High-robustness vehicle passing record identification method based on incomplete ETC data

Abstract

The invention relates to the field of traffic, in particular to a high-robustness vehicle passing record identification method based on incomplete ETC data. The objective of the invention is to provide a high-robustness vehicle passing record identification method based on incomplete ETC data so as to deal with the problems of data incompleteness and reliability and realize accurate and efficient identification of vehicle passing records. According to the technical scheme, the high-robustness vehicle passing record recognition method based on the incomplete ETC data comprises the following steps that S1, entrance flow data, exit flow data and portal transaction data are integrated, and travel information data are obtained; s2, integrating the portal transaction data and the portal license identification data to obtain vehicle passing record data; s3, performing travel segmentation, travel merging and travel rejection on the vehicle passing record data to obtain travel track data; and S4, integrating the travel information data and the travel track data to obtain traffic record data.

Description

Technical Field

[0001] The present invention relates to the field of transportation, and in particular to a highly robust vehicle traffic record recognition method based on incomplete ETC data. Background Art

[0002] With the rapid development of intelligent transportation systems, ETC (Electronic Toll Collection) data has become an indispensable data source for highway operations. Identifying and characterizing vehicle traffic records based on ETC data can help highway owners repeatedly mine data value in multiple fields such as toll auditing, vehicle digital archive construction, and travel services. Using ETC data to accurately and efficiently identify vehicle traffic records is of great significance for improving highway management efficiency and optimizing traffic resource allocation.

[0003] Defects and shortcomings of the existing technology:

[0004] 1) The problem of data incompleteness is prominent: Although ETC data has extremely high application value, in actual application, due to various restrictions, few people can obtain complete road network data. In most cases, the ETC data obtained is incomplete, and the incompleteness of this data seriously restricts the accurate identification of vehicle traffic records. Especially for the entrance and exit data of toll stations, although the entry and exit information of vehicles can be recorded, there is a lack of specific driving trajectory data, and there are special cases such as entry without exit, exit without entry, and transit vehicles, which further increases the difficulty of data processing.

[0005] 2) Insufficient reliability of gantry transaction data: As an important part of the ETC system, the reliability of gantry transaction data directly affects the accuracy of vehicle pass records. However, in actual applications, gantry transaction data is not always completely reliable. Vehicles may interfere with the normal detection of the gantry system through various means (such as deliberately shielding the CPC card signal), resulting in missing or incorrect data. In addition, the phenomenon of multiple signatures for one vehicle also occurs from time to time, that is, the same vehicle holds multiple CPC cards, thereby generating multiple transaction records in the system, which brings great trouble to the identification of vehicle pass records.

[0006] 3) There is a misrecognition rate in the license plate recognition data of the gantry bayonet camera: Although the license plate recognition data of the gantry bayonet camera performs well in terms of integrity, there is still a misrecognition rate problem that cannot be ignored. Due to the influence of various factors such as lighting conditions, vehicle speed, license plate damage, etc., the camera may misrecognize the license plate number or license plate color. This misrecognition will not only reduce the accuracy of vehicle traffic records, but also may mislead subsequent data processing and analysis.

[0007] Existing technologies lack effective response strategies: Existing technologies often lack effective response strategies for the above-mentioned data incompleteness and reliability issues. Traditional data processing methods are unable to cope with incomplete and erroneous data, and it is difficult to improve data accuracy while ensuring data integrity. Therefore, there is an urgent need for a vehicle traffic record recognition method that can combine toll station entrance data, toll station exit data, gantry transaction data, and gantry bayonet camera license plate recognition data to maintain high robustness under partial road network data. Summary of the invention

[0008] The purpose of the present invention is to overcome the deficiencies in the above-mentioned background technology and provide a highly robust vehicle traffic record identification method based on incomplete ETC data to address data incompleteness and reliability issues and achieve accurate and efficient identification of vehicle traffic records.

[0009] The technical solution of the present invention is:

[0010] A highly robust vehicle traffic record recognition method based on incomplete ETC data comprises the following steps:

[0011] S1: Integrate the inlet flow data, outlet flow data and gantry transaction data to obtain travel information data;

[0012] S2: Integrate the gantry transaction data and the gantry license plate recognition data to obtain the vehicle passing record data;

[0013] S3: Perform itinerary segmentation, itinerary merging and itinerary elimination on the vehicle passing record data to obtain itinerary trajectory data;

[0014] S4: Integrate travel information data and travel trajectory data to obtain travel record data.

[0015] The entry flow data includes a pass mark, a license plate, an entry toll station mark, and an entry time; the exit flow data includes a pass mark, a license plate, an exit toll station mark, and an exit time;

[0016] The gantry transaction data includes the pass identification, gantry identification, license plate, and vehicle passing time; the travel information data includes the pass identification, license plate, entrance toll station identification, entrance time, exit toll station identification, exit time, and travel type;

[0017] The mast identification data includes mast identification, license plate, and vehicle passing time; the vehicle passing record data includes passage identification, mast identification, license plate, and vehicle passing time;

[0018] The travel trajectory data includes the pass identification, license plate, starting gantry identification, starting gantry passing time, ending gantry identification, ending gantry passing time, and passing trajectory; the pass record data includes the pass identification, license plate, starting gantry identification, starting gantry passing time, ending gantry identification, ending gantry passing time, passing trajectory string, entrance toll station identification, entrance time, exit toll station identification, exit time, and travel type.

[0019] The step S1 includes: matching with the pass identification, fully connecting the inlet flow data, the outlet flow data and the gantry transaction data, and obtaining the travel information data.

[0020] The travel types of the travel information data include both in and out, in but no out, both in and out, and neither in nor out.

[0021] The step S2 includes: matching the license plate and the vehicle passing time, left-connecting the gantry transaction data and the gantry license plate recognition data, sorting by license plate and vehicle passing time, and obtaining vehicle passing record data.

[0022] The trip segmentation in step S3 includes: treating the vehicle passing record data of the same license plate and the vehicle passing through the front and rear adjacent gantries with a time less than the trip segmentation threshold as the same trip, and combining the vehicle passing record data of each trip into a trip trajectory data;

[0023] In the travel trajectory data: the first gantry passed is used as the starting gantry identifier and the starting gantry passing time, and the last gantry passed is used as the ending gantry identifier and the ending gantry passing time; the passing trajectory includes first combining the gantry identifier and the passing time of each passing record data into a trajectory point, and then combining all the trajectory points in the same trip into a passing trajectory in chronological order. The passing trajectory is a json string.

[0024] The trip merging in step S3 includes: merging two trip trajectory data with the same passage identification and adjacent head and tail portals into one trip trajectory data.

[0025] The travel elimination in step S3 is to delete the travel trajectory data that only passes through one gantry and lacks a passage mark.

[0026] The step S4 comprises:

[0027] When the travel type of the travel information data is both in and out or both in and out: if the travel trajectory data has a pass identifier, match it with the pass identifier, connect the travel information data and the travel trajectory data to obtain the pass record data; if the travel trajectory data has no pass identifier, match it with the license plate, the entry time is less than the starting gantry passing time, and the exit time is greater than the ending gantry passing time, connect the travel information data and the travel trajectory data to obtain the pass record data;

[0028] When the travel type of the travel information data is entry without exit, the travel information data is matched with the travel identification, and the travel information data and the travel trajectory data are connected to obtain the travel record data;

[0029] When the travel type of the travel information data is no entry and no exit, the travel trajectory data is used as the travel record data.

[0030] The beneficial effects of the present invention are:

[0031] 1. High robustness: This method can maintain high recognition accuracy and stability even when data is incomplete or erroneous. By integrating multiple ETC data sources (including entrance and exit flow data of toll stations, gantry transaction data, and gantry license plate recognition data), the present invention can effectively overcome data incompleteness and reliability issues and ensure accurate construction of vehicle traffic records.

[0032] 2. Comprehensive coverage of vehicle traffic information: By integrating multiple ETC data sources, the present invention can record every trip information of the vehicle as completely as possible, including key information such as the starting point, the end point, and the driving trajectory; even in the absence of full road network data, the continuity and integrity of the vehicle traffic record can be ensured by reasonably setting the trip termination conditions (such as the vehicle leaving the area under the jurisdiction of the operating unit is regarded as the end of the current trip);

[0033] 3. Intelligent trip division and merging: In view of the differences in trip definitions and data integrity issues, the present invention introduces the "trip segmentation threshold" and "trip merging" mechanisms; by scientifically setting the trip segmentation threshold and finely splitting and merging the passing data of the same vehicle, the vehicle's travel segments can be accurately divided, while avoiding the trip mis-segmentation caused by special circumstances (such as rest in service areas, roadside stops, etc.). This intelligent trip division and merging strategy greatly improves the accuracy and practicality of vehicle traffic records;

[0034] 4. Efficient data cleaning and verification: In order to solve the reliability problem of gantry transaction data and the accuracy challenge of gantry license plate recognition data, the present invention adopts a series of efficient data cleaning and verification measures. By conducting reliability assessment on gantry transaction data and additional verification and correction on gantry license plate recognition data, the high quality of the final vehicle traffic record is ensured.

[0035] 5. Flexible response to actual business scenarios: The present invention fully considers various possibilities in actual business scenarios, such as the redefinition of passID, the processing of various flow data situations, etc.; by reasonably setting data processing rules (such as the introduction of travel types, the integration of gantry license plate recognition data and gantry transaction data, and the association through time and space correlation to synthesize vehicle pass record data), the present invention can flexibly respond to data integration needs in different situations and ensure the accuracy and practicality of vehicle pass records;

[0036] 6. Reduce the cost of manual intervention: Through the automated data integration and verification process, the present invention greatly reduces the cost of manual intervention; monitoring personnel only need to manually review and handle a few complex or abnormal situations, thereby improving work efficiency and reducing error rates;

[0037] In summary, the highly robust vehicle traffic record identification method based on incomplete ETC data provided by the present invention has shown significant advantages and potential in solving the defects and deficiencies of the prior art. Through highly robust data processing technology, comprehensive vehicle traffic information coverage, intelligent itinerary division and merging strategies, efficient data cleaning and verification measures, and the ability to flexibly respond to actual business scenarios, it provides an innovative solution for vehicle traffic record identification in highway operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of the present invention.

[0039] Figure 2 Schematic diagram of the trip segmentation threshold of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] 1. Overall Objectives

[0042] The overall goal of the present invention is to integrate the inlet flow data, the outlet flow data, the gantry transaction data and the gantry plate recognition data, to construct the most complete traffic record data possible, and to record each trip information of the vehicle. In view of the fact that the existing data only covers the information of some sections of the road, it is impossible to accurately track the journey of the vehicle after it leaves the section under the jurisdiction of the operating unit. Therefore, the following strategy is adopted: once the vehicle leaves the area under the jurisdiction of the operating unit, it is regarded as the end of the current journey. Even if the vehicle enters the area under the jurisdiction of the operating unit again at a later time (the pass ID passID remains the same), it is regarded as two independent journeys (after exiting the highway section under the jurisdiction of the current operating unit and entering the highway again, the common situation is that after leaving the highway, driving a section of urban road, and then getting on the highway, or entering the highway section under the jurisdiction of the A operating unit from the highway section under the jurisdiction of the B operating unit and then entering the highway section under the jurisdiction of the A operating unit, both of which are counted as two journeys).

[0043] 2. Technical Difficulties

[0044] (1) Lack of infrastructure geographic information data

[0045] There is currently a lack of detailed geographic information data on key infrastructure such as gantries and toll booths. This lack of information makes it difficult to accurately define the boundary nodes of the operator's area, which in turn makes it complicated and difficult to determine the start and end points of a trip.

[0046] (2) Differences in itinerary definitions

[0047] In the existing toll collection system, the pass ID passID is designed to identify a complete trip from the entrance toll station to the exit toll station. However, this is different from the requirements of the present invention in definition, and the pass ID passID cannot be used to uniquely identify a trip.

[0048] (3) Incomplete flow data

[0049] The existing expressway network is usually managed by multiple operating units, and the data between the operating units are not interconnected. When a vehicle enters and leaves the jurisdiction of an operating unit, there will be no data incompleteness problem. When a vehicle travels across operating units, there will be data incompleteness problems. Currently, only the entrance and exit flow data of toll stations within the jurisdiction of the operating unit can be obtained. In actual data, four situations may be encountered: clear entrance and exit records, only entrance records but no exit records, only exit records but no entrance records, and no entrance or exit records. This brings challenges to the construction of accurate vehicle traffic records.

[0050] (4) Reliability of gantry transaction data

[0051] Although gantry transaction data can be obtained, these data may not be completely reliable, especially in some cases where vehicles may deliberately block CPC card signals, resulting in data loss, which in turn affects the accuracy of constructing vehicle traffic records.

[0052] (5) Challenges in the accuracy of door frame plate recognition data

[0053] Compared with the gantry transaction data, the gantry brand recognition data has better integrity, but there is still a certain misrecognition rate, which requires us to perform additional verification and correction during data processing and analysis.

[0054] In summary, we faced many technical challenges when constructing the vehicle traffic record table, including the lack of infrastructure geographic information data, differences in itinerary definitions, integrity issues with flow data, reliability of gantry transaction data, and accuracy of gantry license plate recognition data. In order to overcome these difficulties, we conducted detailed data cleaning and verification work in combination with actual business scenarios.

[0055] like Figure 1As shown, a highly robust vehicle traffic record recognition method based on incomplete ETC data includes the following steps:

[0056] S1: Integrate the inlet flow data, outlet flow data and gantry transaction data to obtain travel information data;

[0057] Specifically, it includes: matching with the pass identification, fully connecting the entrance flow data, exit flow data and gantry transaction data to obtain travel information data.

[0058] The entry flow data includes the pass identification, license plate, entry toll station identification, and entry time. When a vehicle enters the toll station entrance, the pass identification and the entry toll station information (including the entry toll station identification and entry time) will be written into the pass medium. The entry time is the time when the vehicle enters the toll station entrance.

[0059] The exit flow data includes the pass identification, license plate, exit toll station identification, and exit time. When the vehicle leaves the toll station exit, the pass identification and entrance toll station information will be written into the exit flow data. Therefore, the exit flow data includes the entrance flow data. The exit time is the time when the vehicle leaves the toll station exit.

[0060] The gantry transaction data includes the pass identification, gantry identification, license plate, and vehicle passing time. The gantry transaction data is the data generated by the ETC device when the vehicle passes through the gantry. The gantry reads the pass identification and writes the gantry transaction data. When the vehicle enters the toll station, the license plate data is written into the ETC device.

[0061] The travel information data includes a traffic sign, a license plate, an entrance toll station sign, an entrance time, an exit toll station sign, an exit time, and a travel type.

[0062] The pass identifier is the pass medium ID (OBU serial number code / CPC card code) + entry time (YYYYMMDDHHmmss).

[0063] The travel types of travel information data include both in and out, in but no out, both in and out, and neither in nor out.

[0064] When the travel information data has both inbound and outbound flow data, the travel type is both inbound and outbound;

[0065] When the travel information data has entry flow data but no exit flow data, the travel type is entry without exit, and the exit toll station ID and exit time are null values;

[0066] When the travel information data has no entry flow data but has exit flow data, the travel type is no entry but exit, and the entry toll station ID and entry time are null values;

[0067] When the travel information data has no entry flow data and no exit flow data, the travel type is no entry and no exit, and except for the pass sign and license plate, the rest are empty values.

[0068] S2: Integrate the gantry transaction data and the gantry license plate recognition data to obtain the vehicle passing record data;

[0069] Specifically, it includes: matching the license plate and the passing time, left-connecting the gantry transaction data and the gantry license plate recognition data, sorting by license plate and passing time, and obtaining the passing record data.

[0070] The travel trajectory data records the path information of the trip in detail and is one of the core records of high-speed travel data. In order to construct this data, it is necessary not only to rely on the exit flow data, entrance flow data and gantry transaction data, but also to pay special attention to the integrity and accuracy of the data. Considering that the recognition rate of gantry transaction data is limited, and there may be data loss or the driver deliberately shielding the signal, it is necessary to combine the gantry license plate recognition data to enhance the integrity of the data. The gantry license plate recognition data is the data generated by the gantry bayonet camera when the vehicle passes through the gantry.

[0071] The gantry license plate recognition data includes the gantry logo, license plate, and vehicle passing time.

[0072] The vehicle passing record data includes a pass mark, a gantry mark, a license plate, and a vehicle passing time. Since the gantry license plate recognition data does not include a pass mark, the pass mark of the vehicle passing record data may be a null value.

[0073] S3: Perform itinerary segmentation, itinerary merging and itinerary elimination on the vehicle passing record data to obtain itinerary trajectory data;

[0074] Specifically include:

[0075] (1) Setting the trip segmentation threshold;

[0076] like Figure 2 As shown, through comprehensive statistics of the time intervals of continuous passing through the gantries within the jurisdiction, more than 95% of the vehicles pass through adjacent gantries in less than 10 minutes. Under normal driving conditions, at least 3 gantries can be passed in 20 minutes. Therefore, the trip segmentation threshold is set to 20 minutes, and the trip segmentation is performed on the records of vehicles with the same license plate passing through the gantry.

[0077] (2) Itinerary segmentation

[0078] Based on the determined trip segmentation threshold, the vehicle passing record data of the same vehicle is finely split. This split operation will be grouped according to the license plate and sorted according to the passing time. When it is detected that the time interval between consecutive gantry passes of the same vehicle exceeds the set threshold, it means that a new trip has begun, so the trip is split at this point, and the trips before and after the split point are different trips. The vehicle passing record data of the same trip are combined into a trip trajectory data.

[0079] The specific process of trip segmentation includes: treating the vehicle passing record data with the same license plate and the time of passing the front and rear adjacent gantries less than the trip segmentation threshold (20 minutes) as the same trip, and combining the vehicle passing record data of each trip into one trip trajectory data.

[0080] The travel trajectory data includes a traffic sign, a license plate, a starting gantry sign, a starting gantry vehicle passing time, an ending gantry sign, an ending gantry vehicle passing time, and a vehicle passing track;

[0081] In the trip trajectory data: the starting gantry identifier and the starting gantry passing time are the gantry identifier and passing time of the first gantry passed by the vehicle in the same trip; the ending gantry identifier and the ending gantry passing time are the gantry identifier and passing time of the last gantry passed by the vehicle in the same trip; the passing trajectory includes first combining the gantry identifier and the passing time of each passing record data into a trajectory point, and then combining all the trajectory points in the same trip into a passing trajectory in chronological order, and the passing trajectory is a json string; if the access identifiers of the passing record data are all empty values, the access identifier of the trip trajectory data also takes an empty value; if the access identifiers of non-empty values ​​are the same value, the access identifier of the non-empty value is used as the access identifier of the trip trajectory data; if the access identifiers of non-empty values ​​are different, it means that there is a phenomenon of multiple signatures for one vehicle. At this time, the access identifier that appears the most times in the passing record data is taken as the access identifier of the trip trajectory data.

[0082] Assume that the travel trajectory data passes through 3 gantries in total. The gantry ID and passing time of the first gantry are 0001 and 20240930101000 respectively, the gantry ID and passing time of the second gantry are 0002 and 20240930101500 respectively, and the gantry ID and passing time of the third gantry are 0003 and 20240930102000 respectively. The obtained json string is as follows:

[0083] [{"pass_timestamp":"20240930101000","gantry_id":"0001"},{"pass_timestamp":"20240930 101500","gantry_id":"0002"},{"pass_timestamp":"20240930102000","gantry_id":"0003"}].

[0084] (3) Itinerary Merger

[0085] Taking into account special situations such as vehicles resting in service areas or stopping on the roadside, the time interval between the front and rear adjacent gantries will exceed the preset trip segmentation threshold. In order to ensure the accuracy of trip division, the trip merging step is added.

[0086] The trip merging specifically includes: merging two trip trajectory data with the same traffic identification and adjacent head and tail door frames into one trip trajectory data.

[0087] The number of adjacent first and last gantry frames, including the interval gantry between the end gantry of the previous stroke and the start gantry of the next stroke, is less than 3.

[0088] When merging trips: the starting gantry identifier and the starting gantry passing time of the previous trip trajectory data are used as the starting gantry identifier and the starting gantry passing time of the merged trip trajectory data; the ending gantry identifier and the ending gantry passing time of the next trip trajectory data are used as the ending gantry identifier and the ending gantry passing time of the merged trip trajectory data; the passing trajectories of the two previous and next trip trajectory data are combined into a new json string.

[0089] (4) Stroke elimination

[0090] Since invalid records will be generated by gantry plate recognition errors, invalid data needs to be eliminated to further improve data quality and ensure the accuracy and effectiveness of subsequent analysis.

[0091] The specific steps of trip elimination include: deleting the trip trajectory data that only has a license plate but no pass mark and only passes through one gantry.

[0092] S4: Integrate travel information data and itinerary trajectory data to obtain travel record data;

[0093] (1) When the travel type of the travel information data is both inbound and outbound or both inbound and outbound

[0094] (1.1) If the travel trajectory data has a pass identifier, match it with the pass identifier, connect the travel information data and the travel trajectory data, and obtain the pass record data;

[0095] (1.2) If the travel trajectory data does not have a traffic sign, match the data based on the same license plate, the entry time is less than the starting gantry time, and the exit time is greater than the ending gantry time, connect the travel information data and the travel trajectory data to obtain the traffic record data;

[0096] The entry time is less than the starting gantry time and the exit time is greater than the ending gantry time, indicating that the driving trajectory is between the entrance and exit of the toll station;

[0097] (2) When the travel type of the travel information data is entry without exit, the travel information data is matched with the travel identifier, and the travel information data and the travel trajectory data are connected to obtain the travel record data;

[0098] (3) When the travel type of the travel information data is no entry and no exit, the travel trajectory data is used as the travel record data (the part corresponding to the travel information data is a null value).

[0099] The traffic record data includes the traffic identification, license plate, starting gantry identification, starting gantry passing time, ending gantry identification, ending gantry passing time, passing track string, entrance toll station identification, entrance time, exit toll station identification, exit time, and travel type.

[0100] Example

[0101] From 20241001 to 20241007, the following data were generated on the highways under the jurisdiction of a highway operating company in Zhejiang Province: 5,298,669 entries flow data, 7,673,848 exit flow data, 89,852,585 gantry transaction flow data, 152,235,194 gantry license plate recognition data, 153,248,088 vehicle passing record data, 15,437,067 travel trajectory data, and 16,387,691 traffic record data.

[0102] In the final travel record data table, 52.73% of the vehicles had no entry and no exit, 14.05% had both entry and exit, 12.28% had entry but no exit, and 20.94% had exit but no entry. It is difficult to fully describe the trips with no entry and no exit and with entry but no exit through traditional isolated data tables, but the present invention can obtain complete trip information, effectively supplement the vehicle driving information on the road section under the jurisdiction of the expressway operating company, and restore the travel characteristics of the transit vehicles.

[0103] The preferred embodiments of the present invention are shown in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in this specification. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.

Claims

1. A highly robust vehicle traffic record recognition method based on incomplete ETC data, comprising the following steps: S1: Integrate the inlet flow data, outlet flow data and gantry transaction data to obtain travel information data; S2: Integrate the gantry transaction data and the gantry license plate recognition data to obtain the vehicle passing record data; S3: Perform itinerary segmentation, itinerary merging and itinerary elimination on the vehicle passing record data to obtain itinerary trajectory data; S4: Integrate travel information data and travel trajectory data to obtain travel record data.

2. According to claim 1, a highly robust vehicle traffic record recognition method based on incomplete ETC data is characterized by: The entry flow data includes a pass mark, a license plate, an entry toll station mark, and an entry time; the exit flow data includes a pass mark, a license plate, an exit toll station mark, and an exit time; The gantry transaction data includes the pass identification, gantry identification, license plate, and vehicle passing time; the travel information data includes the pass identification, license plate, entrance toll station identification, entrance time, exit toll station identification, exit time, and travel type; The mast identification data includes mast identification, license plate, and vehicle passing time; the vehicle passing record data includes passage identification, mast identification, license plate, and vehicle passing time; The travel trajectory data includes the pass identification, license plate, starting gantry identification, starting gantry passing time, ending gantry identification, ending gantry passing time, and passing trajectory; the pass record data includes the pass identification, license plate, starting gantry identification, starting gantry passing time, ending gantry identification, ending gantry passing time, passing trajectory string, entrance toll station identification, entrance time, exit toll station identification, exit time, and travel type.

3. According to claim 2, a highly robust vehicle traffic record recognition method based on incomplete ETC data is characterized in that: The step S1 includes: matching with the pass identification, fully connecting the inlet flow data, the outlet flow data and the gantry transaction data, and obtaining the travel information data.

4. The highly robust vehicle traffic record recognition method based on incomplete ETC data according to claim 3 is characterized by: The travel types of the travel information data include both in and out, in but no out, both in and out, and neither in nor out.

5. The highly robust vehicle traffic record recognition method based on incomplete ETC data according to claim 1, characterized in that: The step S2 includes: matching the license plate and the vehicle passing time, left-connecting the gantry transaction data and the gantry license plate recognition data, sorting by license plate and vehicle passing time, and obtaining vehicle passing record data.

6. The highly robust vehicle traffic record recognition method based on incomplete ETC data according to claim 1, characterized in that: The trip segmentation in step S3 includes: treating the vehicle passing record data of the same license plate and the vehicle passing through the front and rear adjacent gantries with a time less than the trip segmentation threshold as the same trip, and combining the vehicle passing record data of each trip into a trip trajectory data; In the travel trajectory data: the first gantry passed is used as the starting gantry identifier and the starting gantry passing time, and the last gantry passed is used as the ending gantry identifier and the ending gantry passing time; the passing trajectory includes first combining the gantry identifier and the passing time of each passing record data into a trajectory point, and then combining all the trajectory points in the same trip into a passing trajectory in chronological order. The passing trajectory is a json string.

7. The highly robust vehicle traffic record recognition method based on incomplete ETC data according to claim 6, characterized in that: The trip merging in step S3 includes: merging two trip trajectory data with the same passage identification and adjacent head and tail portals into one trip trajectory data.

8. The highly robust vehicle traffic record recognition method based on incomplete ETC data according to claim 7, characterized in that: The travel elimination in step S3 is to delete the travel trajectory data that only passes through one gantry and lacks a passage mark.

9. The highly robust vehicle traffic record recognition method based on incomplete ETC data according to claim 1, characterized in that: The step S4 comprises: When the travel type of the travel information data is both in and out or both in and out: if the travel trajectory data has a pass identifier, match it with the pass identifier, connect the travel information data and the travel trajectory data to obtain the pass record data; if the travel trajectory data has no pass identifier, match it with the license plate, the entry time is less than the starting gantry passing time, and the exit time is greater than the ending gantry passing time, connect the travel information data and the travel trajectory data to obtain the pass record data; When the travel type of the travel information data is entry without exit, the travel information data is matched with the travel identification, and the travel information data and the travel trajectory data are connected to obtain the travel record data; When the travel type of the travel information data is no entry and no exit, the travel trajectory data is used as the travel record data.

Images