A rail transit passenger feature evaluation method based on trajectory similarity

Abstract

The application discloses a kind of trajectory passenger feature evaluation methods based on trajectory similarity, comprising the following steps: S1, based on the site connection relationship in subway network, establish subway network as unweighted graph;S2, take all travel trajectory data of passenger in a period of time, obtain the station of each trajectory and the number of station on each trajectory path;S3, according to the similarity of each trajectory on the station, the trajectory similarity between all trajectories of the passenger is calculated, and the trajectory similarity matrix of the passenger is obtained;S4, the similarity matrix is further sorted and combined to obtain the travel rule matrix of the passenger;S5, the mathematical features of travel rule matrix are extracted, and the travel characteristics of the passenger are evaluated.According to the application, the regularity degree of passenger travel is more accurately reflected, the numerical index of passenger travel regularity judgment has great significance for urban rail transit big data analysis.

Description

Technical Field

[0001] This invention relates to the technical field of urban rail transit data, and in particular to a method for evaluating the characteristics of rail transit passengers based on trajectory similarity. Background Technology

[0002] Rail transit is the artery of a city, undertaking vital passenger transport and forming a crucial component of the urban public transportation system. During the operation of urban rail transit, operators collect and analyze passenger trajectory data to deduce and categorize individual or group travel patterns. Based on these patterns, they improve rail transit operation methods, optimize resource allocation, and enhance service satisfaction. The analysis of passenger travel trajectories to evaluate passenger characteristics in urban rail transit has received significant attention from relevant research institutions.

[0003] In urban rail transit, passenger characteristic evaluation typically serves a specific operational screening metric. For example, various cities in China are exploring new passenger classification and security screening technologies. A key function of Shanghai's research-and-development rail transit big data security screening system is passenger credit classification and screening. Based on passenger identity and trajectory data, and according to indicators such as the frequency and patterns of passenger subway trips, it analyzes subway passengers to identify those with high credibility, allowing them to pass through security quickly, thus improving passenger entry efficiency and alleviating peak-hour congestion.

[0004] Trajectory data is crucial for evaluating passenger characteristics. Current research on trajectory data includes trajectory similarity calculation and trajectory clustering. Existing trajectory data processing methods are often based on ground positioning data with trajectory point records, resulting in a relative lack of research on trajectories within subway networks and a lack of targeted optimization for subway network characteristics and passenger travel patterns. Studies on the regularity of passenger subway travel typically focus only on origin and destination, and when extracting regular features from trajectories, group analysis is usually used for passengers. Research on quantitatively evaluating individual travel trajectories and summarizing travel characteristics is relatively scarce. When classifying passenger characteristics, such as whether they are regular commuters, existing work mostly uses a combination of statistical and clustering methods, lacking numerical evaluation thresholds, making it difficult to achieve quantitative evaluation and comparison of different travel trajectories.

[0005] While subway networks can be represented graphically, travel trajectory data only records entry and exit stations, resulting in a very small amount of data compared to ground positioning. Based solely on entry and exit station records, it's difficult to effectively measure the similarity between different trajectories. For example, if a passenger's residence is between two stations, both stations could serve as the starting point for their daily travel trajectory, making it challenging to determine the true similarity between their trajectories based solely on entry and exit stations. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a rail transit passenger feature evaluation method based on trajectory similarity, which more accurately reflects the regularity of passenger travel patterns. This numerical indicator for judging passenger travel patterns is of great significance for big data analysis of urban rail transit. To achieve the above-mentioned objective and other advantages of this invention, a rail transit passenger feature evaluation method based on trajectory similarity is provided, comprising:

[0007] S1. Based on the station connection relationships in the subway network, establish the subway network as an unweighted graph;

[0008] S2. Collect all travel trajectory data of passengers within a certain period of time, and calculate the stations passed through by each trajectory and the number of stations on each trajectory path;

[0009] S3. Calculate the pairwise similarity between all the passenger's trajectories based on the similarity of the stations along each trajectory, and obtain the trajectory similarity matrix of the passenger.

[0010] S4. Further sort and merge the similarity matrix to obtain the passenger travel pattern matrix;

[0011] S5. Extract the mathematical features of the travel pattern matrix and evaluate the travel characteristics of passengers.

[0012] Preferably, the similarity calculation method in step S3 is as follows: the similarity between station A and station B is defined as sim(A,B), the similarity between a station and itself is sim(A,A)=1, the similarity between different stations is the reciprocal of the number of stations on the shortest route between the two stations, and the trajectory similarity is the average of the similarity of the stations along the two trajectories. If the two trajectories are the same, the similarity calculation result is 1. If the two trajectories have a complete inclusion relationship, the similarity calculation result is 1. The farther apart the two trajectories are, the closer the similarity calculation result is to 0.

[0013] Preferably, the trajectory similarity algorithm includes the following steps:

[0014] 1) Define two trajectory sequences, the longer trajectory sequence being T. L T L Includes L1 to L M There are M stations in total, and the shorter trajectory sequence is T. S T S Includes from S1 to S N There are N stations in total, and the trajectory sequence is denoted as: T L =[L1,L2,...,L M ],T S =[S1,S2,...,S N ], (M≥N);

[0015] 2) For shorter trajectories T s Each site S in n Calculate the relationship with another trajectory T L Each station L m similarity sim(L m ,S n And retain the maximum value:

[0016]

[0017] 3) The similarity between the two trajectories is

[0018] Preferably, in step S4, the travel pattern matrix can effectively display the travel patterns of passengers within that time period, including the number of pattern categories, the number of trajectories contained in each pattern category, and the similarity between patterns. The sorting process includes the following steps:

[0019] S41. Count the number of 1s in each row. The number 1 indicates that the similarity between the two trajectories corresponding to that position is 1. The more 1s there are in a row, the more similar trajectories the travel trajectory represented by that row number has. That is, the travel trajectory represented by this row number can be used as the core of a set of similar trajectories.

[0020] S42. Sort the trajectory numbers from high to low according to the number of 1s in each row. Rearrange the original trajectory similarity matrix according to the sorted number order. The output result is a symmetric matrix after swapping the rows and columns of the original matrix. The number of 1s in the new matrix decreases from the first row to the last row.

[0021] Preferably, the merging process in step S4 includes the following steps:

[0022] S411. Starting from the first row of the matrix, record the column numbers of all columns with a value of 1 in that row as a similar trajectory group. The core trajectory of the group is the trajectory represented by the row number. A trajectory can only be recorded in one similar trajectory group.

[0023] S412. Calculate the similarity matrix after merging using the core trajectories of the groups. The number of rows or columns of the merged matrix is ​​the number of similar trajectory groups. The similarity between trajectory groups is taken as the similarity value between the core trajectories of the groups.

[0024] S413. The similarity value is amplified by the number of trajectories in the trajectory group; for the merged similarity matrix, the value on the diagonal of the matrix is ​​the number of trajectories contained in the trajectory group;

[0025] S414. Output the similarity matrix of the trajectory group, where the diagonal is the number of trajectories contained in the similar trajectory group and the remaining positions are the similarity between the core trajectories. This matrix represents the travel pattern of the passenger.

[0026] Preferably, in step S5, by setting relevant requirements for regularity, the number of regular trips can be obtained by the sum of some elements of the trip pattern matrix, and the proportion of regular trips can be obtained by the ratio of the number of regular trips to the total number of trips.

[0027] Compared with existing technologies, the advantages of this invention are: by using the similarity of transit stations, it more accurately reflects the regularity of passenger travel patterns. The passenger trajectory similarity matrix can fully reflect the correlation between their trajectories, and a sorting and merging algorithm can be set according to different judgment criteria to obtain the passenger travel pattern matrix. Due to the numerical characteristics of similarity calculation, the screening threshold can be freely controlled when mining passenger travel characteristics. The numerical index for judging passenger travel patterns designed in this invention is of great significance for big data analysis of urban rail transit. Attached Figure Description

[0028] Figure 1 A flowchart of the rail transit passenger feature evaluation method based on trajectory similarity according to the present invention;

[0029] Figure 2 This is an example diagram illustrating trajectory similarity calculation in the rail transit passenger feature evaluation method based on trajectory similarity according to the present invention;

[0030] Figure 3 This is a diagram showing the results of passenger travel characteristic evaluation based on an embodiment of the rail transit passenger characteristic evaluation method according to the present invention. Detailed Implementation

[0031] 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.

[0032] In the similarity evaluation algorithm of this application, travel trajectories with highly overlapping stops and those exhibiting inclusion relationships are identified as similar trajectories. The similarity of the stops along the trajectories is used to evaluate the similarity of the trajectories. A passenger may have multiple travel trajectories within a certain period. By calculating the similarity between each pair of trajectories, an overall trajectory similarity matrix can be obtained. This matrix visually displays the quantified similarity relationships among all the trajectories of a single passenger. If the number of passenger travel trajectories within a certain period is N, then there are N similarity relationships between the trajectories. 2 The number of elements N in the initial similarity matrix. 2The similarity evaluation algorithm in this invention sorts and merges the matrix to simplify it, reducing its size while transforming the trajectory similarity matrix into a matrix that reflects passenger travel patterns. Generally, a passenger's travel trajectory may contain multiple trajectories or highly overlapping trajectories between two subway stations. The sorting and merging process of the similarity matrix can extract and merge these trajectories, displaying them intuitively after simplification.

[0033] Reference Figure 1 A method for evaluating the characteristics of rail transit passengers based on trajectory similarity includes the following steps:

[0034] Step 1: Based on the topological connections between adjacent stations in the subway network, establish the subway network as an unweighted graph G = {V, E}.

[0035] Step 2: Retrieve all passenger travel trajectory data over a period of time, and use the shortest path algorithm to find the stations along each trajectory and the number of stations on each trajectory path. The shortest path algorithm is implemented using a breadth-first search algorithm, and the specific implementation method is as follows:

[0036] 1) Starting from the inbound station V0 of the trajectory, first visit this station and mark it as visited;

[0037] 2) Visit all unvisited sites adjacent to site V0 in sequence {V i ,…,V j} and mark it as visited;

[0038] 3) For {V i ,…,V j The access method for station V0 is repeated for each station until an exit station is found or the traversal of the entire metro network is completed.

[0039] Considering passenger volume, the shortest paths between all pairs of subway stations within the entire subway network are pre-calculated and stored. This allows for direct querying when calculating the shortest route between subway stations on two travel trajectories, avoiding redundant calculations for the same station and shortening program execution time. Since round-trip routes between the same pair of stations within the subway network may have different calculated results with the same length, a pre-comparison of round-trip routes is performed when storing the shortest paths between subway stations, and results for stations passing through the same route are stored.

[0040] Step 3: Calculate the pairwise similarity between all the passenger's trajectories based on the similarity of the stations along each trajectory, and obtain the trajectory similarity matrix for the passenger.

[0041] Trajectory similarity is the average of the similarities of stations along the path between two trajectories. The similarity between station A and station B is defined as sim(A,B), the similarity between a station and itself is sim(A,A) = 1, and the similarity between different stations is the reciprocal of the number of stations on the shortest path between the two stations. The trajectory similarity algorithm is as follows:

[0042] 1) Define two trajectory sequences, the longer trajectory sequence being T. L T L Includes L1 to L M There are M stations in total, and the shorter trajectory sequence is T. S T S Includes from S1 to S N There are N stations in total. The trajectory sequence is denoted as: T L =[L1,L2,...,L M ],T S =[S1,S2,...,S N ],(M≥N)

[0043] 2) For shorter trajectories T s Each site S in n Calculate the relationship with another trajectory T L Each station L m similarity sim(L m ,S n And retain the maximum value:

[0044]

[0045] 3) The similarity between the two trajectories is:

[0046]

[0047] exist Figure 2 In the example shown, trajectory 1 contains stations ABCD, and trajectory 2 contains stations BCDE. The similarity is (0.5+1+1+1) / 4 = 0.875.

[0048] According to the algorithm, if there is an inclusion relationship between two trajectories, their similarity calculation result is 1 (i.e., they are completely identical); if passengers travel back and forth on the same route, the similarity between trajectories with opposite entry and exit points is 1.

[0049] Step 4: Further process the similarity matrix to obtain the passenger travel pattern matrix, and effectively merge travel trajectories that satisfy the same pattern. The specific processing involves sorting and merging the trajectory similarity matrix.

[0050] The sorting process is summarized as follows:

[0051] 1) Count the number of 1s in each row. A 1 indicates that the similarity between the two trajectories corresponding to that position is 1. The more 1s there are in a row, the more similar trajectories the travel trajectory represented by that row number has. That is, the travel trajectory represented by this row number can be used as the core of a set of similar trajectories.

[0052] 2) Sort the trajectory numbers from high to low according to the number of 1s in each row. Rearrange the original trajectory similarity matrix according to the sorted number order. The output result is a symmetric matrix after swapping the rows and columns of the original matrix. The number of 1s in the new matrix decreases from the first row to the last row.

[0053]

[0054] As shown in Example T1, since the 6th row of the similarity matrix contains the most 1s, after swapping the rows and columns, the trajectories represented by the elements in the 6th row are sequentially swapped with the trajectories represented by the elements in the 1st row. The number of 1s in each row of the resulting sorted matrix decreases sequentially.

[0055] The merger process can be summarized as follows:

[0056] 1) Starting from the first row of the matrix, record the column numbers of all columns that are 1 in that row as a similar trajectory group. The core trajectory of the group is the trajectory represented by the row number. A trajectory can only be recorded in one similar trajectory group.

[0057] 2) Calculate the similarity matrix after merging using the core trajectories of the groups. The number of rows (columns) in the merged matrix is ​​the number of similar trajectory groups. The similarity between trajectory groups is taken as the similarity value between the core trajectories of the groups.

[0058] 3) The similarity value is amplified by the number of trajectories within a trajectory group. For the merged similarity matrix, the values ​​on the diagonal are the number of trajectories contained in the trajectory group;

[0059] 4) Output a trajectory group similarity matrix where the diagonal represents the number of trajectories contained within the similar trajectory group, and the remaining positions represent the similarity between core trajectories. This matrix represents the travel pattern of the passenger.

[0060]

[0061] As shown in Example T2, the similarity matrix contains three similar trajectory groups. Trajectory 1 is the core trajectory of the similar trajectory group consisting of trajectories 1, 2, and 3; Trajectory 4 is the core trajectory of the similar trajectory group consisting of trajectories 4 and 5; and Trajectory 6 forms a separate similar trajectory group. The relationship between the three similar trajectory groups is intuitively displayed by the travel pattern matrix.

[0062] Step 5: Set the requirement for meeting the regular travel pattern requirement as follows: the number of similar trajectories for each passenger must be no less than 5. Construct two evaluation indicators related to passenger travel characteristics for the travel pattern matrix: the average number of trips per route and the percentage of regular trips. The average number of trips per route is the average number of trajectories contained in each similar trajectory group, i.e., the average number of trips for each regular trajectory. The percentage of regular trips is the ratio of the number of regular trips to the total number of trips.

[0063] The results of evaluating passenger travel characteristics according to this embodiment can be as follows: Figure 3 express.

[0064] In summary, this invention considers the characteristics of passenger travel trajectories in subway networks, identifies travel trajectories with highly overlapping stations as similar trajectories, and designs a trajectory similarity evaluation algorithm. Based on trajectory similarity, a passenger feature evaluation method is proposed. A similarity matrix of passenger travel trajectories over a period of time is calculated through trajectory similarity, and the similarity matrices are sorted and merged to obtain a passenger travel pattern matrix. The passenger feature evaluation index constructed using the travel pattern matrix can intuitively display the regularity characteristics of passenger travel, and the numerical evaluation can provide an intuitive and reliable basis for passenger classification and screening in urban rail transit.

[0065] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention, and applications, modifications and variations thereof will be apparent to those skilled in the art.

[0066] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A method for evaluating the characteristics of rail transit passengers based on trajectory similarity, characterized in that, Includes the following steps: S1. Based on the station connection relationships in the subway network, establish the subway network as an unweighted graph; S2. Collect all travel trajectory data of passengers within a certain period of time, and calculate the stations passed through by each trajectory and the number of stations on each trajectory path; S3. Calculate the pairwise similarity between all the passenger's trajectories based on the similarity of the stations along each trajectory, thus obtaining the passenger's trajectory similarity matrix; the similarity calculation method in step S3 is to define the similarity between station A and station B as... The similarity between a certain website and itself is: The similarity between different stations is the reciprocal of the number of stations on the shortest route between the two stations. The trajectory similarity is the average of the similarity of the stations along the two trajectories. If the two trajectories are the same, their similarity is 1. If the two trajectories have a complete inclusion relationship, their similarity is 1. The farther apart the two trajectories are, the closer the similarity is to 0. S4. Further sort and merge the similarity matrix to obtain the passenger travel pattern matrix; S5. Extract the mathematical features of the travel pattern matrix and evaluate the travel characteristics of passengers.

2. The method for evaluating the characteristics of rail transit passengers based on trajectory similarity as described in claim 1, characterized in that, The algorithm for trajectory similarity includes the following steps: 1) Define two trajectory sequences, the longer trajectory sequence being... , Includes from arrive common For each station, the shorter trajectory sequence is: , Includes from arrive common The trajectory sequence for each station is denoted as: ; 2) For shorter trajectories Each site in Calculate with another trajectory Each site similarity And retain the maximum value: ； 3) The similarity between the two trajectories is .

3. The method for evaluating the characteristics of rail transit passengers based on trajectory similarity as described in claim 2, characterized in that, Step S4 effectively displays the travel patterns of passengers within a given time period using a travel pattern matrix, including the number of pattern categories, the number of trajectories contained in each pattern category, and the similarity between patterns. The sorting process includes the following steps: S41. Count the number of 1s in each row. The number 1 indicates that the similarity between the two trajectories corresponding to that position is 1. The more 1s there are in a row, the more similar trajectories the travel trajectory represented by that row number has. That is, the travel trajectory represented by this row number can be used as the core of a set of similar trajectories. S42. Sort the trajectory numbers from high to low according to the number of 1s in each row. Rearrange the original trajectory similarity matrix according to the sorted number order. The output result is a symmetric matrix after swapping the rows and columns of the original matrix. The number of 1s in the new matrix decreases from the first row to the last row.

4. The method for evaluating the characteristics of rail transit passengers based on trajectory similarity as described in claim 3, characterized in that, The merging process in step S4 includes the following steps: S411. Starting from the first row of the matrix, record the column numbers of all columns with a value of 1 in that row as a similar trajectory group. The core trajectory of the group is the trajectory represented by the row number. A trajectory can only be recorded in one similar trajectory group. S412. Calculate the similarity matrix after merging using the core trajectories of the groups. The number of rows or columns of the merged matrix is ​​the number of similar trajectory groups. The similarity between trajectory groups is taken as the similarity value between the core trajectories of the groups. S413. The similarity value is amplified by the number of trajectories in the trajectory group; for the merged similarity matrix, the value on the diagonal of the matrix is ​​the number of trajectories contained in the trajectory group; S414. Output the similarity matrix of the trajectory group, where the diagonal is the number of trajectories contained in the similar trajectory group and the remaining positions are the similarity between the core trajectories. This matrix represents the travel pattern of the passenger.

5. The method for evaluating the characteristics of rail transit passengers based on trajectory similarity as described in claim 1, characterized in that, In step S5, by setting a set of similar trajectories with no less than 5 lines, the number of regular trips can be obtained by the sum of some elements of the trip pattern matrix, and the proportion of regular trips can be obtained by the ratio of the number of regular trips to the total number of trips.

Images