A method and system for evaluating a dynamic job-housing balance state in a planning scenario

Abstract

The application discloses a kind of dynamic job-housing balance state evaluation method and system under planning scene, it is related to urban planning technical field;Build the "bridge" of planning land and commuting volume to carry out dynamic job-housing balance evaluation under planning scene, the connection between planning scene variable and dynamic job-housing balance key variable is broken through, make dynamic job-housing balance evaluation under planning state possible, and then provide quantitative support and evaluation means for the planning measures of job-housing balance;Meanwhile, the scheme also constructs the graph neural network algorithm based on spatial domain to distribute commuting flow, the topological neighborhood effect between plots is considered, the influence of local structure and neighbor attribute on commuting flow between plots is described, effectively improve the accuracy, operability, algorithm efficiency and result accuracy of prediction.

Description

Technical Field

[0001] This invention relates to the field of urban planning technology, specifically to a method and system for assessing the dynamic job-housing balance under planning scenarios. Background Technology

[0002] The rational layout of urban work-residence spaces to promote a balance between work and residence has always been a core issue in urban development. Among these, the scientific and quantitative assessment of this balance has been a crucial research topic. Scholars generally believe that the measurement of work-residence balance is not the traditional static relationship between "work" and "residence"; rather, the dynamic correlation between residents' places of employment and residence more accurately reflects the state of work-residence balance, i.e., the dynamic work-residence balance status. The dynamic work-residence balance index is considered the most representative and effective indicator for measuring dynamic work-residence balance. The dynamic work-residence balance index reflects the degree of self-sufficiency in employment and residence within a certain region; that is, the ratio of locally residing and employed individuals to the total number of employed persons in a given geographical area.

[0003] There are two traditional methods for obtaining work-residence relationship data: questionnaire surveys and statistical or census data. Both inevitably have data limitations. While questionnaire surveys can obtain work-residence correspondences, their high cost results in low sampling rates and relatively poor data representativeness. Statistical or census data, while providing large sample sizes, primarily reflect household or employment-related data and cannot reflect individual work-residence correspondences. Big data technology provides an effective way to obtain large-sample work-residence correspondences (dynamic work-residence relationships), enabling an understanding of the spatial distribution and correspondence of urban residents' work and residence at the individual level. Mobile phone signaling and internet location data are the most important big data sources; their combined use with statistical and traditional survey data can yield a relatively complete work-residence distribution and correspondence. Generally, long-term mobile phone signaling data can be used with specific algorithms to obtain work-residence distribution data and characteristics for specific areas (resident users and employed users in each area and their work-residence area connections), and then the dynamic work-residence balance index for a given area can be calculated.

[0004] Existing technological solutions typically utilize big data to study the job-housing balance, including work and residence spaces, land use, and population groups, but these are limited to assessing the current situation. However, in practice, we need not only to assess the current situation but also to proactively determine the rationality of the job-housing balance under future planning scenarios and to evaluate whether a series of planning measures have improved the dynamic job-housing balance of a city or region. Currently, however, there is a lack of effective methods for simulating the dynamic job-housing index under planning scenarios.

[0005] On the one hand, the lack of a "bridge" connecting changes in land use and roads to changes in commuting volume under planning scenarios makes it impossible to quantitatively assess the dynamic job-housing balance under planning scenarios. On the other hand, existing commuting flow prediction models can be broadly divided into two categories. One category is traditional spatial interaction models, such as gravity models and radiation models. These models have relatively fixed inputs and simple expressions, and do not consider rich geographic semantic information, making it difficult to model complex human mobility patterns. The second category is machine learning models, such as neural networks and tree models, which mostly simply utilize the features of the starting and ending points without fully considering the topological structure of the spatial interaction network and the spatial correlation brought by neighboring areas. Summary of the Invention

[0006] The technical problem this invention aims to solve is that traditional commuter traffic prediction models fail to reflect the relationship between an individual's job and residence, making it difficult to quantitatively assess the dynamic job-residence balance under planning scenarios. This invention provides a method and system for assessing the dynamic job-residence balance under planning scenarios. It quantifies data for planning scenarios and combines it with a spatial domain graph neural network algorithm to complete the assessment of the dynamic job-residence balance under planning scenarios. It builds a "bridge" between planned land use and commuter volume to assess the dynamic job-residence balance under planning scenarios, establishing a connection between planning scenario variables and key variables of dynamic job-residence balance. This makes dynamic job-residence balance assessment under planning conditions possible, thereby providing quantitative support and assessment methods for job-residence balance planning measures.

[0007] This invention is achieved through the following technical solution:

[0008] This solution provides a method for assessing the dynamic job-housing balance under a planning scenario, including:

[0009] Step 1: Obtain planning scenario data for the target area, including land use data and road network data under the planning scenario;

[0010] Step 2: Construct a commuting volume model, a commuting attraction model, and a commuting cost model for the target area;

[0011] Step 3: Based on the commuting occurrence model, commuting attraction model, and commuting cost model, construct a spatial domain-based graph neural network algorithm to complete commuting allocation, thereby obtaining the commuting flow between different plots under the planning scenario;

[0012] Step 4: Calculate the dynamic job-housing balance index based on commuting traffic between different plots under the planning scenario.

[0013] The planning scenario data mainly includes land use settings and road network settings such as roads and rail transit under future scenarios. Among them, the latest control plan land use is used as the land use scheme under the planning scenario; the road network scheme under the planning scenario is constructed by combining the latest road planning data, rail planning data and existing road network data.

[0014] The working principle of this solution: Existing commuter traffic prediction models can be broadly categorized into two types: The first type is traditional spatial interaction models, such as gravity models and radiation models. These models have relatively fixed inputs and simple expressions, and do not consider rich geographic semantic information, making it difficult to model complex human mobility patterns. The second type is machine learning models, such as neural networks and tree models. These mostly simply utilize the characteristics of the starting and ending points, without fully considering the topological structure of the spatial interaction network and the spatial correlation brought by neighboring areas. Existing technical solutions lack a "bridge" to commuter volume changes based on land use changes, road changes, etc., under planning scenarios, making it impossible to quantitatively assess the dynamic job-housing balance under planning scenarios.

[0015] This solution provides a dynamic job-housing balance assessment method under planning scenarios, offering a novel technical concept. It quantifies data within the planning scenario and combines it with a spatial domain graph neural network algorithm to complete the dynamic job-housing balance assessment. This establishes a "bridge" between planned land use and commuting volume, bridging the gap between planning scenario variables and key dynamic job-housing balance variables, making dynamic job-housing balance assessment under planning conditions possible. This provides quantitative support and assessment tools for job-housing balance planning measures. Furthermore, this solution constructs a spatial domain-based graph neural network algorithm for commuting traffic allocation, incorporating the topological proximity effect between plots to characterize the influence of local structure and neighbor attributes on commuting traffic between plots, effectively improving prediction accuracy, operability, algorithm efficiency, and result accuracy.

[0016] The further optimized plan is to establish a relationship between land use and commuting volume based on the "commuting occurrence rate"; where the commuting occurrence rate refers to the commuting demand generated per unit area of ​​land per unit time.

[0017] Methods for constructing commuting volume models include:

[0018] S1, classify the land use types of the target area, and divide the land where commuting occurs into residential land and commercial service land, with each type of land including multiple plots;

[0019] S2, identify benchmark cities and calculate the commuting incidence rate of each type of land use in the benchmark cities under the current conditions, and the target value of the commuting incidence rate of the target area under the planning scenario;

[0020] S3, calculates the actual commuting rate for each type of land use based on the current commuting data of the target area;

[0021] S4, analyze the commuting incidence rate of benchmark cities and the actual commuting incidence rate of the target area, and determine the commuting incidence rate of each type of land use in the target area under the planning scenario;

[0022] S5 multiplies the commuting occurrence rate of each land use type under the planning scenario by the building area of ​​the land use area to obtain the planned commuting occurrence of each plot.

[0023] The further optimized solution is that step S3 includes the following sub-steps:

[0024] S31, obtain current commuting data for the target area; such as commuting data obtained through mobile signaling or Baidu Smart Eye;

[0025] S32, separate the starting point location and the corresponding number of commuters from the current commuting data;

[0026] S33, based on the spatial overlay analysis of the starting point and the existing land parcels, the actual commuting volume of each parcel in the target area is determined;

[0027] S34, calculate the actual commuting rate for different streets and land use categories based on the actual commuting volume of each plot:

[0028]

[0029] Where, r i For the current commuting incidence rate of land use type i in the street, B ij S represents the commuting volume of the j-th plot of land in the current street (class i). ij Let n be the building area of ​​the j-th plot of land in the current street of type i, and n be the total number of plots of land in the current street of type i.

[0030] A further optimized solution is that step S4 includes the following sub-steps:

[0031] S41. By comparing the spatial relationship and land use categories of the current land use data with the land use data under the planning scenario, the target categories of land use to be added in the plan are identified. The target categories of land use include residential land and commercial service land; (i.e., newly added residential land and commercial service land).

[0032] S42, use the average commuting rate of the same type of land in the street corresponding to the newly added target type of land plot to assign the commuting rate of the newly added type of land plot;

[0033] S43, according to formula Calculate the expansion coefficient x between the actual commuting rate and the planned target value;

[0034] Where n represents the total number of plots of a certain type of land use, a i S represents the commuting volume of plot i. i A represents the building area of ​​plot i, and A represents the target value for commuting incidence rate.

[0035] S44, according to (a i ×x) / s i Calculate the initial commuting rate for each plot of land under the planning scenario;

[0036] S45, Correct the initial value of the commuting occurrence rate to obtain the commuting occurrence rate under the planning scenario of each plot: assign the minimum value of the preset reference range to the initial value of the commuting occurrence rate that is less than the preset reference range, and assign the maximum value of the preset reference range to the initial value of the commuting occurrence rate that is greater than the preset reference range.

[0037] A further optimized approach is to establish a relationship between land use and commuting attraction based on the "commuting attraction rate." The commuting attraction rate refers to the amount of commuting attracted per unit of building area per unit of time. The method for constructing the commuting attraction model includes:

[0038] T1 classifies the land use types in the target area, dividing land with commuting attraction into: commercial and service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land. Each type of land includes multiple plots.

[0039] T2, identify benchmark cities and calculate the commuting occurrence rate of each type of land use in the benchmark cities under the current conditions, as well as the target value of commuting attraction rate of the target area under the planning scenario;

[0040] T3 calculates the actual commuting attraction rate for each type of land use based on the current commuting data of the target area; specifically, it includes the following steps:

[0041] T31 acquires current commuting data for the target area; such as commuting data obtained through mobile signaling or Baidu Smart Eye.

[0042] T32 separates the destination location and the corresponding number of commuters from the current commuting data;

[0043] T33 analyzes the actual commuter attraction of each plot within the target area by overlaying the endpoint location and the existing plots.

[0044] T34 calculates the actual commuter attraction rate for different streets and land use categories based on the actual commuter attraction of each plot:

[0045]

[0046] Where, r i To determine the commuter attraction rate of current street type i land, B ijS represents the commuter attraction of the j-th plot of land in the current street's i-th land use category. ij Let n be the building area of ​​the j-th plot of land in the current street of type i, and n be the total number of plots of land in the current street of type i.

[0047] T4 analyzes the commuting attraction rate of benchmark cities and the actual commuting attraction rate of the target area to determine the commuting attraction rate of each land use type in the target area under the planning scenario; specifically, it includes the following sub-steps:

[0048] T41. By comparing the spatial relationship and land use categories between the current land use data and the land use data under the planning scenario, the target land use categories to be added in the plan are identified. The target land use categories include: commercial and service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land.

[0049] T42 uses the average commuting rate of the same type of land in the street corresponding to the newly added target type of land plot to assign the commuting rate of the newly added type of land plot;

[0050] T43, according to formula Calculate the expansion coefficient x between the actual commuting rate and the planned target value;

[0051] Where n represents the total number of plots of a certain type of land use, b i S represents the commuter attraction of plot i. i B represents the building area of ​​plot i, and B represents the target value for commuter attraction rate.

[0052] T44, according to (b) i ×x) / s i Calculate the initial commuting rate for each plot of land under the planning scenario;

[0053] T45, the initial value of the commuting attraction rate is corrected to obtain the commuting attraction rate under the planning scenario of each plot: the initial value of the commuting attraction rate that is less than the preset reference range is assigned the minimum value of the preset reference range, and the initial value of the commuting attraction rate that is greater than the preset reference range is assigned the maximum value of the preset reference range.

[0054] T5 calculates the planned commuter attraction of each land use type by multiplying its commuter attraction rate under the planning scenario by the building area of ​​the land use area.

[0055] A further optimization scheme is proposed, which uses time cost as the commuting cost, is based on the road network, and employs a graph search algorithm to calculate the shortest commuting time between commuting point O and employment center point D (at the plot scale), forming a commuting cost OD matrix; the method for constructing the commuting cost model includes:

[0056] G1, which allocates commuting costs based on time costs;

[0057] G2, Route Planning: Based on graph search algorithms, it finds the best route from the origin to the destination for each commuter trip in the road network data of the target area under the planning scenario / current situation;

[0058] G3 determines the commuting speed for different road types and uses a graph search algorithm to determine the commuting cost between any two plots, thus obtaining the commuting cost OD matrix between all plots under the planning scenario / current situation.

[0059] The further optimized solution is that step G2 includes the following sub-steps:

[0060] G21, construct a set O containing the center points of all commuting origin plots, and a set D containing the center points of all commuting destination plots;

[0061] G22, based on the point-pair matrix formed by the corresponding sets O and D (o i d j ); where o i Let d represent the center point of the i-th starting plot. j This represents the center point of the j-th terminal plot;

[0062] G23, determine the optimal commuting cost between all origins and destinations: iterate through o i to d j Record all possible routes and their corresponding commuting costs, and select the route with the lowest commuting cost as o. i to d j The optimal commuting cost (cost(o)) i d j );

[0063] G24 uses the route with the optimal commuting cost as the best commuting route.

[0064] A further optimization scheme involves constructing a spatial domain-based graph neural network algorithm to complete commuter allocation, thereby obtaining commuter traffic flow between different plots under the planning scenario, including the following methods:

[0065] H1, construct an NN adjacency matrix A between plots based on the adjacency relationship between plots, where N is the total number of plots in the target area;

[0066] H2, constructing a feature matrix X based on the location, commuting attraction, commuting occurrence, and land use type of each plot;

[0067] H3 performs graph embedding operations on the adjacency matrix A and the feature matrix X based on an encoder containing L layers of graph convolutional layers;

[0068] H4. Select the starting point and the destination to construct the commuting cost OD matrix, that is, home and workplace constitute the OD matrix. The commuting cost OD matrix representation vector is concatenated with the corresponding commuting cost and used as the input of the random forest regression model to predict the commuting flow between the starting point and the destination.

[0069] H5, based on H4, trains a graph neural network model;

[0070] H6 inputs the land use type, commuting attraction, commuting occurrence, and commuting cost under the planning scenario into a pre-trained graph neural network model to obtain the commuting flow between different plots under the planning scenario, and adjusts the commuting flow between different plots under the planning scenario.

[0071] The further optimized solution is that step four includes the following process:

[0072] Using the total commuting attraction as the total number of employed people (sumpop), we determine the total number of commuters (pop) at the starting and ending points of the commuting cost OD matrix one by one.

[0073] The dynamic job-housing balance index r is calculated based on the total number of employed persons (sumpop) and the total number of commuters (pop).

[0074]

[0075] Select the dynamic work-life balance assessment unit (determined according to actual needs, which can be any unit as needed, commonly used ones include districts, counties, streets, etc.);

[0076] Identify the industrial plots included in each dynamic job-housing balance assessment unit, and calculate the total commuting attraction of the industrial plots in each dynamic job-housing balance assessment unit as the total number of employees in the assessment unit, denoted as sumpop.

[0077] For each planning dynamic work-housing balance assessment unit, the total number of commuters (i.e., people who live and work locally) in the OD of the commuting cost OD matrix obtained above, where both the starting and ending plots are in the planning dynamic work-housing balance assessment unit, is determined and counted, and is denoted as pop.

[0078] For each planning dynamic job-housing balance assessment unit, the planning scenario dynamic job-housing balance index r is obtained by calculating the ratio of pop to sumpop.

[0079] This solution also provides a dynamic job-housing balance assessment system under a planning scenario, used to implement the above-mentioned dynamic job-housing balance assessment method under a planning scenario, the system comprising:

[0080] The acquisition module is used to acquire planning scenario data for the target area, including land use data and road network data under the planning scenario.

[0081] The model building module is used to build models of commuting occurrence, commuting attraction, and commuting cost for the target area.

[0082] The commuting allocation module is used to construct a spatial domain-based graph neural network algorithm based on the commuting occurrence model, commuting attraction model, and commuting cost model to complete commuting allocation, thereby obtaining the commuting flow between different plots under the planning scenario.

[0083] The calculation module is used to calculate the dynamic job-housing balance index based on commuting traffic between different plots under the planning scenario.

[0084] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0085] 1. The present invention provides a method and system for assessing the dynamic job-housing balance under planning scenarios, which builds a "bridge" between planned land use and commuting volume to conduct dynamic job-housing balance assessment under planning scenarios, opens up the connection between planning scenario variables and key variables of dynamic job-housing balance, makes dynamic job-housing balance assessment under planning conditions possible, and provides quantitative support and assessment means for planning measures for job-housing balance.

[0086] 2. The present invention provides a method and system for assessing the dynamic job-housing balance under planning scenarios. It constructs a graph neural network algorithm based on the spatial domain for commuter traffic allocation, takes into account the topological proximity effect between plots, and characterizes the influence of local structure and neighbor attributes on commuter traffic between plots, effectively improving the accuracy, operability, algorithm efficiency and result accuracy of prediction. Attached Figure Description

[0087] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0088] Figure 1 This is a schematic diagram of the process for assessing the dynamic job-housing balance based on spatial graph neural networks.

[0089] Figure 2 This is a schematic diagram illustrating the principle of a dynamic job-housing balance assessment method based on spatial graph neural networks. Detailed Implementation

[0090] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0091] Example 1

[0092] This embodiment provides a method for assessing the dynamic job-housing balance under a planning scenario, such as... Figure 1 and Figure 2 As shown, it includes:

[0093] Step 1: Obtain planning scenario data for the target area. The planning scenario data includes land use data and road network data under the planning scenario. In this embodiment, the planning scenario data mainly includes land use settings and road network settings such as roads and rail transit under the future scenario. The latest control plan land use is used as the land use scheme under the planning scenario. The road network scheme under the planning scenario is constructed by combining the latest road planning data, rail planning data and existing road network data.

[0094] Step 2: Construct a commuting volume model, a commuting attraction model, and a commuting cost model for the target area;

[0095] The relationship between land use and commuting volume is established using "commuting occurrence rate," where commuting occurrence rate refers to the commuting demand generated per unit area of ​​land per unit time. The methods for constructing the commuting volume model include:

[0096] S1, classify the land use types of the target area, and divide the land where commuting occurs into: residential land and commercial service land, with each type of land including multiple plots;

[0097] S2, identify benchmark cities and calculate the commuting incidence rate of each type of land use in the benchmark cities under the current conditions, and the target value of the commuting incidence rate of the target area under the planning scenario;

[0098] This embodiment selects two domestic cities that rank higher than the current city in the "Global Urban Competitiveness Report" published by the Chinese Academy of Social Sciences that year as benchmark cities.

[0099] The current commuting incidence rate of residential and commercial service land in selected benchmark cities was calculated using commuting big data. The average commuting incidence rate of the two benchmark cities was then used as the target value for commuting incidence rate under the current urban planning scenario, as shown in Table 1.

[0100] Table 1. Examples of Reference Standards for Commuting Incidence Rate

[0101]

[0102] S3, calculate the actual commuting rate for each type of land use based on the current commuting data of the target area; specifically including the following sub-steps:

[0103] S31, obtain current commuting data for the target area; such as commuting data obtained through mobile signaling or Baidu Smart Eye;

[0104] S32, separate the starting point location and the corresponding number of commuters from the current commuting data;

[0105] S33, based on the spatial overlay analysis of the starting point and the existing land parcels, the actual commuting volume of each parcel in the target area is determined;

[0106] S34, calculate the actual commuting rate for different streets and land use categories based on the actual commuting volume of each plot:

[0107]

[0108] Where, r i For the current commuting incidence rate of land use type i in the street, B ij S represents the commuting volume of the j-th plot of land in the current street (class i). ij Let n be the building area of ​​the j-th plot of land in the current street of type i, and n be the total number of plots of land in the current street of type i.

[0109] S4, analyze the commuting incidence rate of benchmark cities and the actual commuting incidence rate of the target area, and determine the commuting incidence rate of each type of land use in the target area under the planning scenario;

[0110] Step S4 includes the following sub-steps:

[0111] S41, compare the spatial relationship and land use categories of the current land use data with the land use data under the planning scenario, and identify the new target land use categories to be added in the plan. The target land use categories include residential land and commercial service land.

[0112] S42, use the average commuting rate of the same type of land in the street corresponding to the newly added target type of land plot to assign the commuting rate of the newly added type of land plot;

[0113] S43, according to formula Calculate the expansion coefficient x between the actual commuting rate and the planned target value;

[0114] Where n represents the total number of plots of a certain type of land use, a i S represents the commuting volume of plot i. i A represents the building area of ​​plot i, and A represents the target value for commuting incidence rate.

[0115] S44, according to (a i ×x) / si Calculate the initial commuting rate for each plot of land under the planning scenario;

[0116] S45, Correct the initial value of the commuting occurrence rate to obtain the commuting occurrence rate under the planning scenario of each plot: assign the minimum value of the preset reference range to the initial value of the commuting occurrence rate that is less than the preset reference range, and assign the maximum value of the preset reference range to the initial value of the commuting occurrence rate that is greater than the preset reference range.

[0117] This embodiment refers to the Ministry of Housing and Urban-Rural Development's "Technical Standard for Traffic Impact Assessment of Construction Projects" (CJJ / T141-2010), and uses 50% of the traffic travel rate as the reference range for calculating the commuting occurrence rate as the preset reference range.

[0118] S5 multiplies the commuting occurrence rate of each land use type under the planning scenario by the building area of ​​the land use area to obtain the planned commuting occurrence of each plot.

[0119] The relationship between land use and commuting attraction is established using the "commuting attraction rate," where the commuting attraction rate refers to the amount of commuting attraction attracted per unit building area per unit time. The methods for constructing the commuting attraction model include:

[0120] T1 categorizes land use types in the target area, classifying land with commuting attraction into five types: commercial and service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land; each type of land includes multiple plots.

[0121] T2, identify benchmark cities and calculate the commuting occurrence rate of each type of land use in the benchmark cities under the current conditions, as well as the target value of commuting attraction rate of the target area under the planning scenario;

[0122] This embodiment uses the city competitiveness ranking in the "Global City Competitiveness Report" published by the Chinese Academy of Social Sciences that year to select two domestic cities that rank higher than the current city and are closest in rank to the current city as benchmark cities. Through commuting big data calculation, the commuting attraction rate of various types of industrial land (five major categories of industrial land: commercial service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land) of the selected benchmark cities is obtained, and the average of the commuting attraction rates of the two cities is taken as the target value of commuting attraction rate under the current urban planning scenario.

[0123] Table 2. Examples of Reference Standards for Commuting Attraction Rate

[0124]

[0125] T3 calculates the actual commuting attraction rate for each type of land use based on the current commuting data of the target area; specifically, it includes the following steps:

[0126] T31 retrieves current commuting data for the target area; such as commuting data obtained via mobile signaling or Baidu Smart Eye.

[0127] T32 separates the destination location and the corresponding number of commuters from the current commuting data;

[0128] T33 analyzes the actual commuter attraction of each plot within the target area by overlaying the endpoint location and the existing plots.

[0129] T34 calculates the actual commuter attraction rate for different streets and land use categories based on the actual commuter attraction of each plot:

[0130]

[0131] Where, r i To determine the commuter attraction rate of current street type i land, B ij S represents the commuter attraction of the j-th plot of land in the current street's i-th land use category. ij Let n be the building area of ​​the j-th plot of land in the current street of type i, and n be the total number of plots of land in the current street of type i.

[0132] T4 analyzes the commuting attraction rate of benchmark cities and the actual commuting attraction rate of the target area to determine the commuting attraction rate of each land use type in the target area under the planning scenario; specifically, it includes the following sub-steps:

[0133] T41. By comparing the spatial relationship and land use categories between the current land use data and the land use data under the planning scenario, the target land use categories to be added in the plan are identified. The target land use categories include: commercial and service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land.

[0134] T42 uses the average commuting rate of the same type of land in the street corresponding to the newly added target type of land plot to assign the commuting rate of the newly added type of land plot;

[0135] T43, according to formula Calculate the expansion coefficient x between the actual commuting rate and the planned target value;

[0136] Where n represents the total number of plots of a certain type of land use, b i S represents the commuter attraction of plot i. i B represents the building area of ​​plot i, and B represents the target value for commuter attraction rate.

[0137] T44, according to (b) i ×x) / s i Calculate the initial commuting rate for each plot of land under the planning scenario;

[0138] T45, the initial value of the commuting attraction rate is corrected to obtain the commuting attraction rate under the planning scenario of each plot: the initial value of the commuting attraction rate that is less than the preset reference range is assigned the minimum value of the preset reference range, and the initial value of the commuting attraction rate that is greater than the preset reference range is assigned the maximum value of the preset reference range.

[0139] This embodiment refers to the Ministry of Housing and Urban-Rural Development's "Technical Standard for Traffic Impact Assessment of Construction Projects" (CJJ / T141-2010), and uses 50% of the traffic travel rate as the reference range for calculating the commuting attraction rate as the preset reference range.

[0140] T5 calculates the planned commuter attraction of each land use type by multiplying its commuter attraction rate under the planning scenario by the building area of ​​the land use area.

[0141] Using time cost as the commuting cost, and based on the road network, a graph search algorithm is used to calculate the shortest commuting time between commuting point O and employment center point D (at the plot scale), forming a commuting cost OD matrix; the construction method of the commuting cost model includes:

[0142] G1 uses time cost to configure commuting costs; the time cost is the most closely related to the changes in land use and transportation brought about by planning and can be measured during the commuting process. The time cost is measured by dividing the commuting distance by the average speed to obtain the commuting time.

[0143] G2, Route Planning: Based on graph search algorithms, it finds the best route from the origin to the destination for each commuter trip in the road network data of the target area under the planning scenario / current situation;

[0144] Step G2 includes the following sub-steps:

[0145] G21, construct a set O containing the center points of all commuting origin plots (origin plots include residential land and commercial service land), and a set D containing the center points of all commuting destination plots (destination plots include commercial service land, industrial and mining land, public service and administrative land, logistics and warehousing land and transportation service station land).

[0146] G22, based on the point-pair matrix formed by the corresponding sets O and D (o i d j ); where o i Let d represent the center point of the i-th starting plot. j This represents the center point of the j-th terminal plot;

[0147] G23, determine the optimal commuting cost between all origins and destinations: iterate through o i to d jRecord all possible routes and their corresponding commuting costs, and select the route with the lowest commuting cost as o. i to d j The optimal commuting cost (cost(o)) i d j );

[0148] G24 uses the route with the optimal commuting cost as the best commuting route.

[0149] G3 determines the commuting speed for different road types and uses a graph search algorithm to determine the commuting cost between any two plots, thus obtaining the commuting cost OD matrix between all plots under the planning scenario / current situation.

[0150] In the process of converting path distance into the time cost required for the path, we first assume that the planned speed of various roads in the future will be comparable to the current level. Then, we obtain the speed standards corresponding to different types of roads based on the existing traffic planning standards and road design specifications (see Tables 3 and 4).

[0151] Table 3 Minimum speed limit for average journey of motor vehicles during peak hours (GB / T51328-2018) (km / h)

[0152] Road Class City center expressway 30 Main road 20

[0153] Table 4 Design speeds for roads of all levels (CJJ37-2012) (km / h)

[0154] Road Class expressway Main road Secondary arterial road branch road Design speed 60 40 30 20

[0155] First, based on the existing subway operating speed, the planned commuter speed is set at 40 km / h. Expressways and expressways are classified into one category, and the peak-hour speed standard for this type of road is set at 30 km / h, referring to the "Urban Comprehensive Transportation System Planning Standard" (GB / T51328-2018). National highways, provincial highways, and arterial roads are classified into another category, and the peak-hour speed standard for this type of road is set at 20 km / h, referring to the "Urban Comprehensive Transportation System Planning Standard" (GB / T51328-2018). Referring to the "Urban Comprehensive Transportation System Planning Standard" (GB / T51328-2018) and the "Urban Road Engineering Design Code" (CJJ37-2012), the peak-hour speed standard for county roads and secondary arterial roads is set at 15 km / h. Branch roads, auxiliary roads, and highway ramps are classified into one category, and the peak-hour speed standard for this type of road is set at 10 km / h, referencing the "Urban Comprehensive Transportation System Planning Standard" (GB / T51328-2018) and the "Urban Road Engineering Design Code" (CJJ37-2012). To reflect the differences in road network levels, the peak-hour speed standard for ramps is set at 5 km / h. Internal roads and pedestrian roads are classified into another category, and their speed standard is set at 3 km / h based on the pedestrian speed standard. Ultimately, roads are divided into 7 categories, refining the commuting speed settings for each level of the road network, and converting these into commuting time as commuting cost.

[0156] Table 5. Commuting speed conversion for roads of all levels

[0157] Commute speed (km / h) Road type 40 subway 30 highways and expressways 20 National highways, provincial highways, and main roads 15 Secondary arterial road 10 Auxiliary roads, expressway ramps, county roads, and branch roads 5 ramp 3 Internal roads and pedestrian walkways

[0158] Table 6 Example of Commuting Cost OD Matrix Results

[0159]

[0160]

[0161] The commuting cost model under the planning scenario for the entire study area has now been completed.

[0162] In addition, the OD matrix of commuting costs under the current conditions can be constructed by following the same steps based on the existing road network and land use.

[0163] Step 3: Based on the commuting occurrence model, commuting attraction model, and commuting cost model, construct a spatial domain-based graph neural network algorithm to complete commuting allocation, thereby obtaining the commuting flow between different plots under the planning scenario;

[0164] Including methods:

[0165] H1 constructs an N-dimensional adjacency matrix A between plots based on their adjacency relationships, where N is the total number of plots in the target area; two plots are adjacent if they share an edge, otherwise they are not adjacent.

[0166] H2, constructing a feature matrix X based on the location, commuting attraction, commuting occurrence, and land use type of each plot;

[0167] H3 performs graph embedding operations on the adjacency matrix A and the feature matrix X based on an encoder containing L layers of graph convolutional layers;

[0168] Specifically, each layer includes two steps: for each plot, message passing collects the feature representations of its neighboring nodes, while state update calculates its new representation vector. This can be expressed by formula (2):

[0169]

[0170] middle It is the l-th layer node v j The representation vector, l = 1, 2, ..., L, It is the feature vector input of the first layer. This is the encoder's output. W l It is the weight matrix of the l-th layer, k i It is the normalization coefficient, N i It is v i The set of neighbor indices (i.e., with v) i Geographical units with spatial interactions.

[0171] H4, select the starting point and the destination to construct the commuting cost OD matrix, that is, the home and the workplace constitute the OD matrix. The commuting cost OD matrix representation vector is concatenated with the corresponding commuting cost and used as the input of the random forest regression model to predict the commuting flow between the starting point and the destination; expressed by formula (3):

[0172] f ij =RF(e i ||c ij ||e j (3)

[0173] Among them, e i and e j Represents the embedding of nodes i and j, c ij The commuting cost between nodes is represented by ||, where || denotes a join operation.

[0174] H5, based on H4, trains a graph neural network model;

[0175] Specifically, the actual commuting OD data between existing plots (such as mobile phone signaling or Baidu commuting OD data) is input into the model. Based on the predicted and actual values ​​of commuting traffic, the loss (mean square error, MSE, see formula (4)) is calculated. The loss is returned to train the parameters of the model, and then the optimal values ​​of each parameter in the model are solved to obtain the trained graph neural network model.

[0176]

[0177] Among them, f ij Predict commuter traffic between nodes i and j. This represents actual commuter traffic.

[0178] H6 inputs the land use type, commuting attraction, commuting occurrence, and commuting cost under the planning scenario into a pre-trained graph neural network model to obtain the commuting flow between different plots under the planning scenario, and adjusts the commuting flow between different plots under the planning scenario.

[0179] The adjustment method provided in this embodiment includes: statistically analyzing the commuter traffic volume simulated for each industrial plot, and then constraining the simulated commuter traffic volume based on the commuter attraction under the planning scenario of each industrial plot.

[0180]

[0181] Where, f′ ij f represents the adjusted predicted commuter traffic between plots i and j. i f represents the commuter attraction of plot i. ij This represents the predicted commuter traffic between plots i and j, and n represents the total number of residential plots.

[0182] Step four involves calculating the dynamic job-housing balance index based on commuter traffic flow between different plots under the planning scenario; specifically including:

[0183] Using the total commuting attraction as the total number of employed people (sumpop), we determine the total number of commuters (pop) at the starting and ending points of the commuting cost OD matrix one by one.

[0184] The dynamic job-housing balance index r is calculated based on the total number of employed persons (sumpop) and the total number of commuters (pop).

[0185]

[0186] Select the planning dynamic work-life balance assessment unit (determined according to actual needs, it can be any unit as needed, commonly used ones include districts, counties, streets, etc.).

[0187] Identify the industrial plots included in each dynamic job-housing balance assessment unit, and calculate the total commuting attraction of the industrial plots in each dynamic job-housing balance assessment unit as the total number of employees in the assessment unit, denoted as sumpop.

[0188] For each planning dynamic work-housing balance assessment unit, the total number of commuters (i.e., people who live and work locally) in the OD of the commuting cost OD matrix obtained above, where both the starting and ending plots are in the planning dynamic work-housing balance assessment unit, is determined and counted, and is denoted as pop.

[0189] For each planning dynamic job-housing balance assessment unit, the planning scenario dynamic job-housing balance index r is obtained by calculating the ratio of pop to sumpop.

[0190] Example 2

[0191] This embodiment provides a dynamic job-housing balance assessment system under a planning scenario, used to implement the dynamic job-housing balance assessment method under a planning scenario in Embodiment 1. The system includes:

[0192] The acquisition module is used to acquire planning scenario data for the target area, including land use data and road network data under the planning scenario.

[0193] The model building module is used to build models of commuting occurrence, commuting attraction, and commuting cost for the target area.

[0194] The commuting allocation module is used to construct a spatial domain-based graph neural network algorithm based on the commuting occurrence model, commuting attraction model, and commuting cost model to complete commuting allocation, thereby obtaining the commuting flow between different plots under the planning scenario.

[0195] The calculation module is used to calculate the dynamic job-housing balance index based on commuting traffic between different plots under the planning scenario.

[0196] Example 3

[0197] This embodiment provides a computer-readable medium having a computer program stored thereon. The computer program, when executed by a processor, can implement a dynamic job-housing balance assessment method under a planning scenario as described in Embodiment 1.

[0198] This invention innovatively proposes to build a "bridge" between land use and commuting volume by using commuting occurrence rate and commuting attraction rate, thus establishing a connection between planning scenario variables and key variables of dynamic job-housing balance. This makes dynamic job-housing balance assessment under planning conditions possible, and provides quantitative support and evaluation methods for job-housing balance planning measures. By constructing a spatial domain graph neural network model for commuting flow prediction, the topological proximity effect between plots is taken into account, characterizing the influence of local structure and neighbor attributes on commuting flow between plots, effectively improving the accuracy of prediction. Furthermore, the parameters of the commuting prediction model based on the graph neural network model are calibrated using current commuting big data such as mobile phone signaling and Baidu commuting OD, thereby ensuring the accuracy and reliability of the simulation.

[0199] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for assessing the dynamic job-housing balance under a planning scenario, characterized in that, include: Step 1: Obtain planning scenario data for the target area, including land use data and road network data under the planning scenario; Step 2: Construct a commuting volume model, a commuting attraction model, and a commuting cost model for the target area; Step 3: Based on the commuting occurrence model, commuting attraction model, and commuting cost model, construct a spatial domain-based graph neural network algorithm to complete commuting allocation, thereby obtaining the commuting flow between different plots under the planning scenario; Step 4: Calculate the dynamic job-housing balance index based on commuting traffic between different plots under the planning scenario; The method for constructing the commuting volume model includes: S1, classify the land use types of the target area, divide the areas in the target area where commuting occurs into residential land and commercial service land, with each type of land including multiple plots; S2, identify benchmark cities and calculate the commuting incidence rate of each type of land use in the benchmark cities under the current conditions, so as to obtain the target value of commuting incidence rate of the target area under the planning scenario; S3, calculates the actual commuting rate for each type of land use based on the current commuting data of the target area; S4, analyze the commuting incidence rate of benchmark cities and the actual commuting incidence rate of the target area, and determine the commuting incidence rate of each type of land use in the target area under the planning scenario; S5, multiply the commuting occurrence rate of each type of land use under the planning scenario by the building area of ​​the land use to obtain the planned commuting occurrence of each plot; Step S4 includes the following sub-steps: S41, compare the spatial relationship and land use categories of the current land use data with the land use data under the planning scenario, and identify the new target land use categories to be added in the plan. The target land use categories include residential land and commercial service land. S42, use the average commuting rate of the same type of land in the street corresponding to the newly added target type of land plot to assign the commuting rate of the newly added type of land plot; S43, according to formula Calculate the expansion coefficient x between the actual commuting rate and the planned target value; Where n represents the total number of land parcels of a certain type of land use. This represents the commuting volume of plot i. A represents the building area of ​​plot i, and A represents the target commuting rate of the target area under the planning scenario. S44, according to formula Calculate the initial commuting rate for each plot of land under the planning scenario; S45, Correct the initial value of the commuting occurrence rate to obtain the commuting occurrence rate under the planning scenario of each plot: assign the minimum value of the preset reference range to the initial value of the commuting occurrence rate that is less than the preset reference range, and assign the maximum value of the preset reference range to the initial value of the commuting occurrence rate that is greater than the preset reference range. The method for constructing a spatial domain-based graph neural network algorithm to complete commuter allocation and thus obtain commuter traffic flow between different plots under the planning scenario includes: H1, construct an N×N adjacency matrix A between plots based on the adjacency relationship between plots, where N is the total number of plots in the target area; H2, constructing a feature matrix X based on the location, commuting attraction, commuting occurrence, and land use type of each plot; H3 performs graph embedding operations on the adjacency matrix A and the feature matrix X based on an encoder containing L layers of graph convolutional layers; H4. Select the starting point and the destination to construct the commuting cost OD matrix. Concatenate the commuting cost OD matrix representation vector with the corresponding commuting cost and use it as the input of the random forest regression model to predict the commuting flow between the starting point and the destination. H5, based on H4, trains a graph neural network model; H6 inputs the land use type, commuting attraction, commuting occurrence and commuting cost under the planning scenario into the pre-trained graph neural network model to obtain the commuting flow between different plots under the planning scenario, and adjusts the commuting flow between different plots under the planning scenario. Step four includes the following process: Using the total commuting attraction as the total number of employed people (sumpop), we determine the total number of commuters (pop) at the starting and ending points of the commuting cost OD matrix one by one. The dynamic job-housing balance index r is calculated based on the total number of employed persons (sumpop) and the total number of commuters (pop). 。 2. The method for assessing the dynamic job-housing balance under a planning scenario according to claim 1, characterized in that, Step S3 includes the following sub-steps: S31, Obtain current commuting data for the target area; S32, separate the starting point location and the corresponding number of commuters from the current commuting data; S33, based on the spatial overlay analysis of the starting point and the existing land parcels, the actual commuting volume of each parcel in the target area is determined; S34, calculate the actual commuting rate for different streets and land use categories based on the actual commuting volume of each plot: ； in, The current commuting rate for Class i land use in the street. This represents the commuting volume of the j-th plot of land in the i-th type of street. Let n be the building area of ​​the j-th plot of land in the current street of type i, and n be the total number of plots of land in the current street of type i.

3. The method for assessing the dynamic job-housing balance under a planning scenario according to claim 1, characterized in that, The method for constructing the commuting attraction model includes: T1 categorizes the land use types within the target area, classifying areas with commuter attraction into: commercial and service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land; each type of land use includes multiple plots. T2. Identify benchmark cities and calculate the commuting occurrence rate of each type of land use in the benchmark cities under the current conditions, so as to obtain the target value of commuting attraction rate of the target area under the planning scenario. T3 calculates the actual commuting attraction rate for each type of land use based on current commuting data for the target area; specifically, it includes the following steps: T31, obtain current commuting data for the target area; T32 separates the destination location and the corresponding number of commuters from the current commuting data; T33 analyzes the actual commuter attraction of each plot within the target area by overlaying the endpoint location and the existing plots. T34 calculates the actual commuter attraction rate for different streets and land use categories based on the actual commuter attraction of each plot: ； in, To determine the commuter attraction rate of the current street-level Class I land use, The commuter attraction of the j-th plot of land in the current street (class i). Let n be the building area of ​​the j-th plot of land in the current street of type i, and n be the total number of plots of land in the current street of type i. T4 analyzes the commuting attraction rate of benchmark cities and the actual commuting attraction rate of the target area to determine the commuting attraction rate of each land use type in the target area under the planning scenario; specifically, it includes the following sub-steps: T41. By comparing the spatial relationship and land use categories between the current land use data and the land use data under the planning scenario, the target land use categories to be added in the plan are identified. The target land use categories include: commercial and service land, industrial and mining land, public service and administrative land, logistics and warehousing land, and transportation service station land. T42 uses the average commuting rate of the same type of land in the street corresponding to the newly added target type of land plot to assign the commuting rate of the newly added type of land plot; T43, according to formula Calculate the expansion coefficient x between the actual commuting rate and the planned target value; Where n represents the total number of land parcels of a certain type of land use. This represents the commuter attraction of plot i. B represents the building area of ​​plot i, and B represents the target commuter attraction rate of the target area under the planning scenario. T44, according to Calculate the initial commuting rate for each plot of land under the planning scenario; T45, the initial value of the commuting attraction rate is corrected to obtain the commuting attraction rate under the planning scenario of each plot: the initial value of the commuting attraction rate that is less than the preset reference range is assigned the minimum value of the preset reference range, and the initial value of the commuting attraction rate that is greater than the preset reference range is assigned the maximum value of the preset reference range. T5 calculates the planned commuter attraction of each land use type by multiplying its commuter attraction rate under the planning scenario by the building area of ​​the land use area.

4. The method for assessing the dynamic job-housing balance under a planning scenario according to claim 1, characterized in that, The method for constructing the commuting cost model includes: G1, which allocates commuting costs based on time costs; G2, Route Planning: Based on graph search algorithms, it finds the best route from the origin to the destination for each commuter trip in the road network data of the target area under the planning scenario / current situation; G3 determines the commuting speed for different road types and uses a graph search algorithm to determine the commuting cost between any two plots, thus obtaining the commuting cost OD matrix between all plots under the planning scenario / current situation.

5. The method for assessing the dynamic job-housing balance under a planning scenario according to claim 4, characterized in that, Step G2 includes the following sub-steps: G21, construct a set O containing the center points of all commuting origin plots, and a set D containing the center points of all commuting destination plots; G22, based on the point-pair matrix formed by the corresponding sets O and D (o i d j ); where o i Let d represent the center point of the i-th starting plot. j This represents the center point of the j-th terminal plot; G23, determine the optimal commuting cost between all origins and destinations: iterate through o i to d j Record all possible routes and their corresponding commuting costs, and select the route with the lowest commuting cost as o. i to d j The optimal commuting cost (cost(o)) i d j ); G24 uses the route with the optimal commuting cost as the best commuting route.

6. A dynamic job-housing balance assessment system under a planning scenario, used to implement the dynamic job-housing balance assessment method under a planning scenario as described in any one of claims 1-5, the system comprising: The acquisition module is used to acquire planning scenario data for the target area, including land use data and road network data under the planning scenario. The model building module is used to build models of commuting occurrence, commuting attraction, and commuting cost for the target area. The commuting allocation module is used to construct a spatial domain-based graph neural network algorithm based on the commuting occurrence model, commuting attraction model, and commuting cost model to complete commuting allocation, thereby obtaining the commuting flow between different plots under the planning scenario. The calculation module is used to calculate the dynamic job-housing balance index based on commuting traffic between different plots under the planning scenario.

Images