An urban rail transit passenger flow corridor identification method based on OD clustering and application thereof

Abstract

The application belongs to the technical field of urban traffic planning, and in particular relates to a city rail transit passenger flow corridor identification method based on OD clustering and application thereof. The city rail transit passenger flow corridor identification method based on OD clustering performs OD point clustering processing based on prepared and preprocessed data, and proposes a unique clustering variable for OD clustering. The O point distance, D point distance and path coincidence degree are considered and a reasonable proportion is set, so that the influence factors of passenger travel behavior are more directly reflected. In the clustering process, the OD amount is taken as the weight of the clustering sample point, so as to ensure the accuracy and reliability of the OD clustering. In the clustering center point selection stage, the O point and the D point are selected as the two clustering center points. By distributing the passenger flow to different paths between the O point and the D point, the city rail transit passenger flow corridor is quickly and accurately identified, so as to provide a decision basis for the planning and layout of the rail transit line and the cross-line operation of the rail transit train.

Description

Technical Field

[0001] This invention belongs to the field of urban transportation planning technology, specifically relating to a method for identifying urban rail transit passenger flow corridors based on OD clustering and its application. Background Technology

[0002] Due to modernization, urban populations are gradually increasing, and urban areas are expanding outwards, leading to a growing phenomenon of "separation of work and residence." This phenomenon causes a rapid increase in passenger flow and concentrations of varying sizes. The randomness of passenger flow along the lines is enhanced, with passengers often concentrating at a particular station and then dispersing in different directions. The original single-line train operation model can no longer adequately meet the demands of this passenger flow. Barriers between lines severely reduce service levels and operational efficiency, necessitating relevant modifications. Therefore, many cities are now opting for cross-line operation. To better organize transportation and provide a basis for cross-line modifications, it is essential to accurately analyze passenger flow patterns and precisely identify urban rail transit passenger corridors.

[0003] Urban rail transit passenger corridors refer to the backbone lines in urban rail transit that connect major passenger origins and share a common flow direction. Their formation is based on the convergence of passenger flows from different directions towards the corridor due to its advantages. Urban rail transit passenger corridors connect major functional areas of the city, constructing a mature rail transit network and having a close, mutually reinforcing relationship with urban development. The ability to correctly identify rail transit passenger corridors directly relates to whether rail transit can effectively guide urban development, effectively distribute urban passenger flow, effectively plan land use along rail transit lines, and effectively balance the spatial layout along the lines, thus having significant practical implications. Furthermore, for existing lines, rail transit passenger corridors are an important basis for adjusting train operation plans. Summary of the Invention

[0004] This invention aims to provide a method for identifying urban rail transit passenger flow corridors based on OD clustering and its application, so as to provide guidance for urban rail transit planning and thus optimize urban traffic.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for identifying urban rail transit passenger flow corridors based on OD clustering is provided, including:

[0007] Data preparation and preprocessing: Based on the physical network information of rail transit and passenger entry and exit information, OD information is compiled, which includes OD station information and passenger flow information;

[0008] OD point clustering: Based on OD information, OD sites are clustered and divided into clusters. Each cluster contains multiple OD pairs. Then, the best clustering scheme is selected to obtain several O-point cluster centers and D-point cluster centers.

[0009] Identify passenger flow corridors in rail transit: Based on the passenger flow of the O-point cluster center and the D-point cluster center within each cluster, the passenger flow is distributed to each path between the O-point cluster center and the D-point cluster center within that cluster, thereby realizing the identification of passenger flow corridors.

[0010] Preferably, the OD station information includes the station names for passengers entering and exiting the station and their corresponding time periods, the shortest path between the OD stations, and the walking distance required for passengers to complete a journey between two OD stations. The passenger flow information includes passenger flow data for each OD station during morning and evening peak hours and throughout the day.

[0011] Preferably, the clustering and grouping of OD sites based on OD information includes:

[0012] Different OD pair information is obtained based on the OD information, wherein the OD pair information includes O point information, D point information and OD quantity;

[0013] Using the walking distance required for passengers to complete trips between points O, the walking distance required for passengers to complete trips between points D, and the path overlap between OD pairs as clustering variables, a distance matrix dist is constructed between OD pairs:

[0014]

[0015] Where d ij This represents the distance between the i-th and j-th OD pairs calculated after standardization and weighting of the clustering variable results. This distance covers the walking distance required for passengers to complete travel between points O, the walking distance required for passengers to complete travel between points D, and the path overlap between OD pairs.

[0016] Set the range of values ​​for the clustering neighborhood radius eps and the minimum number of cluster points MinPts. Use the distance matrix as the clustering variable to measure distance and the OD quantity as the variable to measure the weight of sample points. Use the DBSCAN method to perform clustering processing on each O point and D point to obtain multiple clustering schemes.

[0017] Determine the cluster center point for each cluster in each clustering scheme;

[0018] Then select the best clustering scheme.

[0019] Preferably, the walking distance required for travel between points O, the walking distance required for travel between points D, and the path overlap between OD pairs are calculated as follows:

[0020] Query the information of the i-th and j-th OD pairs, and record the walking distance required for the passenger to complete the journey between points O. The walking distance required to complete the journey between point D and point D.

[0021] Calculate the path overlap between the i-th OD pair and the j-th OD pair:

[0022]

[0023] Among them, P ij S represents the number of stations repeatedly traversed by the OD path between the i-th and j-th OD pairs, where the OD path is the shortest path between the OD pairs. i S represents the number of stations traversed by the OD path of the i-th OD pair. j This represents the number of stations traversed by the OD path of the j-th OD pair.

[0024] Preferably, the standardization and weighting of the calculation results for each clustering variable includes:

[0025] Dimensionless processing of the distance:

[0026] The processed walking distance between points O is:

[0027] in S represents the average walking distance between points O. O This represents the standard deviation of the walking distance between points O;

[0028] The processed walking distance between points D is as follows:

[0029] in S represents the average walking distance between points D. D This represents the standard deviation of the walking distance between points D;

[0030] Inverse processing of path overlap between OD pairs: γ ij =1-α ij ;

[0031] Integrating clustering variables: Each standardized clustering variable was assigned a weight of 25%, 25%, and 50% respectively to obtain the final variable d representing the distance between OD pairs. ij :

[0032]

[0033] Preferably, determining the cluster center point of each cluster in each clustering scheme includes:

[0034] Cluster centers are selected based on the principle of minimizing the sum of walking distances from a given point to all points within the same cluster and maximizing the proportion of origin-destination (OD) values. Within each cluster, two cluster centers, O and D, are determined, where:

[0035] Selection of O-point cluster center points:

[0036]

[0037] Where, d O (p i ,p j V represents the walking distance required to travel between points O in the i-th and j-th OD pairs. i Let m represent the OD value of the i-th OD pair, and m represent the number of sample points within the cluster.

[0038] Selection of cluster center at point D:

[0039]

[0040] Where, d D (p i ,p j V represents the walking distance required to travel between point D in the i-th and j-th OD pairs. i This represents the OD value of the i-th OD pair.

[0041] Preferably, the selection of the optimal clustering scheme includes:

[0042] The compactness index (CP), spacing index (SP), and Davidson-Bolding index (DBI) were used to evaluate the various clustering schemes, among which:

[0043] The cluster compactness index (CP) reflects the average distance within a cluster, and is calculated as follows:

[0044]

[0045]

[0046] In the above formula, CP l Ω represents the compactness of the l-th cluster, n represents the number of OD pairs within the l-th cluster, and Ω represents the density of the cluster. l Let μ represent the set of OD pairs within the l-th cluster. O O represents the cluster center. μ represents the walking distance required to travel between point O and the cluster center of point O for the i-th OD pair. D D represents the cluster center. σ represents the walking distance required to travel between point D and the cluster center of point D in the i-th OD pair; mLet represent the compactness of the m-th scheme, and k represent the number of clusters in the m-th scheme;

[0047] SP reflects the distance between different clusters, and is calculated as follows:

[0048]

[0049] in, This represents the walking distance required to travel between the i-th cluster and the j-th cluster, specifically the cluster center O. This represents the walking distance required to travel between the cluster centers at point D;

[0050] DBI represents the maximum sum of the average intra-cluster distances between any two clusters divided by the distance between the cluster centers of the two clusters, calculated as follows:

[0051]

[0052] Where k represents the number of clusters, CP i CP represents the compactness of the i-th cluster. j This indicates the compactness of the j-th cluster;

[0053] The weight of each indicator is determined by the entropy weight method, and the comprehensive score of each scheme is calculated to obtain the optimal clustering scheme.

[0054] Preferably, the OD pair information is stored in a constructed two-dimensional table.

[0055] The present invention also provides an application of the above-described OD clustering-based urban rail transit passenger flow corridor identification method in urban rail transit passenger flow corridor identification.

[0056] Compared with existing technologies, the beneficial effects of this invention are: 1. This method for identifying urban rail transit passenger flow corridors based on OD clustering performs OD point clustering based on prepared and preprocessed data, thereby achieving rapid and accurate identification of urban rail transit passenger flow corridors and providing a decision-making basis for the planning and layout of rail transit lines and the cross-line operation of rail transit trains. 2. This method for identifying urban rail transit passenger flow corridors based on OD clustering proposes unique clustering variables for OD clustering, taking into account the distance between O points, the distance between D points, and the path overlap, and setting reasonable weights, thus more intuitively reflecting the influencing factors of passenger travel behavior. 3. This method for identifying urban rail transit passenger flow corridors based on OD clustering uses the OD value as the weight of the clustering sample points during the clustering process, ensuring the accuracy and reliability of OD clustering. 4. In the OD clustering-based method for identifying urban rail transit passenger flow corridors, two cluster centers, O and D, are selected during the cluster center selection stage. These centers have practical physical meaning. By distributing passenger flow to different paths between the two cluster centers, O and D, the method can quickly and accurately identify urban rail transit passenger flow corridors. Attached Figure Description

[0057] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0058] Figure 1 This is a flowchart of an embodiment of the urban rail transit passenger flow corridor identification method based on OD clustering of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] In one embodiment, a method for identifying urban rail transit passenger flow corridors based on OD clustering is provided, such as... Figure 1 As shown, the urban rail transit passenger flow corridor identification method based on OD clustering includes the following steps:

[0061] Step S1: Data Preparation and Preprocessing

[0062] This method for identifying urban rail transit passenger flow corridors based on OD clustering requires the following data: urban rail transit physical network information and passenger entry / exit information. The physical network information mainly includes the shortest path between stations in the urban rail transit network, the walking distance dist(i,j) required for a passenger to travel between two stations, and the line to which each station belongs. The passenger entry / exit information includes the passenger's departure station name, destination station name, and departure time t. ij .

[0063] The OD information and traffic volume information are compiled. O (Origin) represents the departure station, and D (Destination) represents the destination. OD information includes OD station information and passenger flow information. Specifically, based on the previously obtained rail transit physical network information and passenger entry and exit information, the names of passenger entry and exit stations and their corresponding departure times are extracted. Data with the same OD station are merged to obtain passenger flow data for each OD station during morning and evening peak hours and throughout the day. Based on the rail transit physical network information, the shortest path between two OD stations and the walking distance required for passengers to complete the journey between two OD stations are calculated, and the data is integrated.

[0064] Step S2: OD point clustering processing

[0065] This step involves clustering OD sites based on OD information and selecting the optimal clustering scheme to obtain several O-point cluster centers and D-point cluster centers, thus providing a basis for the identification of traffic flow corridors.

[0066] Step S21: Based on the OD information integrated in the previous steps, obtain different OD pair information. OD pair information includes O point information, D point information, and OD quantity. An OD pair includes O point (departure station) and D point (arrival station). There may be multiple paths between OD pairs. OD pair information is stored in a two-dimensional list. O point information is the passenger departure station information, D point information is the passenger arrival station information, and OD quantity is the passenger flow between O point and D point. Table 1 below is an example of a two-dimensional list storing OD pair information.

[0067] Table 1 - Summary Table of Passenger Entry and Exit Information

[0068]

[0069] As shown in the two-dimensional list 1 above, the first and second columns represent the passenger's entry station number and exit station number, respectively. The third column represents the walking distance required for a passenger to complete a trip from the entry station to the exit station. The fourth column represents the shortest path between the entry and exit stations, i.e., the OD path. The fifth column represents the OD volume (passenger flow), i.e., the passenger flow from point O to point D. The above data is preprocessed to remove erroneous and redundant information, such as data with the same entry and exit station numbers, and data where both entry and exit stations belong to the same line.

[0070] Step S22: Select appropriate clustering variables and construct the distance matrix between OD pairs.

[0071] Unlike other clustering methods, in the OD passenger flow clustering of urban rail transit, the clustering variables should not only include physical distance, but also path overlap. OD pairs clustered into the same category should have close O-point distance, D-point distance and high path overlap.

[0072] In this step, the walking distance required for passengers to complete trips between points O, the walking distance required for passengers to complete trips between points D, and the path overlap between OD pairs are used as clustering variables to construct the distance matrix dist between OD pairs. The method for constructing the distance matrix dist between OD pairs is as follows:

[0073] Step S221: Query the i-th and j-th OD pairs in the two-dimensional table storing OD pair information, and find the walking distance required for the passenger to complete the journey between points O. The walking distance required to complete the journey between point D and point D.

[0074] Step S222: Query the OD path for each OD pair in the two-dimensional table, and calculate the path overlap between the i-th OD pair and the j-th OD pair as follows:

[0075]

[0076] Among them, P ij S represents the number of stations repeatedly traversed by the OD paths of the i-th and j-th OD pairs. i S represents the number of stations traversed by the i-th OD path. j This represents the number of stations traversed by the j-th OD path.

[0077] Step S223: Standardize the calculation results of the clustering variables and assign weights to each clustering variable.

[0078] This step mainly involves dimensionless processing of distances and reverse processing of path overlap. Compared with path overlap, the walking distances required for travel between points O and D are significantly larger and of different magnitudes. Therefore, it is necessary to dimensionless process the walking distances required for travel to unify data of different magnitudes into the same magnitude, ensuring comparability between data. Step S223 further includes:

[0079] Step S2231: Dimensionlessly process the walking distances required for travel between points O and D.

[0080] The specific processing method is as follows:

[0081] The processed walking distance required for travel between points O is:

[0082] in S represents the average walking distance required to travel between points O. O This represents the standard deviation of the walking distance required to travel between points O.

[0083] The processed walking distance required for travel between points D is:

[0084] in S represents the average walking distance required to travel between points D and D. D This represents the standard deviation of the actual walking distance required to travel between points D.

[0085] Step S2232: Path overlap degree reverse processing:

[0086] As mentioned earlier, a higher path overlap value indicates a greater proportion of OD path overlap between two OD pairs, making them more likely to be clustered into one class in cluster analysis—a positive indicator. However, all clustering variables in cluster analysis should be inverse indicators; the lower the value of a clustering variable, the easier it is for them to be clustered into one class. Therefore, we performed an inverse processing on the path overlap value, with the following results:

[0087] The result after path overlap processing is: γ ij =1-α ij .

[0088] Step S2233: Integrate the clustering variables and assign weights to obtain the final distance variables:

[0089] Distance and path overlap are equally important, so we assign weights of 25%, 25%, and 50% to distance and path overlap, respectively, resulting in the following final distance variable results:

[0090]

[0091] in This represents the standard value indicating the walking distance required to travel between points O. γ represents the standard value of the walking distance required to travel between points D. ij Standard values ​​representing path overlap.

[0092] Step S224: Calculate the values ​​of the distance variable between OD pairs to obtain the final distance matrix for clustering, represented as:

[0093]

[0094] Where d ij This represents the distance between the i-th and j-th OD pairs after standardization. This distance encompasses the distance between points O, the distance between points D, and path overlap. This matrix is ​​a real symmetric matrix with zeros on the diagonal and is the final distance matrix used for clustering. M and K represent other distance quantities not shown in this distance matrix.

[0095] Taking the OD data of the morning rush hour between two OD pairs on February 21, 2022 as an example, the calculated distance matrix is ​​as follows:

[0096]

[0097] Step S225: Cluster the OD pairs using the DBSCAN method

[0098] Unlike other DBSCAN clustering methods, in OD-based DBSCAN clustering, the distance between clusters is integrated into the distance matrix dist obtained in step S224, and the weight of each sample point is determined by the OD value. During DBSCAN clustering, the ranges for the cluster neighborhood radius eps and the minimum number of cluster points MinPts are pre-defined. The distance matrix is ​​used as the clustering variable to measure the distance between OD pairs, and the OD value is used as the variable to measure the weight of each sample point, resulting in multiple clustering schemes.

[0099] In the DBSCAN clustering process, based on the input distance matrix dist and the pre-set range of clustering neighborhood radius eps and minimum cluster size MinPts, all OD pairs are clustered into multiple pre-selected sample points by identifying the O and D points. Finally, the sample points are determined based on the OD values; O or D points that meet certain OD values ​​are identified as sample points. Each cluster contains multiple O points and multiple D points that satisfy the distance range. The DBSCAN clustering output includes the number of clusters, the number of sample points within each cluster, the number of outliers, and the cluster number to which each sample point belongs.

[0100] For example, based on the distance matrix dist calculated above, the neighborhood radius is selected to be in the range of [0.3, 0.8], and the minimum number of cluster points is selected to be in the range of [850, 1200]. After multiple iterations, different clustering schemes (partial) are obtained as shown in the table below.

[0101] Table 2 Clustering Schemes

[0102]

[0103] The first column, eps, represents the neighborhood radius; the second column, min_samples, represents the minimum number of cluster points; the third column, n_clusters, represents the number of clusters; the fourth column, outlines, represents the number of noise points; and the fifth column, stats, represents the number of elements within each cluster. After selecting a clustering scheme, the category number of each OD pair can be obtained. The OD information and category numbers for some OD pairs are shown in the table below.

[0104] Table 3 - List of Partial OD Pair Information and Category Numbers

[0105] origin destination Distance volume cluster S01002 S10029 910 375 0 S01002 S10026 1273 127 0 S01002 S10025 1376 134 0 S01003 S02002 1699 136 -1 S01003 S02004 1527 174 -1 S01005 S02002 1429 109 1 S01005 S02004 1257 204 1

[0106] The first column, origin, represents the departure station number; the second column, destination, represents the arrival station number; the third column, distance, represents the walking distance required for a passenger to complete a trip between the arrival station and the departure station; the fourth column, volume, represents the OD quantity; and the fifth column, cluster, represents the OD pair's category number. When the value of this column is -1, it indicates that the OD pair belongs to noise. OD pairs with the same value in this column belong to the same category.

[0107] Step S226: Determine the cluster center point of each cluster in each clustering scheme.

[0108] During DBSCAN clustering, the cluster radius and minimum number of cluster points are continuously adjusted, resulting in multiple clustering schemes. It is necessary to evaluate each scheme to find the optimal one. Before finding the optimal scheme, the cluster centroids for each cluster in each scheme need to be determined. The selection principle for cluster centroids is to minimize the sum of walking distances required to travel from that point to all points within the same cluster, and to ensure that the point's origin-destination (OD) value accounts for a large proportion. Each cluster has two cluster centroids, O and D. The selection of cluster centroids includes:

[0109] Step S2261: Selection of cluster centers at point O:

[0110]

[0111] Where, d O (p i ,p j V represents the walking distance required to travel between points O in the i-th and j-th OD pairs. i Let represent the OD quantity of the i-th OD pair. Each OD pair includes multiple O points and multiple D points. There are multiple paths between OD pairs. The OD quantity refers to the shortest path between OD pairs. m represents the number of sample points within the cluster. This represents the percentage of the total OD value of all points except point i. The smaller this value, the larger the percentage of the OD value of point i.

[0112] Step S2262: Selection of cluster center at point D:

[0113]

[0114] Where, d D (p i ,p j Let represent the walking distance required for travel between point D in the i-th OD pair and the j-th OD pair.

[0115] For example, the cluster centers of some different schemes are shown in Table 4 below.

[0116] Table 4 - List of cluster centers for some different schemes

[0117]

[0118] The first column represents the number of different schemes, the second column eps represents the neighborhood radius, the third column min_samples represents the minimum number of cluster points, the fourth column represents the number of clusters, the fifth column represents the number of noise points, the sixth column represents the number of O-point cluster centers, and the seventh column represents the number of D-point cluster centers. The number of O-point and D-point cluster centers is the same as the number of clusters.

[0119] Step S227: Select the best clustering scheme

[0120] As mentioned above, the values ​​of cluster radius and minimum number of cluster points are continuously adjusted during DBSCAN clustering. For multiple clustering schemes, it is necessary to evaluate each clustering scheme to find the optimal one.

[0121] In the evaluation of clustering schemes, compactness (CP), spacing (SP), and the Davidson-Bolding index (DBI) were used to evaluate each clustering scheme, among which:

[0122] The cluster compactness index (CP) reflects the average distance within a cluster, and is calculated as follows:

[0123]

[0124]

[0125] Among them, CP l Ω represents the compactness of the l-th cluster, n represents the number of OD pairs within the l-th cluster, and Ω represents the density of the cluster. l Let μ represent the set of OD pairs within the l-th cluster. O O represents the cluster center. μ represents the walking distance required to travel between point O and the cluster center of point O for the i-th OD pair. D D represents the cluster center. σ represents the walking distance required to travel between point D and the cluster center of point D in the i-th OD pair. m Let m represent the compactness of the m-th scheme, and k represent the number of clusters in the m-th scheme.

[0126] The interval index SP reflects the distance between different clusters, and is calculated as follows:

[0127]

[0128] in This represents the walking distance required to travel between the i-th cluster and the j-th cluster, specifically the cluster center O. Let represent the walking distance required to travel between the i-th cluster and the j-th cluster, and the cluster center at point D.

[0129] The Davidson-Bauer index (DBI) represents the maximum sum of the average intra-cluster distances between any two clusters divided by the distance between the cluster centers of the two clusters. The formula is as follows:

[0130]

[0131] Where k represents the number of clusters, CP i CP represents the compactness of the i-th cluster. j This indicates the compactness of the j-th cluster.

[0132] Then, the weight of each indicator is determined by the entropy weight method, the comprehensive score of each scheme is calculated, and the optimal clustering scheme is obtained.

[0133] We selected clustering schemes with fewer noise points and a moderate number of clusters, calculated the corresponding indicators for each scheme and assigned weights, and finally obtained the results shown in the table.

[0134] Table 5 - Evaluation Index Values ​​for Each Scheme

[0135] eps minpts Cluster number Noise count CP SP DBI comprehensive Option 1 0.47 850 13 230 386.94 2973.11 6.43 0.747 Option 2 0.46 990 13 310 467.95 3021.5 16.30 0.218 Option 3 0.46 920 16 283 415.77 3160.8 22.05 0.189 Option 4 0.47 960 10 259 430.82 3601.5 10.12 0.710 Option 5 0.48 850 10 206 425.96 3213.96 4.93 0.780 Option 6 0.48 1130 9 262 456.7 3531.20 10.09 0.625

[0136] In this dataset, CP represents the compactness of the cluster, SP represents the spacing, and DBI represents the Davidson-Borgin index. Spacing and the DBI index are negative indicators, while compactness is a positive indicator. The three indicators are standardized and weighted accordingly. The resulting composite data is shown in column nine. Based on the composite index results, cluster 5 is the optimal solution, with a cluster radius of 0.48, a minimum number of cluster points of 850, forming 10 clusters, and containing 206 noisy points.

[0137] Step S3: Identify rail transit passenger flow corridors based on clustering results.

[0138] After determining the optimal clustering scheme, multiple clusters are generated. Each cluster contains multiple OD pairs within a pre-defined clustering radius. Each cluster has one O-point cluster center and one D-point cluster center, resulting in several O-point and D-point cluster centers. The passenger flow to the O-point and D-point cluster centers within each cluster is calculated. Based on the generalized cost and selection probability of each path between the O-point and D-point cluster centers in a cluster, the passenger flow is allocated to each path between the O-point and D-point cluster centers. Information on each path between the O-point and D-point cluster centers and the corresponding passenger flow are obtained, enabling the identification of passenger flow corridors. Paths with the same passenger origin (O-point cluster center), a common destination (D-point cluster center), and high passenger flow (OD volume) are identified as passenger flow corridors.

[0139] Based on the clustering results, five passenger flow corridors during the morning peak hours were identified, as shown in Table 6 below.

[0140] Table 6 - List of Passenger Flow Corridors During Morning Peak

[0141] corridor Specific cross-line situation 1 Line 9 - Fangshan Line - Line 4: Changyang → National Library → Renmin University 2 Line 13 - Line 2: Wudaokou → Xizhimen → Jishuitan → Jianguomen 3 Line 1-Line 2: Yuquan Road → Fuxingmen → Changchun Street → Jianguomen 4 Lines 1-10: Huazhuang → Guomao → Beitucheng 5 Line S1-Line 6: Xiaoyuan → Pingguoyuan → Chegongzhuang

[0142] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for identifying urban rail transit passenger flow corridors based on OD clustering, characterized in that, include: Data preparation and preprocessing: Based on the physical network information of rail transit and passenger entry and exit information, OD information is compiled, where O represents the departure station and D represents the destination; OD information includes OD station information and passenger flow information. OD point clustering: Based on OD information, OD sites are clustered and divided into clusters. Each cluster contains multiple OD pairs. Then, the best clustering scheme is selected to obtain several O-point cluster centers and D-point cluster centers. Identifying passenger flow corridors in rail transit: Based on the passenger flow of the O-point cluster center and the D-point cluster center within each cluster, the passenger flow is distributed to each path between the O-point cluster center and the D-point cluster center within that cluster, thereby realizing the identification of passenger flow corridors; The clustering and grouping of OD sites based on OD information includes: Different OD pair information is obtained based on the OD information, wherein the OD pair information includes O point information, D point information and OD quantity; Using the walking distance required for passengers to complete trips between points O, the walking distance required for passengers to complete trips between points D, and the path overlap between OD pairs as clustering variables, a distance matrix between OD pairs is constructed. : ; in This indicates the calculation of the first clustering variable after standardization and weighting. The first OD pair and the first The distance between each OD pair includes the walking distance required for a passenger to complete a journey between points O, the walking distance required for a passenger to complete a journey between points D, and the path overlap between OD pairs; Set the cluster neighborhood radius and minimum number of cluster points The range of values ​​is determined by using the distance matrix as the clustering variable to measure distance and the OD value as the variable to measure the weight of sample points. The DBSCAN method is used to perform clustering processing on each O point and D point to obtain multiple clustering schemes. Determine the cluster center point for each cluster in each clustering scheme; Then select the best clustering scheme.

2. The urban rail transit passenger flow corridor identification method based on OD clustering according to claim 1, characterized in that, The OD station information includes the station names for passenger entry and exit and their corresponding time periods, the shortest path between OD stations, and the walking distance required for passengers to complete a journey between two OD stations. The passenger flow information includes passenger flow data for each OD station during morning and evening peak hours and throughout the day.

3. The urban rail transit passenger flow corridor identification method based on OD clustering according to claim 1, characterized in that, The walking distance required for travel between points O, the walking distance required for travel between points D, and the path overlap between OD pairs are calculated as follows: Query No. The first OD pair information and the first Each O-point pair records the walking distance required for a passenger to complete a journey between O-points. The walking distance required to complete the journey between point D and point D. ; Calculate the first The first OD pair information and the first Path overlap between information pairs of OD pairs: ; in, Indicates the first The first OD pair and the first The number of stations repeatedly traversed by the OD path of an OD pair, where the OD path is the shortest path between the OD pairs. Indicates the first The number of sites traversed by the OD path of an OD pair Indicates the first The number of sites traversed by the OD path of an OD pair.

4. The urban rail transit passenger flow corridor identification method based on OD clustering according to claim 3, characterized in that, The standardization and weighting of the calculation results for each clustering variable include: Dimensionless processing of the distance: The processed walking distance between points O is: ; in This represents the average walking distance between points O. This represents the standard deviation of the walking distance between points O; The processed walking distance between points D is as follows: ; in This represents the average walking distance between points D. This represents the standard deviation of the walking distance between points D; Reverse processing of path overlap between OD pairs: ; Integrating clustering variables: Each standardized clustering variable was assigned a weight of 25%, 25%, and 50% respectively to obtain the final variable representing the distance between OD pairs. :

5. The urban rail transit passenger flow corridor identification method based on OD clustering according to claim 3, characterized in that, The determination of the cluster center point for each cluster in each clustering scheme includes: Cluster centers are selected based on the principle of minimizing the sum of walking distances to all points within the same cluster and maximizing the proportion of origin-destination (OD) values. Within each cluster, two cluster centers, O and D, are determined, where: Selection of O-point cluster center points: ; in, Indicates the first The first OD pair and the first The walking distance required for each OD to travel between its respective O points. Indicates the first The OD amount of each OD pair Indicates the number of sample points within a cluster; V j Indicates the first j The OD amount of each OD pair; Selection of cluster center at point D: ; in, Indicates the first The first OD pair and the first Each OD represents the walking distance required to travel between its respective D point. Indicates the first The OD amount of each OD pair; V j Indicates the first j The OD amount of each OD pair.

6. The urban rail transit passenger flow corridor identification method based on OD clustering according to claim 5, characterized in that, The selection of the optimal clustering scheme includes: The compactness index (CP), spacing index (SP), and Davidson-Bolding index (DBI) were used to evaluate the various clustering schemes, among which: The cluster compactness index (CP) reflects the average distance within a cluster, and is calculated as follows: ; ; In the above formula, Indicates the first The compactness of a cluster, Indicates the first The number of OD pairs within a cluster Indicates the first The set of OD pairs within a cluster O represents the cluster center. Indicates the first The walking distance required for travel between the O points and the O point cluster centers of an OD pair. D represents the cluster center. Indicates the first The walking distance required for travel between point D and the cluster center of point D for each OD pair; Indicates the first The tightness of the scheme, Indicates the first The number of clusters in each scheme; Indicates the first The starting point O of each OD pair; Indicates the first The endpoint D of each OD pair; SP reflects the distance between different clusters, and is calculated as follows: ; in, Indicates the first The cluster and the first The walking distance required to travel between the O-point cluster centers of each cluster. This represents the walking distance required to travel between the cluster centers at point D; Indicates the first The cluster center O of each cluster; Indicates the first The cluster center O of each cluster; Indicates the first The cluster center D of each cluster; Indicates the first The cluster center D of each cluster; DBI represents the maximum sum of the average intra-cluster distances between any two clusters divided by the distance between the cluster centers of the two clusters, calculated as follows: ; in, Indicates the number of clusters, Indicates the first The compactness of a cluster, Indicates the first The compactness of individual clusters; The weight of each indicator is determined by the entropy weight method, and the comprehensive score of each scheme is calculated to obtain the optimal clustering scheme.

7. The urban rail transit passenger flow corridor identification method based on OD clustering according to claim 1, characterized in that, The OD pair information is stored in a constructed two-dimensional table.

8. The application of the urban rail transit passenger flow corridor identification method based on OD clustering as described in any one of claims 1-7 in the identification of urban rail transit passenger flow corridors.

Images