Method for identifying substantial boundary of metropolitan area or urban agglomeration based on multi-modal transportation network

Abstract

The application proposes a metropolitan area or urban agglomeration substantial boundary identification method based on a multi-modal traffic network, which comprises the following steps: obtaining and preprocessing multi-source travel information data; calculating point accessibility based on improved Mahalanobis distance; calculating node accessibility based on abnormal influence factors; calculating node comprehensive accessibility; identifying the substantial boundary of the metropolitan area or urban agglomeration; and verifying the identification result based on the function polycentricity. The method designs a new point accessibility, node accessibility and node comprehensive accessibility calculation system based on emerging big data analysis technology, and proposes a simple, efficient, accurate and easy-to-understand metropolitan area or urban agglomeration substantial boundary identification method, which has a profound theoretical and practical significance for the application of related big data technology in the field of city and traffic planning and engineering.

Description

Technical Field

[0001] This invention relates to a method for identifying the actual boundaries of metropolitan areas or urban clusters in the context of multimodal transportation network travel, belonging to the interdisciplinary field of transportation and urban planning. Background Technology

[0002] As a new unit of development that transcends administrative boundaries in the process of modern urbanization, the socio-economic characteristics, urban planning, and transportation planning of multi-center urban areas such as metropolitan areas and urban agglomerations have received widespread attention from the industry. However, there is still a lack of methods for identifying the substantive boundaries of metropolitan areas and urban agglomerations. With the increasing convenience for residents to obtain information on land travel, emerging big data analysis technologies have made it possible to identify the substantive boundaries of multi-center urban areas such as metropolitan areas and urban agglomerations, which has important practical significance for urban and transportation planning, design, and other engineering practices. Summary of the Invention

[0003] The purpose of this invention is to provide a method for identifying the actual boundaries of metropolitan areas or urban agglomerations in the context of multimodal transportation networks, providing a scientific and quantitative basis for the formulation, adjustment, and implementation of future transportation planning and policies related to metropolitan areas and urban agglomerations. At the same time, this method can also, to a certain extent, reduce the deviation between the implementation of planning schemes and development policies and their actual application scope.

[0004] This invention identifies the physical boundaries of metropolitan areas or urban clusters by evaluating the comprehensive accessibility of nodes in a multimodal transportation network, and verifies the identification results of physical boundaries by combining functional polycentricity.

[0005] To address the aforementioned technical problems, this invention provides a method for identifying the substantial boundaries of metropolitan areas or urban clusters based on multimodal transportation networks, using big data analytics. The method includes the following steps:

[0006] Step 1: Acquisition and Preprocessing of Multi-Source Travel Information Data

[0007] Multi-source travel information data mainly includes origin and destination information, ticketing information, vehicle information (including capacity), schedule information, and trip information (including travel time and distance) for various common modes of transportation. Based on the above information and necessary assumptions, and following the basic principle of minimizing travel time, travel chain information dominated by different modes of transportation can be obtained. Each travel chain includes information such as travel time, travel distance, travel cost, number of trips, and passenger capacity. Necessary assumptions include: in a combined travel network dominated by any mode of transportation, the effective path between any two points in the network is the path with the shortest travel time and the fewest transfers between the dominant modes of transportation; when a travel path consists of multiple modes of transportation, the transfer time between different modes of transportation is assumed to be a fixed value; the passenger capacity and total number of trips involved in any travel chain should be equal to the mode of transportation with the smallest passenger capacity and total number of trips in the travel chain; the theoretical "capacity and total number of trips" of private transportation in a broad sense (such as private cars, ride-hailing services, taxis, etc.) are assumed to be infinite; in travel chains where railways (such as intercity railways, high-speed railways, and ordinary railways, excluding urban rail transit) are the dominant mode of transportation, passengers generally do not need to transfer within the dominant mode of transportation.

[0008] Step 2: Point reachability calculation based on improved Mahalanobis distance

[0009] Using the travel chain data obtained in step 1 as input data, after dimensionality reduction processing, the inter-point reachability between any two points in the network is calculated according to the Mahalanobis distance formula.

[0010]

[0011] In the formula, MMD is the Mahalanobis distance calculated after data dimensionality reduction, i.e., the improved Mahalanobis distance; i and j are the start and end points of the travel chain, respectively, and satisfy (i∈I=J,j∈J,j≠i); m is the dataset number classified according to the dominant method; x is the data vector of the target travel chain; μ is the mean of the benchmark dataset. In this method, the benchmark dataset should preferably be a travel chain dataset that relies solely on urban rail transit.

[0012] Step 3: Node reachability calculation based on anomaly impact factors

[0013] Using the point-to-point reachability obtained in step 2 as input data for step 3, the node reachability in the travel network m for different dominant modes of transportation is calculated based on the anomaly impact factor algorithm, including:

[0014] 3.1 Delete one-way trip chain data to ensure data flatness.

[0015] 3.2 Determine the number N of neighboring points within the neighborhood of target node i in network m dominated by different travel modes. (i-nearest,m) :

[0016]

[0017] In the formula, N m Let m be the total number of nodes in network m. Additionally, let P be the set of points i and their neighbors. (i-nearest,m) All points have a "distance" smaller than the nearest neighbor distance (MMD). (i-nearest,m) The latter is usually the distance between point i and its farthest neighbor.

[0018] 3.3 In network m, based on nearest neighbor distance (MMD) (i-nearest,m) Determine the reachable distance RMMD from point i to any other point j. (i,j,m) :

[0019] RMMD (i,j,m) =max{MMD (i-nearest,m) MMD (i,j,m)} (3)

[0020] 3.4 Determine the Locally Reachable Density (LRD) of point i in network m (i,m) :

[0021] LRD (i,m) =1 / mean(∑ j∈J＝I,j≠i RMMD (i,j,m) (4)

[0022] 3.5 Determine the reachability (LOF) of each sample point i in network m. (i,m) :

[0023]

[0024] In step 3, in the network m dominated by any mode of travel, the default value for the reachability of non-connected nodes is infinity.

[0025] Step 4: Calculate the overall reachability of nodes

[0026] Based on the node reachability of different dominant transportation modes in step 3, the overall node reachability of the network (composed of superimposed different transportation networks) is calculated, including:

[0027] 4.1: Normalize the reciprocal of the reachability of each node i in each dominant travel mode network m, n (i,m) The normalized value:

[0028]

[0029] 4.2: Determine the entropy h of each dominant travel mode network m. m :

[0030]

[0031] 4.3: Calculate the weights w of each dominant travel mode network m. m :

[0032]

[0033] 4.4: Based on w m The overall reachability index SLOF of computing node i i :

[0034]

[0035] Step 5: Identification of the actual boundaries of metropolitan areas or urban clusters

[0036] Based on the SLOF (Synthetic Accessibility Index) of multimodal travel network nodes obtained in step 4 i The absolute median deviation method is used to screen nodes with poor (or good) accessibility. Furthermore, the actual boundaries of metropolitan areas or urban clusters are depicted based on these nodes with poor (or good) overall accessibility.

[0037] MAD = b * median(|SLOF) i -median(SLOF i (10)

[0038] U = median(SLOF) i )+T×MAD (11)

[0039] L = median(SLOF) i )-T×MAD (12)

[0040] In the formula, b is a scaling factor constant, typically taken as 1.4826; U and L are the upper and lower thresholds for judging the overall reachability, respectively. When the overall reachability of a node is greater than U (or less than L), it is considered a node with poor reachability; T is the recognition accuracy, typically taken as 2, 2.5, or 3.

[0041] Step 6: Result Validation Based on Functional Multicenter

[0042] Step 6 primarily utilizes external resident activity data (such as work-residence distribution and travel demand) to calculate the functional multicentrality of the metropolitan area (or urban agglomeration) obtained in Step 5, and compares it with the functional multicentrality of the larger region in which the metropolitan area (or urban agglomeration) is located, thus achieving the purpose of verification. The functional multicentrality is calculated using the following formula:

[0043]

[0044]

[0045] In the formula, ρF σ represents functional multicentrism; Δ represents network density; σ represents functional multicentrism. F σ is the standard deviation of the degree of each node in the network; Fmax L is the standard deviation of the maximum degree in a two-node network (a two-node network consists of nodes with non-zero degree and nodes with zero degree); L is the number of connected edges in the network; L max The theoretical maximum number of edges in a network.

[0046] The result is reliable if the functional multicentrality of the metropolitan area (or urban agglomeration) obtained in step 5 is greater than the functional multicentrality of the larger area in which the metropolitan area (or urban agglomeration) is located, as determined in step 5. Otherwise, the identification scope is too small and fails to encompass the entire metropolitan area (or urban agglomeration).

[0047] The method of this invention can be applied to urban planning and transportation planning departments and engineering consulting fields related to urban planning or city clusters.

[0048] This invention presents a novel calculation system for point-to-point accessibility, node accessibility, and comprehensive node accessibility based on emerging big data analytics technologies. Based on this system, it proposes a simple, efficient, accurate, and easy-to-understand method for identifying the actual boundaries of metropolitan areas or urban clusters. This method has profound theoretical and practical significance for the application of related big data technologies in urban and transportation planning and engineering. Attached Figure Description

[0049] Figure 1 These are the overall steps provided in the embodiments of the present invention, namely, the method for identifying the actual boundaries of metropolitan areas or urban clusters.

[0050] Figure 2 This is the result of the actual boundary identification in an embodiment of the present invention. Note: Due to the low network density, the functional multicenter of the entire Yangtze River Delta region ("three provinces and one municipality") is only about 0.0001.

[0051] Figure 3 This is the verification result of the identification of the actual boundary of the metropolitan area or urban agglomeration in the embodiments of the present invention. *The identification results based on the evening peak hour data show that the Hangzhou metropolitan area has been integrated into the urban agglomeration represented by Shanghai. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0053] The application principle of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] like Figure 1As shown, this embodiment of the invention provides a method for identifying the actual boundaries of metropolitan areas or urban clusters in the context of multimodal transportation network travel.

[0055] The application principle of the present invention will be further described below with reference to specific embodiments.

[0056] Calculation example:

[0057] The method for identifying the actual boundaries of metropolitan areas or urban clusters provided in this embodiment of the invention uses data from the Yangtze River Delta region (Shanghai, Jiangsu, Zhejiang, and Anhui) during the entire day, morning peak (6:00-10:00), and evening peak (16:00-20:00) periods as an example for calculation.

[0058] (1) Acquire multi-source travel information data and perform necessary preprocessing.

[0059] Based on data from the internet regarding road networks, national railways, urban rail transit, and long-distance buses within the Yangtze River Delta region, all possible travel chain information is obtained by piecing together network trips from different modes of transportation. The data overview used in this embodiment and the travel chain data set obtained through piecing together are shown in Tables 1 and 2.

[0060] Table 1 Overview of Basic Data

[0061]

[0062] a: The fuel consumption is uniformly taken as 0.6 CNY / km.

[0063] b: Includes three modes: intercity railway, high-speed railway, and conventional railway.

[0064] c: The passenger capacity of urban rail transit is uniformly taken as the capacity of the most common train model, approximately 2,046 people.

[0065] d: Only bus routes to major stations are included; routes to remote rural areas are not included. Furthermore, long-distance buses are generally round-trip routes, but only one-way data is available for some routes between origin and destination stations. Considering that the stations with missing data are often small stations at the township, town, or district level, for the sake of seriousness, only 122,099 bus routes were ultimately used in the study.

[0066] e: The passenger capacity of long-distance buses is uniformly based on the most common model, approximately 45 people.

[0067] f: Some stations simultaneously serve multiple functions, including railway stations, urban rail transit stations, and long-distance bus stations.

[0068] g: The user's transfer time within the urban rail transit system is assumed to be 5 minutes, and all other transfer times are assumed to be 10 minutes.

[0069] Table 2. Classification and Data Scale of the Consolidated Travel Chain

[0070]

[0071]

[0072] *: This refers to the number of travel chains available at different times.

[0073] Table 2 contains information for each travel chain, including: origin and destination, travel time, travel distance, travel cost, total number of services (all day, morning peak, evening peak), passenger capacity (all day, morning peak, evening peak), and number of transfers between urban rail transit and non-urban rail transit.

[0074] (2) Dimensionality reduction is performed on the travel chain data, and point accessibility based on improved Mahalanobis distance is calculated using Equation (1).

[0075] Principal component analysis was performed on the urban rail transit (single mode) dataset in Table 2, and all other datasets were dimensionality reduced based on the urban rail transit (single mode) dataset in Table 2. The inter-point reachability between service stations in Table 1 was calculated by Equation (1).

[0076] In this embodiment, the principal component analysis determined the number of principal components based on a variance contribution rate of no less than 90%, and the number and coefficients of principal components for each travel period are shown in Table 3. Due to the limited data used in this embodiment, only one principal component could be extracted for each period.

[0077] Table 3. Principal component analysis results

[0078] Travel time Trip Time Travel distance Trip Expenses Total number of trains Staffing Transfer time Morning rush hour 0.0000 -0.0007 0.1383 0.9853 -0.0995 0.0092 Evening rush hour 0.0000 -0.0001 0.1383 0.9853 -0.0995 0.0092 All day 0.0000 -0.0001 0.1383 0.9853 -0.0995 0.0092

[0079] (3) Node reachability calculation

[0080] Calculate the node reachability of each node in the network m dominated by different travel modes based on equations (2) to (5).

[0081] (4) Calculation of node comprehensive reachability

[0082] Based on equations (6) to (9), calculate the node comprehensive reachability of a multi-mode transportation network with different travel modes superimposed on network m.

[0083] (5) Identification of the physical boundaries of metropolitan areas or urban clusters

[0084] Based on equations (10) to (12), the overall accessibility of each node in the multimodal transportation network is judged. Based on this, the poorly accessible points on the edge of the region not covered by points with poor overall accessibility are connected sequentially, thus outlining the actual boundary of the metropolitan area or urban agglomeration, such as... Figure 2 As shown.

[0085] (6) Validation of the results of the physical boundary identification

[0086] Calculate the functional multicentrality of the metropolitan area (or urban agglomeration) obtained in step 5 and the functional multicentrality of the larger area in which these metropolitan areas (or urban agglomerations) are located, respectively, according to equations (13) and (14), and compare them. If the functional multicentrality of the metropolitan area (or urban agglomeration) obtained in step 5 is greater than the functional multicentrality of the larger area in which these metropolitan areas (or urban agglomerations) are located, the result is valid. Otherwise, it indicates that the identification range is too small and fails to encompass the entire metropolitan area (or urban agglomeration), requiring an expanded data collection range and a return to step 1 for re-identification. Figure 3 It can be seen that the functional multicentrality of each metropolitan area or urban agglomeration identified in the example is greater than the corresponding value (0.0001) of the larger region ("three provinces and one municipality"), and the result is valid.

Claims

1. A method for identifying the actual boundaries of metropolitan areas or urban agglomerations based on multimodal transportation networks, characterized in that: Includes the following steps: Step 1: Acquisition and preprocessing of multi-source travel information data; Step 2: Point reachability calculation based on improved Mahalanobis distance; including: Multi-source data dimensionality reduction and point-to-point reachability calculation based on principal component analysis; Specifically, the travel chain data obtained in step 1 is used as input data. After dimensionality reduction processing, the inter-point reachability between any two points in the network is calculated according to the Mahalanobis distance formula. (1) In the formula, The Mahalanobis distance is calculated after data dimensionality reduction, i.e., the improved Mahalanobis distance; , These are the start and end points of the travel chain, and they satisfy... ; The dataset is numbered according to the dominant classification method; The data vector for the target travel chain; The mean of the benchmark dataset, which uses a travel chain dataset that relies solely on urban rail transit; Step 3: Node reachability calculation based on anomaly impact factors; specifically: Using the point-to-point reachability data obtained in step 2 as input data for step 3, the travel network for different dominant modes of transportation is calculated based on the anomaly impact factor algorithm. Node reachability in the system includes: 3.1 Delete one-way trip chain data to ensure data flatness; 3.2 Identify the dominant networks for different modes of transportation In the middle, the target node Number of neighboring points within the neighborhood : (2) In the formula, For the network The total number of nodes; points Its neighbor set The "distance" of all points is less than the nearest neighbor distance. The latter is usually a point Distance to the farthest neighbor; 3.3 On the network In the middle, based on nearest neighbor distance Determined point to any other point accessible distance : (3) 3.4 Determine the network various points in the middle Locally accessible density : (4) 3.5 Determine the network Each sample point Node reachability : (5) In step 3, the network dominated by any mode of transportation In this context, the default value for the reachability of non-connected nodes is infinity. Step 4: Calculate the overall reachability of nodes; specifically: Based on the travel networks of different dominant modes of transportation in step 3 The node reachability calculation in the network assesses the overall node reachability of the entire network, including: 4.1: Networks of each dominant travel mode Each node in Normalize the reciprocal of the reachability. The normalized value: (6) 4.2: Determine the networks of each dominant travel mode entropy : (7) 4.3: Calculate the networks of each dominant travel mode weight : (8) 4.4: Based on compute nodes Comprehensive accessibility index : (9)； Step 5: Identification of the actual boundaries of metropolitan areas or urban clusters; Step 6: Validation of results based on functional multicenters.

2. The method as described in claim 1, characterized in that, Step 1, the acquisition and preprocessing of multi-source travel information data, includes: Based on the combination of origin and destination information, ticketing information, vehicle information, schedule information, and itinerary information of various common modes of transportation, feasible combination travel chain information of users in the context of multi-modal transportation is obtained; For private cars, ride-hailing services, and taxis, which are modes of private transportation, their vehicle information and schedule information are not considered, and the corresponding default values ​​are set to infinity. Preprocessing, based on the above information, follows the basic principles of minimizing travel time and minimizing transfers between the main modes of transportation. It combines travel information with necessary assumptions to obtain complete travel chain data.

3. The method as described in claim 1, characterized in that, In step 5, the comprehensive reachability index of multimodal travel network nodes obtained in step 4 is used. The absolute median deviation method is used to screen nodes with poor or good accessibility. Furthermore, based on these nodes with poor or good overall accessibility, the actual boundaries of metropolitan areas or urban clusters are depicted. Specifically: (10) (11) (12) In the formula, The scaling factor constant is taken as 1.4826; , These are the upper and lower thresholds for judging the overall accessibility quality; When the overall reachability of a node is greater than or less These are points with poor node reachability; For accuracy, we use 2, 2.5, and 3.

4. The method as described in claim 1, characterized in that, Step 6 uses external resident activity data to calculate the functional multicentrality of the metropolitan area or urban agglomeration obtained in Step 5, and compares it with the functional multicentrality of the larger area in which the metropolitan area or urban agglomeration is located, to achieve the purpose of verification.

5. The method as described in claim 4, characterized in that, The functional multicenter is calculated using the following formula: (13) (14) In the formula, For functional polycentricity; Network density; represents the standard deviation of the degree of each node in the network; The maximum standard deviation of node degree in a two-node network is given by the standard deviation of node degree, which is composed of nodes with non-zero degree and nodes with zero degree. This represents the number of connected edges in the network. The theoretical maximum number of edges in a network; The result is reliable if the functional multicenter of the metropolitan area or urban agglomeration obtained in step 5 is greater than the functional multicenter of the larger area in which the metropolitan area or urban agglomeration is located; otherwise, the identification range is too small and fails to encompass the entire metropolitan area or urban agglomeration.

Images