjustment suggestions for the land parcel and the road segments contained therein are output based on the predicted traffic flow data.

[0008] Furthermore, a three-dimensional correlation index table is established for land parcels, road sections, and time periods, including: The land space is divided into several traffic zones, and the unique identification code of each traffic zone and its corresponding land use attributes are extracted. The land use attributes include building scale constraint indicators. Extract the road network topology covering the plots, establish unique identification codes for road segments, and map the unique identification codes for road segments to the corresponding unique identification codes for plots based on spatial inclusion relationships; The timeline is segmented to form a time-slice sequence. Dynamic traffic flow data is linked with the corresponding time slices to construct a three-dimensional association index table of unique land parcel identification code - unique road segment identification code - time slice.

[0009] Furthermore, the real-time dynamic traffic flow data is anonymized based on the aforementioned land use attributes, including: Define the original traffic flow dataset to be processed as follows: ,in, This represents the original real-time dynamic traffic flow data of the i-th sampling point in the sequence; Through mapping function The original real-time dynamic traffic flow data is transformed to calculate the anonymized values. Where T represents the adjustment period, Indicates the initial phase shift. Indicates land use sensitivity. Represents the perturbation gain function; The probability distribution of the dataset to be de-identified is calculated using a Gaussian kernel function to construct the original probability distribution space. The specific formula is as follows: in, This indicates that the values ​​in the original sequence are The probability of a data point appearing in the overall distribution. Represents the original traffic flow dataset The mean, This represents the standard deviation of the original traffic flow dataset X; The anonymized traffic flow dataset is mapped back to the original probability distribution space using a histogram specification algorithm to obtain the final anonymized traffic flow dataset Y.

[0010] Furthermore, steady-state filtering prediction is performed on the anonymized real-time dynamic traffic flow data based on the associated features in the three-dimensional association index table, including: Based on the time slice of the three-dimensional association index table, extract the anonymized historical dynamic traffic flow data of the target road segment and its associated plots, and simultaneously retrieve the corresponding building scale constraint index in the index table; The coefficient of variation of the anonymized historical dynamic traffic flow sequence is calculated, and an adaptive weighting operator is constructed based on the coefficient of variation. The weighting operator is negatively correlated with the coefficient of variation, and the building scale constraint index is used as the upper limit threshold for predicting traffic flow data. The historical dynamic traffic flow data is smoothed by the adaptive weighting operator, and the predicted traffic flow data of the road segment in the future time slice is obtained based on the smoothed historical dynamic traffic flow data. The predicted traffic flow data is written into the prediction field of the three-dimensional association index table.

[0011] Furthermore, based on the smoothed historical dynamic traffic flow data, predicted traffic flow data for this road segment in future time slices is obtained, including: Based on the preset historical backtracking window, the historical dynamic traffic flow data is normalized, and the coefficient of variation and the spatial carrying capacity feature components obtained by mapping the building scale constraint index are broadcast and aligned along the time axis. Tensor splicing is performed on the feature dimension to construct a multi-dimensional spatiotemporal feature matrix, which serves as the input of the deep learning model. A deep learning model based on recurrent neural units is constructed. The model integrates an adaptive gating operator layer, and the adaptive weighting operator is used as the control signal for this layer. Define the loss function: Where MSE represents the mean squared error between the predicted value and the actual value. This represents the upper limit threshold corresponding to this road segment, and λ represents the constraint penalty coefficient. y represents the predicted value, and y represents the actual value; The prediction error for each iteration is calculated using the composite loss function, and the physical constraint penalty term is mapped to the gradient correction value of the neuron weights using the error backpropagation algorithm to optimize the parameters of the deep learning model. The multidimensional spatiotemporal feature matrix is ​​input into the trained deep learning model, which outputs predicted traffic flow data for future time slices.

[0012] Furthermore, the periodic scanning of the three-dimensional correlation index table, based on predicted traffic flow data, outputs adjustment suggestions for land parcels and road segments, including: The three-dimensional correlation index table is periodically scanned to calculate the saturation (proportion) of the predicted traffic flow data of each road segment in each time slice to the upper limit threshold corresponding to that road segment. When the saturation exceeds a preset threshold, the unique identification code of the associated land parcel is retrieved in reverse through the three-dimensional association index table to obtain the planning and construction status of the land parcel. If the associated land parcel is in an undeveloped state, a planning adjustment suggestion is output, which includes: reducing the plot ratio of the land parcel and increasing the density of branch roads in the road segments contained in the land parcel.

[0013] The present invention proposes a spatial data processing system for transportation planning based on land spatial elements, the system comprising: The index table creation module is used to create a three-dimensional association index table of land parcels, road sections, and time. The data anonymization module is used to obtain the land use attributes of the plots associated with each road segment through the three-dimensional association index table, and anonymize the real-time dynamic traffic flow data based on the land use attributes. The prediction module is used to perform steady-state filtering and prediction on the desensitized real-time dynamic traffic flow data based on the features associated in the three-dimensional association index table. The output adjustment suggestion module is used to periodically scan the three-dimensional association index table and output adjustment suggestions for the land parcel and the road segments contained in the land parcel based on the predicted traffic flow data.

[0014] Furthermore, the index table creation module includes: The regional division module is used to divide the national land space into several traffic zones, extract the unique identification code of each traffic zone and its corresponding land use attributes, including building scale constraint indicators. The spatial mapping module is used to extract the road network topology covering the plot, establish a unique identification code for each road segment, and map the unique identification code of each road segment to the corresponding unique identification code of the plot according to the spatial inclusion relationship. The associated traffic flow data module is used to segment the time axis to form a time slice sequence, link dynamic traffic flow data with the corresponding time slices, and construct a three-dimensional association index table of unique land parcel identification code - unique road segment identification code - time slice.

[0015] Furthermore, the data desensitization module includes: The "Get Traffic Flow Dataset" module is used to define the original traffic flow dataset to be processed. ,in, This represents the original real-time dynamic traffic flow data of the i-th sampling point in the sequence; Transformation module, used to map functions The original real-time dynamic traffic flow data is transformed to calculate the anonymized values. Where T represents the adjustment period, Indicates the initial phase shift. Indicates land use sensitivity. Represents the perturbation gain function; The probability distribution calculation module is used to calculate the probability distribution of the dataset to be de-identified using a Gaussian kernel function, and to construct the original probability distribution space. The specific implementation formula is as follows: in, This indicates that the values ​​in the original sequence are The probability of a data point appearing in the overall distribution. Represents the original traffic flow dataset The mean, This represents the standard deviation of the original traffic flow dataset X; The data mapping module is used to map the anonymized traffic flow dataset back to the original probability distribution space using a histogram specification algorithm, thereby obtaining the final anonymized traffic flow dataset Y.

[0016] Furthermore, the prediction module includes: An input data module is constructed to normalize historical dynamic traffic flow data based on a preset historical backtracking window. Simultaneously, the coefficient of variation and the spatial carrying capacity feature components obtained by mapping the building scale constraint index are broadcast and aligned along the time axis. Tensor splicing is performed on the feature dimension to construct a multi-dimensional spatiotemporal feature matrix, which serves as the input to the deep learning model. A deep learning model building module is used to build a deep learning model based on recurrent neural units. The model integrates an adaptive gating operator layer, and the adaptive weighting operator is used as the control signal for this layer. Define a function module for defining loss functions: Where MSE represents the mean squared error between the predicted value and the actual value. This represents the upper limit threshold corresponding to this road segment, and λ represents the constraint penalty coefficient. y represents the predicted value, and y represents the actual value; The training model module is used to calculate the prediction error for each iteration using the composite loss function, map the physical constraint penalty term to the gradient correction value of the neuron weights using the error backpropagation algorithm, and optimize the parameters of the deep learning model. The prediction data acquisition module is used to input the multidimensional spatiotemporal feature matrix into the trained deep learning model and output the predicted traffic flow data for future time slices.

[0017] The beneficial effects of this invention are as follows: Unlike traditional purely data-driven models, this invention introduces land use and spatial constraint features through a three-dimensional index table, solving the numerical drift problem that easily occurs in long-sequence predictions in deep learning models, making the prediction results more consistent with the objective physical laws of urban operation. By using land use attributes for desensitization, sensitive individual trajectory information is hidden while group traffic behavior characteristics are preserved through semantic association, achieving a balance between security and practicality in big data applications. Through early intervention in land parcels awaiting development, the system can identify potential supply and demand imbalances before project construction, transforming passive road repair into proactive optimization, and preventing urban congestion at its root. The design of the three-dimensional relational index table transforms complex cross-database relational queries into pre-defined index matching, greatly shortening the response time from problem discovery (saturation warning) to cause location (related land parcels), supporting real-time monitoring and rapid decision-making for large-scale urban road networks. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a spatial data processing method for transportation planning based on land spatial elements, as described in this invention. Detailed Implementation

[0019] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0020] Numerous specific details are set forth in the following description to provide a thorough understanding of the invention. The described embodiments are only a part of, and not all, of the embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0022] One embodiment of the present invention provides a method for processing spatial data of transportation planning based on land spatial elements, the method comprising: Establish a three-dimensional correlation index table of land parcels, road sections, and time; The land use attributes of the plots associated with each road segment are obtained through the three-dimensional association index table, and the real-time dynamic traffic flow data is anonymized based on the land use attributes. Steady-state filtering prediction is performed on the anonymized real-time dynamic traffic flow data based on the associated features in the three-dimensional association index table. The three-dimensional correlation index table is scanned periodically, and adjustment suggestions for the land parcel and the road segments contained therein are output based on the predicted traffic flow data.

[0023] The working principle and effects of the above technical solution are as follows: By establishing a three-dimensional correlation index table of land parcels, road segments, and time, the originally fragmented geographic space (land parcels), physical network (road segments), and dynamic evolution (time) are unified under the same coordinate system. This makes each traffic data point no longer an isolated value, but a structured information with a clear source of generation and evolution path. The system uses land use attributes (such as residential, commercial, and industrial) in the index table to desensitize and extract features from the traffic data. The principle is that different land use properties determine the background characteristics of traffic flow (such as tidal and explosive nature). Through this correlation, the system can preserve the traffic attraction patterns that are crucial for prediction while protecting privacy. In the prediction stage, the system does not simply perform mathematical extrapolation, but injects the spatial carrying capacity characteristics (such as road network density and planning limits) in the three-dimensional index table as external constraints into the filtering algorithm. This dual-driven mode of data + physical rules ensures that the prediction results can maintain steady-state convergence even in complex and fluctuating environments. When potential congestion (saturation exceeding limits) is detected in the forecast, the reverse retrieval capability of the copper pot 3D index table directly maps traffic pressure back to its source plot. By comparing the development status of the plot, the dynamic supply and demand balancing logic is triggered (i.e., adjusting the plot ratio or adding branch roads), thereby realizing cross-dimensional feedback from end-of-pipe governance to source-level planning intervention.

[0024] One embodiment of the present invention establishes a three-dimensional association index table of land parcel-road segment-time, including: The land space is divided into several traffic zones, and the unique identification code of each traffic zone and its corresponding land use attributes are extracted. The land use attributes include building scale constraint indicators. Extract the road network topology covering the plots, establish unique identification codes for road segments, and map the unique identification codes for road segments to the corresponding unique identification codes for plots based on spatial inclusion relationships; The timeline is segmented to form a time-slice sequence. Dynamic traffic flow data is linked with the corresponding time slices to construct a three-dimensional association index table of unique land parcel identification code - unique road segment identification code - time slice.

[0025] One embodiment of the present invention involves desensitizing real-time dynamic traffic flow data based on the land use attributes, including: Define the original traffic flow dataset to be processed as follows: ,in, This represents the original real-time dynamic traffic flow data of the i-th sampling point in the sequence; Through mapping function The original real-time dynamic traffic flow data is transformed to calculate the anonymized values. Where T represents the adjustment period, Indicates the initial phase shift. Indicates land use sensitivity. Represents the perturbation gain function; The probability distribution of the dataset to be de-identified is calculated using a Gaussian kernel function to construct the original probability distribution space. The specific formula is as follows: in, This indicates that the values ​​in the original sequence are The probability of a data point appearing in the overall distribution. Represents the original traffic flow dataset The mean, This represents the standard deviation of the original traffic flow dataset X; The anonymized traffic flow dataset is mapped back to the original probability distribution space using a histogram specification algorithm to obtain the final anonymized traffic flow dataset Y.

[0026] The working principle and effect of the above technical solution are as follows: the original traffic flow data is transformed through a mapping function. Specifically, the sine function introduces periodic fluctuations to prevent the data from being manipulated by simple linear regression. This ensures that different road segments have different disturbance directions and amplitudes even at the same time, eliminating the possibility of spatial correlation attacks. The disturbance gain σ(α) is dynamically adjusted based on the land parcel's attributes (e.g., military zone, core government zone, or ordinary residential zone). Higher sensitivity results in larger disturbance amplitudes, achieving differentiated desensitization. A Gaussian kernel function is used to probabilistically model the original sequence X. The purpose of this step is to extract the characteristics of the traffic distribution of the road segment, namely the mean μ, standard deviation σ, and the shape of the distribution curve. This defines the true traffic fluctuation pattern of the road segment in the physical world. Although the transformed data achieves desensitization, its statistical distribution may deviate from the true physical laws (e.g., mean shift or distribution distortion), which will affect the accuracy of subsequent prediction algorithms. A histogram specification algorithm is used to force the desensitized dataset back to the original probability distribution space. The principle is to... The data is rearranged according to the cumulative distribution function so that it numerically matches the statistical characteristics of the original data, thus generating the final result.

[0027] in, This represents the unique identifier for the road segment. H(·) represents a preset hash function (such as the low-order bits of CRC32 or MD5, or a simple prime number mapping function) used to map the string or long integer ID to uniformly distributed integers. M represents the preset number of phase discretization layers (a value of [value missing] is recommended). to ), used to control the fineness of phase distribution.

[0028] Wherein, α represents the land use sensitivity coefficient of the associated land parcels in the current road segment, with a value range of [0,1]. γ represents the preset maximum sensitivity (usually 1), and γ represents the global perturbation scaling factor, which controls the maximum deviation of the desensitization. In order to ensure that the data still has usability after histogram specification, γ∈[0.05,0.2].

[0029] One embodiment of the present invention involves performing steady-state filtering prediction on desensitized real-time dynamic traffic flow data based on features associated in a three-dimensional association index table, including: Based on the time slice of the three-dimensional association index table, extract the anonymized historical dynamic traffic flow data of the target road segment and its associated plots, and simultaneously retrieve the corresponding building scale constraint index in the index table; The coefficient of variation of the anonymized historical dynamic traffic flow sequence is calculated, and an adaptive weighting operator is constructed based on the coefficient of variation. The weighting operator is negatively correlated with the coefficient of variation, and the building scale constraint index is used as the upper limit threshold for predicting traffic flow data. The historical dynamic traffic flow data is smoothed by the adaptive weighting operator, and the predicted traffic flow data of the road segment in the future time slice is obtained based on the smoothed historical dynamic traffic flow data. The predicted traffic flow data is written into the prediction field of the three-dimensional association index table.

[0030] The working principle and effects of the above technical solution are as follows: One embodiment of the present invention involves obtaining predicted traffic flow data for a road segment in future time slices based on smoothed historical dynamic traffic flow data, including: Based on the preset historical backtracking window, the historical dynamic traffic flow data is normalized, and the coefficient of variation and the spatial carrying capacity feature components obtained by mapping the building scale constraint index are broadcast and aligned along the time axis. Tensor splicing is performed on the feature dimension to construct a multi-dimensional spatiotemporal feature matrix, which serves as the input of the deep learning model. A deep learning model based on recurrent neural units is constructed. The model integrates an adaptive gating operator layer. The adaptive weighting operator is used as the control signal of this layer to dynamically adjust the forgetting ratio of the hidden layer state to historical fluctuation terms and the response tendency to the planning boundary terms.

[0031] Define the loss function: Where MSE represents the mean squared error between the predicted value and the actual value. This represents the upper limit threshold corresponding to this road segment, and λ represents the constraint penalty coefficient. y represents the predicted value, and y represents the actual value; The prediction error for each iteration is calculated using the composite loss function, and the physical constraint penalty term is mapped to the gradient correction value of the neuron weights using the error backpropagation algorithm to optimize the parameters of the deep learning model. The multidimensional spatiotemporal feature matrix is ​​input into the trained deep learning model, which outputs predicted traffic flow data for future time slices.

[0032] The working principle and effects of the above technical solution are as follows: In one embodiment of the present invention, the step of periodically scanning the three-dimensional correlation index table and outputting adjustment suggestions for land parcels and road segments based on predicted traffic flow data includes: The three-dimensional correlation index table is periodically scanned to calculate the saturation (proportion) of the predicted traffic flow data of each road segment in each time slice to the upper limit threshold corresponding to that road segment. When the saturation exceeds a preset threshold, the unique identification code of the associated land parcel is retrieved in reverse through the three-dimensional association index table to obtain the planning and construction status of the land parcel. If the associated land parcel is in an undeveloped state, a planning adjustment suggestion is output, which includes: reducing the plot ratio of the land parcel and increasing the density of branch roads in the road segments contained in the land parcel.

[0033] The working principle and effects of the above technical solution are as follows: One embodiment of the present invention provides a spatial data processing system for transportation planning based on land spatial elements, the system comprising: The index table creation module is used to create a three-dimensional association index table of land parcels, road sections, and time. The data anonymization module is used to obtain the land use attributes of the plots associated with each road segment through the three-dimensional association index table, and anonymize the real-time dynamic traffic flow data based on the land use attributes. The prediction module is used to perform steady-state filtering and prediction on the desensitized real-time dynamic traffic flow data based on the features associated in the three-dimensional association index table. The output adjustment suggestion module is used to periodically scan the three-dimensional association index table and output adjustment suggestions for the land parcel and the road segments contained in the land parcel based on the predicted traffic flow data.

[0034] In one embodiment of the present invention, the index table creation module includes: The regional division module is used to divide the national land space into several traffic zones, extract the unique identification code of each traffic zone and its corresponding land use attributes, including building scale constraint indicators. The spatial mapping module is used to extract the road network topology covering the plot, establish a unique identification code for each road segment, and map the unique identification code of each road segment to the corresponding unique identification code of the plot according to the spatial inclusion relationship. The associated traffic flow data module is used to segment the time axis to form a time slice sequence, link dynamic traffic flow data with the corresponding time slices, and construct a three-dimensional association index table of unique land parcel identification code - unique road segment identification code - time slice.

[0035] In one embodiment of the present invention, the data anonymization module includes: The "Get Traffic Flow Dataset" module is used to define the original traffic flow dataset to be processed. ,in, This represents the original real-time dynamic traffic flow data of the i-th sampling point in the sequence; Transformation module, used to map functions The original real-time dynamic traffic flow data is transformed to calculate the anonymized values. Where T represents the adjustment period, Indicates the initial phase shift. Indicates land use sensitivity. Represents the perturbation gain function; The probability distribution calculation module is used to calculate the probability distribution of the dataset to be de-identified using a Gaussian kernel function, and to construct the original probability distribution space. The specific implementation formula is as follows: in, This indicates that the values ​​in the original sequence are The probability of a data point appearing in the overall distribution. Represents the original traffic flow dataset The mean, This represents the standard deviation of the original traffic flow dataset X; The data mapping module is used to map the anonymized traffic flow dataset back to the original probability distribution space using a histogram specification algorithm, thereby obtaining the final anonymized traffic flow dataset Y.

[0036] In one embodiment of the present invention, the prediction module includes: An input data module is constructed to normalize historical dynamic traffic flow data based on a preset historical backtracking window. Simultaneously, the coefficient of variation and the spatial carrying capacity feature components obtained by mapping the building scale constraint index are broadcast and aligned along the time axis. Tensor splicing is performed on the feature dimension to construct a multi-dimensional spatiotemporal feature matrix, which serves as the input to the deep learning model. A deep learning model building module is used to build a deep learning model based on recurrent neural units. The model integrates an adaptive gating operator layer, and the adaptive weighting operator is used as the control signal for this layer. Define a function module for defining loss functions: Where MSE represents the mean squared error between the predicted value and the actual value. This represents the upper limit threshold corresponding to this road segment, and λ represents the constraint penalty coefficient. y represents the predicted value, and y represents the actual value; The training model module is used to calculate the prediction error for each iteration using the composite loss function, map the physical constraint penalty term to the gradient correction value of the neuron weights using the error backpropagation algorithm, and optimize the parameters of the deep learning model. The prediction data acquisition module is used to input the multidimensional spatiotemporal feature matrix into the trained deep learning model and output the predicted traffic flow data for future time slices.

[0037] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for processing spatial data of transportation planning based on land spatial elements, characterized in that, The method includes: Establish a three-dimensional correlation index table of land parcels, road sections, and time; The land use attributes of the plots associated with each road segment are obtained through the three-dimensional association index table, and the real-time dynamic traffic flow data is anonymized based on the land use attributes. Steady-state filtering prediction is performed on the anonymized real-time dynamic traffic flow data based on the associated features in the three-dimensional association index table. The three-dimensional correlation index table is scanned periodically, and adjustment suggestions for the land parcel and the road segments contained therein are output based on the predicted traffic flow data.

2. The method for processing spatial data of transportation planning based on land spatial elements according to claim 1, characterized in that, Establish a three-dimensional correlation index table of land parcels, road sections, and time, including: The land space is divided into several traffic zones, and the unique identification code of each traffic zone and its corresponding land use attributes are extracted. The land use attributes include building scale constraint indicators. Extract the road network topology covering the plots, establish unique identification codes for road segments, and map the unique identification codes for road segments to the corresponding unique identification codes for plots based on spatial inclusion relationships; The timeline is segmented to form a time-slice sequence. Dynamic traffic flow data is linked with the corresponding time slices to construct a three-dimensional association index table of unique land parcel identification code - unique road segment identification code - time slice.

3. The method for processing spatial data of transportation planning based on land spatial elements according to claim 1, characterized in that, De-identifying real-time dynamic traffic flow data based on the aforementioned land use attributes includes: Define the original traffic flow dataset to be processed as follows: ,in, This represents the original real-time dynamic traffic flow data of the i-th sampling point in the sequence; Through mapping function The original real-time dynamic traffic flow data is transformed to calculate the anonymized values. Where T represents the adjustment period, Indicates the initial phase shift. Indicates land use sensitivity. Represents the perturbation gain function; The probability distribution of the dataset to be de-identified is calculated using a Gaussian kernel function to construct the original probability distribution space. The specific formula is as follows: in, This indicates that the values ​​in the original sequence are The probability of a data point appearing in the overall distribution. Represents the original traffic flow dataset The mean, This represents the standard deviation of the original traffic flow dataset X; The anonymized traffic flow dataset is mapped back to the original probability distribution space using a histogram specification algorithm to obtain the final anonymized traffic flow dataset Y.

4. The method for processing spatial data of transportation planning based on land spatial elements according to claim 1, characterized in that, Steady-state filtering prediction is performed on the anonymized real-time dynamic traffic flow data based on the associated features in the three-dimensional association index table, including: Based on the time slice of the three-dimensional association index table, extract the anonymized historical dynamic traffic flow data of the target road segment and its associated plots, and simultaneously retrieve the corresponding building scale constraint index in the index table; The coefficient of variation of the anonymized historical dynamic traffic flow sequence is calculated, and an adaptive weighting operator is constructed based on the coefficient of variation. The weighting operator is negatively correlated with the coefficient of variation, and the building scale constraint index is used as the upper limit threshold for predicting traffic flow data. The historical dynamic traffic flow data is smoothed by the adaptive weighting operator, and the predicted traffic flow data of the road segment in the future time slice is obtained based on the smoothed historical dynamic traffic flow data. The predicted traffic flow data is written into the prediction field of the three-dimensional association index table.

5. The method for processing spatial data of transportation planning based on land spatial elements according to claim 4, characterized in that, Based on smoothed historical dynamic traffic flow data, predicted traffic flow data for this road segment in future time slices is obtained, including: Based on the preset historical backtracking window, the historical dynamic traffic flow data is normalized, and the coefficient of variation and the spatial carrying capacity feature components obtained by mapping the building scale constraint index are broadcast and aligned along the time axis. Tensor splicing is performed on the feature dimension to construct a multi-dimensional spatiotemporal feature matrix, which serves as the input of the deep learning model. A deep learning model based on recurrent neural units is constructed. The model integrates an adaptive gating operator layer, and the adaptive weighting operator is used as the control signal for this layer. Define the loss function: Where MSE represents the mean squared error between the predicted value and the actual value. This represents the upper limit threshold corresponding to this road segment, and λ represents the constraint penalty coefficient. y represents the predicted value, and y represents the actual value; The prediction error for each iteration is calculated using the composite loss function, and the physical constraint penalty term is mapped to the gradient correction value of the neuron weights using the error backpropagation algorithm to optimize the parameters of the deep learning model. The multidimensional spatiotemporal feature matrix is ​​input into the trained deep learning model, which outputs predicted traffic flow data for future time slices.

6. The method for processing spatial data of transportation planning based on land spatial elements according to claim 1, characterized in that, The periodic scanning of the three-dimensional correlation index table, based on predicted traffic flow data, outputs adjustment suggestions for land parcels and road segments, including: The three-dimensional correlation index table is periodically scanned to calculate the saturation of the predicted traffic flow data of each road segment in each time slice and the corresponding upper limit threshold of that road segment; When the saturation exceeds a preset threshold, the unique identification code of the associated land parcel is retrieved in reverse through the three-dimensional association index table to obtain the planning and construction status of the land parcel. If the associated land parcel is in an undeveloped state, a planning adjustment suggestion is output, which includes: reducing the plot ratio of the land parcel and increasing the density of branch roads in the road segments contained in the land parcel.

7. A spatial data processing system for transportation planning based on land and space elements, characterized in that: The system includes: The index table creation module is used to create a three-dimensional association index table of land parcels, road sections, and time. The data anonymization module is used to obtain the land use attributes of the plots associated with each road segment through the three-dimensional association index table, and anonymize the real-time dynamic traffic flow data based on the land use attributes. The prediction module is used to perform steady-state filtering and prediction on the desensitized real-time dynamic traffic flow data based on the features associated in the three-dimensional association index table. The output adjustment suggestion module is used to periodically scan the three-dimensional association index table and output adjustment suggestions for the land parcel and the road segments contained in the land parcel based on the predicted traffic flow data.

8. The spatial data processing system for transportation planning based on land spatial elements according to claim 7, characterized in that, The index table creation module includes: The regional division module is used to divide the national land space into several traffic zones, extract the unique identification code of each traffic zone and its corresponding land use attributes, including building scale constraint indicators. The spatial mapping module is used to extract the road network topology covering the plot, establish a unique identification code for each road segment, and map the unique identification code of each road segment to the corresponding unique identification code of the plot according to the spatial inclusion relationship. The associated traffic flow data module is used to segment the time axis to form a time slice sequence, link dynamic traffic flow data with the corresponding time slices, and construct a three-dimensional association index table of unique land parcel identification code - unique road segment identification code - time slice.

9. The spatial data processing system for transportation planning based on land spatial elements according to claim 7, characterized in that, The data desensitization module includes: The "Get Traffic Flow Dataset" module is used to define the original traffic flow dataset to be processed. ,in, This represents the original real-time dynamic traffic flow data of the i-th sampling point in the sequence; Transformation module, used to map functions The original real-time dynamic traffic flow data is transformed to calculate the anonymized values. Where T represents the adjustment period, Indicates the initial phase shift. Indicates land use sensitivity. Represents the perturbation gain function; The probability distribution calculation module is used to calculate the probability distribution of the dataset to be de-identified using a Gaussian kernel function, and to construct the original probability distribution space. The specific implementation formula is as follows: in, This indicates that the values ​​in the original sequence are The probability of a data point appearing in the overall distribution. Represents the original traffic flow dataset The mean, This represents the standard deviation of the original traffic flow dataset X; The data mapping module is used to map the anonymized traffic flow dataset back to the original probability distribution space using a histogram specification algorithm, thereby obtaining the final anonymized traffic flow dataset Y.

10. The spatial data processing system for transportation planning based on land spatial elements according to claim 7, characterized in that, The prediction module includes: An input data module is constructed to normalize historical dynamic traffic flow data based on a preset historical backtracking window. Simultaneously, the coefficient of variation and the spatial carrying capacity feature components obtained by mapping the building scale constraint index are broadcast and aligned along the time axis. Tensor splicing is performed on the feature dimension to construct a multi-dimensional spatiotemporal feature matrix, which serves as the input to the deep learning model. A deep learning model building module is used to build a deep learning model based on recurrent neural units. The model integrates an adaptive gating operator layer, and the adaptive weighting operator is used as the control signal for this layer. Define a function module for defining loss functions: Where MSE represents the mean squared error between the predicted value and the actual value. This represents the upper limit threshold corresponding to this road segment, and λ represents the constraint penalty coefficient. y represents the predicted value, and y represents the actual value; The training model module is used to calculate the prediction error for each iteration using the composite loss function, map the physical constraint penalty term to the gradient correction value of the neuron weights using the error backpropagation algorithm, and optimize the parameters of the deep learning model. The prediction data acquisition module is used to input the multidimensional spatiotemporal feature matrix into the trained deep learning model and output the predicted traffic flow data for future time slices.

Images