Vehicle passing track auditing analysis system and method based on spatial-temporal feature clustering

Abstract

The invention relates to the technical field of intelligent traffic, and discloses a vehicle passing track auditing analysis system and method based on spatial-temporal feature clustering, and the system comprises a topology construction module which is configured to construct a highway networking charging attribute graph based on nodes and edge sets, and form a topology network of a node reachable relationship; the feature processing module is configured to perform standardization processing on the spatio-temporal features of the historical passing tracks and eliminate dimensional differences; the clustering analysis module is configured to perform adaptive clustering based on the standardized spatio-temporal characteristics and determine division of a time cluster and a space cluster; and the trajectory reasoning module is configured to generate candidate paths in combination with the space-time cluster characteristics, calculate a comprehensive score through space-time similarity calculation and weight distribution, and output an optimal reasoning path. The method corresponds to the system. According to the invention, the efficiency of the networking charging auditing work is greatly improved, and the national networking charging order is effectively maintained.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent transportation, and specifically to a vehicle passing trajectory auditing and analysis system and method based on spatio-temporal feature clustering. Background Art

[0002] There are significant limitations in the traditional highway network toll auditing and management. The traditional data analysis methods and the way relying on manual verification not only have low work efficiency, but also are prone to missing toll evasion behaviors, and are also easily affected by subjective factors. Although there are currently automated auditing models in application, most of these models are constructed based on explicit rule engines, and the data analysis methods only stay on the comparison of surface features, lacking the ability to deeply and effectively mine implicit information such as the internal laws and correlations of data, so it is difficult to cope with various complex and changeable toll evasion behaviors. In addition, the problems of data limitations and untimely updates further limit the ability to identify toll evasion patterns.

[0003] Therefore, in order to adapt to the new situation of the national network toll operation of highways, it is urgent to develop an auditing technology that can integrate multi-source data, is more intelligent and adapts to business development. Summary of the Invention

[0004] The purpose of the present application is to provide a vehicle passing trajectory auditing and analysis system and method based on spatio-temporal feature clustering, so as to solve the technical problems proposed in the above background art, and achieve a significant improvement in the efficiency of network toll auditing work and effectively maintain the national network toll order.

[0005] To achieve the above purpose, the present application discloses the following technical solutions: In the first aspect, the present application discloses a vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering, including: A topology construction module, configured to: construct a highway network toll attribute graph based on a node and edge set, and form a topological network of node reachability relationships; A feature processing module, configured to: perform standardization processing on the spatio-temporal features of historical passing trajectories to eliminate dimension differences; A clustering analysis module, configured to: perform adaptive clustering based on the standardized spatio-temporal features to determine the time cluster and space cluster partitions; A trajectory inference module, configured to: generate candidate paths by combining spatio-temporal cluster features, and calculate a comprehensive score through spatio-temporal similarity calculation and weight assignment calculation and output the optimal inference path.

[0006] Preferably, the node and edge set includes a node set and a directed edge set; The node set is , where is the toll station node set, is a set of gantry nodes; The set of directed edges includes the edges from toll stations to gantries , edges from gantry to gantry and edges from gantry to toll stations . The reachability relationship between nodes is represented by an adjacent matrix . The matrix elements are marked with 0 or 1 to indicate whether they are connected.

[0007] Preferably, the adjacent matrix includes a toll station to toll station relationship block , a toll station to gantry relationship block , a gantry to toll station relationship block and a gantry to gantry relationship block ; among them, the toll station to toll station relationship block is a matrix with all elements being 0 , the matrix dimension of the toll station to gantry relationship block is , the matrix dimension of the gantry to toll station relationship block is , and the matrix dimension of the gantry to gantry relationship block is .

[0008] Preferably, the feature processing module is further configured to: construct a vehicle passing record matrix based on historical passing trajectories, specifically including: Extract the historical passing trajectory sequence , where each complete passing trajectory flow data is , is the entrance flow, , is the gantry flow, , is the exit flow, ; Convert the trajectory information into a directed edge matrix to complete the conversion of vehicle passing information from cross-sectional information to interval passing state information. The expression of the directed edge matrix is: where , is the passing identification of the nth directed edge of the ith flow, is the directed edge identification, is the entry time, is the exit time, is the passing time, is the passing mileage, is the average passing speed.

[0009] Preferably, the spatio-temporal features include temporal features and spatial features; The temporal features include passing time, the proportion of peak traffic hours, and average speed, and the temporal features are processed by Z-score standardization; The spatial features include mileage, number of lanes, node degree, longitude and latitude, traffic flow density, and adjacent edge correlation, and the spatial features are processed by a hybrid standardization of Min-Max normalization and one-hot encoding.

[0010] Preferably, the temporal feature vector is , where is the average passing time, is the variance of passing time, is the coefficient of variation of passing time, is the proportion of peak traffic hours of passing traffic flow, is the average speed of the whole journey, is the maximum passing time, is the minimum passing time.

[0011] Preferably, the clustering analysis module is specifically configured to: dynamically determine the optimal number of clusters through the elbow method and GapStatistic, calculate the clustering center based on the Euclidean distance and update it iteratively, and output the temporal cluster division and the spatial cluster division .

[0012] Preferably, generating candidate paths by combining spatio-temporal cluster features specifically includes: Missing type definition and preprocessing: Define the missing positions of the flowing water data as the entrance flowing water missing scenario, the exit flowing water missing scenario, and the gantry flowing water missing scenario. Among them, the expression corresponding to the entrance flowing water missing scenario is , the expression corresponding to the exit flowing water missing scenario is , and the expression corresponding to the gantry flowing water missing scenario is ; Generating a candidate path set: If the entrance flowing water is missing, extract the first known gantry , and retrieve all possible entrance candidate sets in reverse to form a candidate path set , where , ; If the exit flowing water is missing, extract the last known gantry , and retrieve all possible exit nodes in the forward direction to form a candidate path set , where , ; If the gantry sequence is missing, extract the starting and ending endpoints of the missing , from forward retrieval to all possible passing nodes in the interval, where the set of directed edges during this period is , ; output the candidate path set .

[0013] Preferably, the calculation of the comprehensive score and the output of the optimal inference path specifically include: Feature alignment: For each candidate path , extract the time feature and the space feature , and align with the historical trajectory feature result, where , ; Trajectory spatio-temporal similarity calculation: Calculate the distance between the time feature of each edge in the path and the center of its time cluster, and calculate the time similarity , and the expression of the time similarity is: where is the time feature vector of the candidate path edge , is the time cluster center feature vector of the historical trajectory edge , is the historical time feature standard deviation; Calculate the distance between the space feature of each edge in the path and the center of its space cluster, and calculate the space similarity , and the expression of the space similarity is: where is the space feature vector of the candidate path edge , is the space cluster center feature vector of the historical trajectory edge , is the space feature standard deviation; Comprehensive score calculation: Combine the time and space similarities to calculate the score and calculate the comprehensive score , and the calculation formula is: where is the weighted high-frequency edge weight of the target vehicle appearing in the directed edge in the historical trajectory, is the time similarity weight coefficient, is the space similarity weight coefficient; ​​​​​​Output the optimal inference path: Output the Top-K paths in descending order according to the comprehensive score.

[0014] In a second aspect, the present application discloses a vehicle passing trajectory auditing and analysis method based on spatio-temporal feature clustering. Applying the vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering as described above, the method includes the following steps: Construct a highway network toll attribute graph based on a set of nodes and edges to form a topological network of node reachability relationships; Standardize the time features and space features of historical passing trajectories; Perform adaptive clustering based on the standardized spatio-temporal features to obtain time clusters and space clusters; Generate candidate paths by combining spatio-temporal cluster features, calculate the comprehensive score through spatio-temporal similarity calculation and weight assignment calculation, and output the optimal inference path.

[0015] Beneficial effects: The vehicle passing trajectory auditing and analysis system and method based on spatio-temporal feature clustering of the present application lay a foundation for the rationality of vehicle passing reachable paths by constructing a highway network toll attribute graph; through spatio-temporal data feature extraction and clustering analysis of historical data, it effectively increases the sensitivity of vehicle passing features and significantly reduces the interference of data analysis noise; by using time local feature cluster division and space local feature cluster division, when screening candidate path edges, it greatly reduces the data retrieval complexity and improves the accuracy of candidate paths; by adopting a weight calculation that balances historical behavior matching and path rationality, combining a scoring function and dynamic constraints, and complementing paths through multi-index quantification, it realizes the effective promotion and restoration of missing paths in feature streams. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0017] Figure 1 It is a structural block diagram of the vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering provided by the embodiment of the present application. Detailed Embodiments

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0019] In this text, the term "comprising" is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a series of elements includes not only those elements but also other elements not expressly listed, or elements that are inherent to such process, method, article, or apparatus. Without further limitation, an element limited by the statement "comprising..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0020] In a first aspect, this embodiment discloses a vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering as shown in Figure 1 which includes a topology construction module, a feature processing module, a clustering analysis module, and a trajectory inference module.

[0021] Specifically The topology construction module is configured to: construct a highway network toll attribute graph based on a set of nodes and edges, and form a topological network of node reachability relationships.

[0022] The feature processing module is configured to: perform standardization processing on the spatio-temporal features of historical passing trajectories to eliminate dimension differences.

[0023] The clustering analysis module is configured to: perform adaptive clustering based on the standardized spatio-temporal features to determine the division of time clusters and space clusters.

[0024] The trajectory inference module is configured to: generate candidate paths by combining spatio-temporal cluster features, calculate a comprehensive score through spatio-temporal similarity calculation and weight assignment calculation, and output the optimal inference path.

[0025] In one implementation, the set of nodes and edges includes a set of nodes and a set of directed edges.

[0026] The set of nodes is , where is the toll station node set, is the gantry node set, and the node attribute definitions include but are not limited to: node number, name, affiliated area, usage status, toll station type, main line stake number, upstream flag, upstream flag, provincial boundary flag, latitude and longitude, etc. The toll station node set is , and the gantry node set: .

[0027] The set of directed edges includes the edges from toll stations to gantries , the edges from gantry to gantry , and the edges from gantries to toll stations . The reachability relationship between nodes is represented by an adjacency matrix , and the matrix elements are marked with 0 or 1 to indicate whether they are connected. , , , the set of directed edges is .

[0028] The adjacent matrix mentioned above includes the toll station-to-toll station relationship block , the toll station-to-gantry relationship block , the gantry-to-toll station relationship block and the gantry-to-gantry relationship block ; among them, the toll station-to-toll station relationship block is a matrix with all elements being 0, and the matrix dimension of the toll station-to-gantry relationship block is , the matrix dimension of the gantry-to-toll station relationship block is , and the matrix dimension of the gantry-to-gantry relationship block is .

[0029] The relevant introduction of the adjacent matrix is specifically as follows: : Matrix, the edges from toll station to toll station (no connection, all 0).

[0030] : Matrix, representing the connection relationship from toll station to gantry.

[0031] : Matrix, representing the connection relationship from gantry to toll station.

[0032] : Matrix, representing the edges from gantry to gantry.

[0033] In one implementation, the feature processing module is further configured to: construct a vehicle passing record matrix based on the historical passing trajectories, specifically including: Extract the historical passing trajectory sequence , where each complete passing trajectory flow data is , is the entrance flow, , is the gantry flow, , is the export flow,[[]] ; Convert the trajectory information into a directed edge matrix to complete the conversion of vehicle passing information from cross-sectional information to interval passing state information. The expression of the directed edge matrix is: where, , is the passing identification of the nth directed edge of the ith flow, is the directed edge identification, is the entry time, is the exit time, is the passing time, is the passing mileage, is the average passing speed.

[0034] In one implementation, the spatio-temporal features include time features and space features; The time features include passing time, the proportion of peak traffic hours, and average speed, and the time features are processed by Z-score standardization; The space features include mileage, number of lanes, node degree, longitude and latitude, traffic density, and adjacent edge correlation, and the space features are processed by a hybrid standardization of Min-Max normalization and one-hot encoding.

[0035] Further, the clustering analysis module is specifically configured to: dynamically determine the optimal number of clusters through the elbow method and GapStatistic, calculate the clustering center based on the Euclidean distance and iteratively update it, and output the time cluster division and the space cluster division .

[0036] Specifically, the time local feature clustering analysis includes: 1. Time feature standardization and enhanced extraction: Based on the vehicle passing flow directed edge matrix , extract each directed edge , extract the local features of time, perform Z-score standardization to eliminate the dimension difference, and form a feature vector . The time feature vector is , is the maximum passing time, is the minimum passing time.

[0037] where is the average passing time, which is used to reflect the overall passing efficiency and passing congestion of the directed edge section.

[0038] n: The number of passing vehicle trips of a directed edge e during the observation period. : The actual time taken for the i-th vehicle trip to pass through this directed edge.

[0039] is the variance of the passing time, which is used to measure the volatility of the passing time. The larger the value, the more unstable the passing time.

[0040] is the coefficient of variation of the passing time. By taking the ratio of the standard deviation to the mean, the dimension is eliminated, which is convenient for comparison across directed edges.

[0041] is the proportion of the peak period of the passing traffic flow.

[0042] is the distribution set of the peak period of the passing of the directed edge e. indicates that the passing time of the i-th vehicle belongs to the peak time. is used to indicate the function to count the total number of vehicles passing during the peak time. is the total traffic flow.

[0043] is the average speed of the whole journey.

[0044] is the passing mileage of the directed edge e. is the average passing time of the directed edge e.

[0045] 2. Determination and clustering of adaptive time clusters: (1) Input parameters: Set of directed edges E, feature matrix , number of classification clusters K, convergence threshold ϵ.

[0046] (2) Algorithm steps: Initialization: Randomly select K initial clustering centers ; Assign clusters: For each edge , calculate its Euclidean distance to each center; Among them, is the standardized eigenvalue of the directed edge e. is the center value of the K-th cluster. is the cluster with the smallest Euclidean distance of the directed edge e.

[0047] Update Center: Recalculate cluster means: where, is the number of edge sets of cluster k in the t-th iteration, is the center point calculated for cluster K in the (t + 1)-th iteration.

[0048] Iteration: Repeat the steps of assigning clusters and updating the center until: where, is the set threshold, is the center point calculated for cluster K in this iteration, is the center point calculated for cluster K in the previous iteration.

[0049] (3)Output result: Temporal cluster partitioning Spatial local feature clustering analysis, including: 1. Spatial feature standardization and extraction: Based on the attribute graph, extract each directed edge and extract the local features of the space, and adopt a hybrid standardization strategy to form the feature vector ; where, is the mileage of the directed edge after Min-Max normalization; is the number of lanes after Min-Max normalization; is the type of the directed edge (0 for toll station to gantry, 1 for gantry to gantry, 2 for gantry to exit toll station), which is one-hot encoded; and are the out-degree and in-degree (the number of other connected edges) of the start and end nodes after Min-Max normalization respectively; and are the longitude and latitude of the midpoint of the standardized edge; is the standardized average daily traffic flow density; is the correlation of adjacent edges after standardization (obtained based on the historical passing conversion probability).

[0050] 2. Adaptive determination of the number of spatial clusters and clustering (1)Input parameters: Set of directed edges E, feature matrix , number of classification clusters Y, convergence threshold ϵ; (2)Process using the algorithm steps in the temporal local feature clustering analysis; (3)Output result: Spatial cluster partitioning 。

[0051] In one implementation, generating candidate paths by combining spatio-temporal cluster features specifically includes: Missing type definition and preprocessing: Define the missing positions of the flowing water data as the inlet flowing water missing scenario, the outlet flowing water missing scenario, and the gantry flowing water missing scenario. Among them, the expression corresponding to the inlet flowing water missing scenario is The expression corresponding to the outlet flowing water missing scenario is The expression corresponding to the gantry flowing water missing scenario is ; Generating a candidate path set: Based on the topological structure and spatio-temporal constraints (time and distance thresholds) of the attribute graph M = (U, A, E), generate a candidate path set. If the inlet flowing water is missing, extract the first known gantry and retrieve all possible inlet candidate sets in reverse to form a candidate path set where ; If the outlet flowing water is missing, extract the last known gantry and retrieve all possible outlet nodes in the forward direction to form a candidate path set where ; If the gantry sequence is missing, extract the starting endpoint of the missing part and retrieve all possible passing nodes in the interval from in the forward direction to where the set of directed edges during this period is ; Output the candidate path set .

[0052] Furthermore, calculating the comprehensive score and outputting the optimal inference path specifically includes: Feature alignment: For each candidate path , extract the time feature and the spatial feature , and align them with the historical trajectory feature results, where ; ; Trajectory spatio-temporal similarity calculation: Calculate the distance between the time feature of each edge in the path and the center of its time cluster to obtain the time similarity The expression for the time similarity is: where is the time feature vector of the candidate path edge , is the time cluster of the historical trajectory edge ​ Central feature vector is the standard deviation of historical time features; Calculate the distance between the spatial features of each edge in the path and the center of the spatial cluster to which it belongs, and calculate the spatial similarity , and the expression of spatial similarity is: where is the spatial feature vector of the candidate path edge , is the historical trajectory edge 's spatial cluster central feature vector, is the standard deviation of spatial features; Comprehensive score calculation: Combine time and spatial similarity to calculate the score and calculate the comprehensive score , and the calculation formula is: where is the weighted high-frequency edge weight of the target vehicle on the directed edge appearing in the historical trajectory, is the time similarity weight coefficient, is the spatial similarity weight coefficient; Output the optimal inference path: Output the Top-K paths in descending order according to the comprehensive score. Top-K is the top K pushed paths, as the scope of key trajectory auditing for target vehicles with missing traffic information Based on the above, the vehicle passing trajectory auditing analysis system based on spatio-temporal feature clustering in this embodiment lays a foundation for the rationality of the vehicle passing reachable path by constructing a highway network toll attribute graph; through spatio-temporal data feature extraction and clustering analysis of historical data, it effectively increases the sensitivity of vehicle passing features and significantly reduces the interference of data analysis noise; by using time local feature cluster division and spatial local feature cluster division, when screening candidate path edges, it greatly reduces the data retrieval complexity and improves the accuracy of candidate paths; by adopting weight calculation that balances historical behavior matching and path rationality, combining a scoring function and dynamic constraints, and complementing paths through multi-index quantification, it realizes the effective advancement and restoration of paths with missing feature streams.

[0053] In the second aspect of this embodiment, a vehicle passing trajectory auditing analysis method based on spatio-temporal feature clustering is provided, which is applied to the vehicle passing trajectory auditing analysis system based on spatio-temporal feature clustering as described above. The method includes the following steps: Construct a highway network toll attribute graph based on node and edge sets to form a topological network of node reachable relationships; Standardize the time characteristics and spatial characteristics of the historical traffic trajectories; Perform adaptive clustering based on the standardized spatio-temporal characteristics to obtain time clusters and spatial clusters; Generate candidate paths by combining spatio-temporal cluster characteristics, calculate the comprehensive score through spatio-temporal similarity calculation and weight assignment calculation, and output the optimal inference path.

[0054] It should be noted that the vehicle traffic trajectory audit analysis method based on spatio-temporal feature clustering in this embodiment corresponds to the vehicle traffic trajectory audit analysis system based on spatio-temporal feature clustering described above. Therefore, for the parts not described in detail in the vehicle traffic trajectory audit analysis method based on spatio-temporal feature clustering in this embodiment (including but not limited to specific technical means, technical effects, etc.), reference can be made to the relevant descriptions in the vehicle traffic trajectory audit analysis system based on spatio-temporal feature clustering described above, and this text will not elaborate here.

[0055] In the embodiments provided in this application, it should be understood that the embodiments described here can be implemented in hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described here, or a combination thereof. For software implementation, part or all of the processes of the embodiments can be completed by a computer program instructing the relevant hardware. When implemented, the above program can be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. The computer-readable storage medium includes computer storage media and communication media, where the communication media includes any medium that facilitates the transmission of a computer program from one place to another. The storage media can be any available medium that can be accessed by a computer. The computer-readable storage medium can include but is not limited to RAM, ROM, EEPROM, CD-ROM or other optical disc storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0056] Finally, it should be noted that the above are only the preferred embodiments of the present application and are not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A vehicle passing trajectory audit and analysis system based on spatio-temporal feature clustering, characterized in that including: A topology construction module, configured to: construct a highway network toll attribute graph based on a set of nodes and edges, and form a topological network of node reachability relationships; A feature processing module, configured to: perform standardization processing on the spatio-temporal features of historical passing trajectories to eliminate dimensional differences; A clustering analysis module, configured to: perform adaptive clustering based on the standardized spatio-temporal features to determine the division of time clusters and space clusters; A trajectory inference module, configured to: generate candidate paths by combining spatio-temporal cluster features, calculate a comprehensive score through spatio-temporal similarity calculation and weight assignment calculation, and output the optimal inference path.

2. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 1, characterized in that The set of nodes and edges includes a set of nodes and a set of directed edges; The node set is , where is the toll station node set, is the gantry node set; The directed edge set includes the edges from toll stations to gantries , the edges from gantry to gantry and the edges from gantry to toll station . The reachability relationship between nodes is represented by an adjacent matrix . The matrix elements are marked with 0 or 1 to indicate whether they are connected 3. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 2, wherein, The adjacent matrix described above includes the toll station-to-toll station relationship block , the toll station-to-gantry relationship block , the gantry-to-toll station relationship block and the gantry-to-gantry relationship block ; among them, the toll station-to-toll station relationship block is a matrix with all elements being 0 , the matrix dimension of the toll station-to-gantry relationship block is , the matrix dimension of the gantry-to-toll station relationship block is , and the matrix dimension of the gantry-to-gantry relationship block is .

4. The vehicle passing trajectory audit and analysis system based on spatio-temporal feature clustering according to claim 1, characterized in that, The feature processing module is further configured to: construct a vehicle passing record matrix based on historical passing trajectories, specifically including: Extract the historical passing trajectory sequence where each complete passing trajectory flow data is , the entrance flow,[[]] , the gantry flow,[[]] , the exit flow,[[]] ; Convert the trajectory information into a directed edge matrix, complete the conversion of vehicle passing information from cross-sectional information to interval passing state information, and the expression of the directed edge matrix is: Among them, , is the passing identification of the nth directed edge of the ith water flow, is the directed edge identification, is the entry time, is the exit time, is the passing time, is the passing mileage, is the average passing speed.

5. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 1 or 4, characterized in that, The spatio-temporal features include time features and space features; The time features include passing time, the proportion of peak traffic hours, and average speed, and the time features are processed using Z-score standardization; The space features include mileage, number of lanes, node degree, longitude and latitude, traffic density, and adjacent edge correlation, and the space features are processed using a hybrid standardization of Min-Max normalization and one-hot encoding.

6. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 5, wherein The time feature vector is , where is the average passing time, is the variance of the passing time, is the coefficient of variation of the passing time, is the proportion of the peak period of the passing vehicle flow, is the average speed of the whole journey, is the maximum passing time, is the minimum passing time.

7. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 1, characterized in that The clustering analysis module is specifically configured to: dynamically determine the optimal number of clusters through the elbow method and GapStatistic, calculate the clustering center based on the Euclidean distance and update it iteratively, and output the time cluster division and the spatial cluster division .

8. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 1, characterized in that The generating candidate paths by combining spatio-temporal cluster features specifically includes: Missing type definition and preprocessing: Define the missing positions of the flowing water data as the inlet flowing water missing scenario, the outlet flowing water missing scenario, and the gantry flowing water missing scenario. Among them, the expression corresponding to the inlet flowing water missing scenario is , the expression corresponding to the outlet flowing water missing scenario is , and the expression corresponding to the gantry flowing water missing scenario is ; Generate a candidate path set: If the incoming flow is missing, extract the first known gantry , and retrieve all possible incoming candidate sets in reverse to form a candidate path set , where , ; if the outgoing flow is missing, extract the last known gantry , and retrieve all possible outgoing nodes in forward to form a candidate path set , where , ; if the gantry sequence is missing, extract the missing starting endpoint , from retrieve all possible passing nodes in the interval from in forward, where the set of directed edges during this period is , ; output the candidate path set .

9. The vehicle passing trajectory auditing and analysis system based on spatio-temporal feature clustering according to claim 8, characterized in that, The calculating the comprehensive score and outputting the optimal inference path specifically includes: Feature alignment: For each candidate path , extract temporal features and spatial features , and align them with the historical trajectory feature results, where , ; Trajectory spatio-temporal similarity calculation: Calculate the distance between the time features of each edge in the path and the center of the time cluster to which it belongs, and obtain the time similarity , and the expression for time similarity is: Among them, is the time feature vector of the candidate path edge , is the time cluster of the historical trajectory edge center feature vector is the historical time feature standard deviation; Calculate the distance between the spatial features of each edge in the calculation path and the center of its affiliated spatial cluster, and the spatial similarity is obtained , and the expression of the spatial similarity is as follows: Among them, is the spatial feature vector of the candidate path edge , is the spatial cluster of the historical trajectory edge center feature vector is the spatial feature standard deviation; Comprehensive score calculation: Calculate the comprehensive score by combining time and spatial similarity for score calculation , and the calculation formula is as follows: Among them, is the weighted high-frequency edge weight that the target vehicle appears in the directed edge in the historical trajectory, is the time similarity weight coefficient, is the space similarity weight coefficient; Output the optimal inference path: Output the Top-K paths in descending order of the comprehensive score.

10. A vehicle passing trajectory audit and analysis method based on spatio-temporal feature clustering, which applies the vehicle passing trajectory audit and analysis system based on spatio-temporal feature clustering as described in any one of claims 1-9, characterized in that, The method includes the following steps: Construct a highway network toll attribute graph based on a set of nodes and edges, and form a topological network of node reachability relationships; Perform standardization processing on the time features and space features of historical passing trajectories; Perform adaptive clustering based on the standardized spatio-temporal features to obtain time clusters and space clusters; Generate candidate paths by combining spatio-temporal cluster features, calculate a comprehensive score through spatio-temporal similarity calculation and weight assignment calculation, and output the optimal inference path.

Images