A method and system for reliably extracting vehicle trajectories based on ETC gantry data

Abstract

The present invention relates to a method and system for reliably extracting vehicle trajectories based on ETC gantry data, belonging to the technical field of the development and utilization of ETC gantry data on expressways. First, a database is constructed, and then the similarity distances between the target license plate number of the target vehicle for which the trajectory is to be extracted and other license plate numbers in the database are calculated respectively to determine the similar license plate numbers of the target license plate number. Further, a first data set and a second data set are constructed. Then, selective fusion is performed on the first data set and the second data sets corresponding to each similar license plate number to obtain a fused data set. Finally, the fused data set is split into single-trip data to obtain several single-trip trajectories of the target vehicle during the analysis period, thereby reliably extracting the vehicle trajectories, accurately identifying the vehicle operation trajectory paths, and providing a data processing technical basis for improving the accuracy of toll auditing and realizing the development and utilization of ETC data.

Description

Technical Field

[0001] The present invention relates to the technical field of the development and utilization of ETC gantry data on expressways, and particularly to a method and system for reliably extracting vehicle trajectories based on ETC gantry data. Background Art

[0002] With the continuous application of new toll collection technologies, in addition to the traditional toll station clearing data, a large amount of ETC gantry data has been accumulated during the operation of expressways for realizing the segmented tolling of expressway vehicles. The ETC gantry obtains information such as license plate information, vehicle information, location, and tolling situation of passing vehicles through the integration of license plate recognition and in-vehicle terminal perception. However, due to factors such as insufficient gantry supplementary lighting, environmental reflection, insufficient license plate recognition accuracy, OBU device shielding, and limited gantry perception ability, situations such as incorrect gantry license plate recognition, missing gantry data, repeated collection of gantry data, and abnormal gantry data often occur, which has a greater negative impact on accurately identifying vehicle trajectories, supporting accurate tolling, and promoting the development and utilization of ETC data.

[0003] In view of the above problems, there is an urgent need for a technology for reliably extracting vehicle trajectories based on ETC gantry data. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for reliably extracting vehicle trajectories based on ETC gantry data, which can reliably extract vehicle trajectories, accurately identify the path of vehicle operation trajectories, and provide a data processing technology basis for improving the accuracy of toll auditing and realizing the development and utilization of ETC data.

[0005] To achieve the above purpose, the present invention provides the following solutions:

[0006] A method for reliably extracting vehicle trajectories based on ETC gantry data, the vehicle trajectory reliable extraction method comprising:

[0007] Constructing a database; the database includes a number of vehicle passing data collected by each ETC gantry arranged on the expressway during the analysis period; the vehicle passing data includes mileage, license plate number, collection time, and traveling direction; the license plate number includes a provincial abbreviation, a city code plus a serial number, and a license plate color; the traveling direction includes exit, entry, up, and down;

[0008] Calculating the similarity distance between the target license plate number of the target vehicle of the trajectory to be extracted and each license plate number other than the target license plate number in the database respectively, and selecting the license plate numbers with the similarity distance less than the similarity threshold as the similar license plate numbers of the target license plate number;

[0009] Select the vehicle passing data with the license plate number being the target license plate number from the database to form a first data set; for each of the similar license plate numbers, select the vehicle passing data with the license plate number being the similar license plate number from the database to form a second data set; perform selective fusion on the first data set and the second data set corresponding to each similar license plate number to obtain a fused data set;

[0010] Perform single-trip data splitting on the fused data set to obtain several single-trip trajectories of the target vehicle during the analysis period.

[0011] A vehicle trajectory credible extraction system based on ETC gantry data, the vehicle trajectory credible extraction system includes:

[0012] A construction module for constructing a database; the database includes several vehicle passing data collected by each ETC gantry arranged on the highway during the analysis period; the vehicle passing data includes mileage, license plate number, collection time, and traveling direction; the license plate number includes a provincial abbreviation, a city code plus a serial number, and a license plate color; the traveling direction includes exit, entry, up, and down;

[0013] A similar license plate number determination module for calculating the similarity distance between the target license plate number of the target vehicle whose trajectory is to be extracted and each license plate number in the database other than the target license plate number, and selecting the license plate numbers with the similarity distance less than the similarity threshold as the similar license plate numbers of the target license plate number;

[0014] A data fusion module for selecting the vehicle passing data with the license plate number being the target license plate number from the database to form a first data set; for each of the similar license plate numbers, selecting the vehicle passing data with the license plate number being the similar license plate number from the database to form a second data set; performing selective fusion on the first data set and the second data set corresponding to each similar license plate number to obtain a fused data set;

[0015] A splitting module for performing single-trip data splitting on the fused data set to obtain several single-trip trajectories of the target vehicle during the analysis period.

[0016] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0017] The present invention is used to provide a method and system for reliably extracting vehicle trajectories based on ETC gantry data. First, a database is constructed, which includes a number of vehicle passing data collected by each ETC gantry arranged on the highway during the analysis period. Then, the similarity distances between the target license plate number of the target vehicle whose trajectory is to be extracted and other license plate numbers in the database are calculated respectively to determine the similar license plate numbers of the target license plate number. Further, a first data set and a second data set are constructed. Then, selective fusion is performed on the first data set and the second data sets corresponding to each similar license plate number to obtain a fused data set. Finally, the fused data set is split into single-trip data to obtain a number of single-trip trajectories of the target vehicle during the analysis period, so as to reliably extract the vehicle trajectories, accurately identify the vehicle operation trajectory path, and provide a data processing technology basis for improving the toll auditing accuracy and realizing the development and utilization of ETC data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 It is the flowchart of the method for reliably extracting vehicle trajectories provided in Embodiment 1 of the present invention;

[0020] Figure 2 It is the technical roadmap of the method for reliably extracting vehicle trajectories provided in Embodiment 1 of the present invention;

[0021] Figure 3 It is the schematic diagram of the initial data provided in Embodiment 1 of the present invention;

[0022] Figure 4 It is the schematic diagram of the initial data of the similar license plate numbers provided in Embodiment 1 of the present invention;

[0023] Figure 5 It is the schematic diagram of the single-trip trajectory provided in Embodiment 1 of the present invention;

[0024] Figure 6 It is the system block diagram of the vehicle trajectory reliable extraction system provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0026] The purpose of the present invention is to provide a method and system for reliably extracting vehicle trajectories based on ETC gantry data, which can reliably extract vehicle trajectories, accurately identify the paths of vehicle operation trajectories, and provide a data processing technology basis for improving the accuracy of toll auditing and realizing the development and utilization of ETC data.

[0027] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] Embodiment 1:

[0029] This embodiment is used to provide a method for reliably extracting vehicle trajectories based on ETC gantry data. As Figure 1 and Figure 2 shown, the method for reliably extracting vehicle trajectories includes:

[0030] S1: Construct a database; the database includes a number of vehicle passing data collected by each ETC gantry arranged on the highway during the analysis period; the vehicle passing data includes mileage, license plate number, collection time, and travel direction; the license plate number includes a provincial abbreviation, a city code plus a serial number, and a license plate color; the travel direction includes out, in, up, and down;

[0031] Specifically, S1 may include:

[0032] (1) Obtain a number of initial data collected by each ETC gantry arranged on the highway during the analysis period; the initial data includes the ETC gantry stake number, the initial license plate number, the collection time, and the traffic flow direction sensed by the ETC gantry;

[0033] The ETC gantries in this embodiment can be ETC gantries at the entrance and exit toll stations, section ETC gantries, and branch ramp ETC gantries arranged on the highway. In this embodiment, the construction information of each ETC gantry will be obtained first, including the stake number where the ETC gantry is located and the traffic flow direction sensed by the ETC gantry (the traffic flow direction sensed by the ETC gantry at the entrance and exit toll stations is out or in, and the traffic flow direction sensed by the ETC gantries at other positions is up or down), etc. Each ETC gantry will also collect vehicle information passing through the ETC gantry. The construction information and vehicle information together form an initial data collected by the ETC gantry. As Figure 3As shown in the figure, the content of the initial data includes but is not limited to: ETC gantry toll point name, ETC gantry toll point number, ETC gantry mileage, initial license plate number, vehicle type, vehicle category, special situation value, collection time, traffic flow direction (in / out / up / down), vehicle load, etc.

[0034] (2) For each piece of initial data, convert the ETC gantry mileage to kilometers; perform data standardization on the initial license plate number to obtain the license plate number; determine the driving direction based on the traffic flow direction sensed by the ETC gantry to obtain vehicle passing data; all vehicle passing data forms a database.

[0035] Specifically, the initial data undergoes data preprocessing to obtain vehicle passing data. This data preprocessing process includes mileage position calculation of the initial data, marking of the driving direction of the collected data, standardization of license plate data, elimination of duplicate data, and database creation. Specifically as follows:

[0036] (2.1) Mileage conversion: Standardize the ETC gantry mileage data, converting the data format of "KY+XXX" to the mileage data format of Y.XXX to convert the ETC gantry mileage to kilometers for subsequent data processing. The unit of Y is km, and the unit of XXX is m.

[0037] (2.2) Standardization of license plate data. Perform data standardization processing for different storage structures of license plate data and abnormal license plate information in the initial data. For license plates with complete information such as the abbreviation of the province where the license plate is located, the letter of the first-level code of the city, the serial number, and the license plate color in the license plate recognition result, standardize them according to the data structure of the province abbreviation, city code plus serial number, and license plate color. For license plates with missing structured data, supplement them by referring to videos and license plate images.

[0038] (2.3) Marking of the driving direction. Mark the driving direction (in, out, up, down) for each piece of initial data according to the traffic flow direction sensed by the ETC gantry.

[0039] (2.4) Creation of a database for the collected data. After standardizing all the initial data during the analysis period, obtain vehicle passing data to construct a database. The structure of the vehicle passing data includes but is not limited to data number, ETC gantry toll point name, ETC gantry toll point number, ETC gantry mileage, kilometers, license plate number (province abbreviation, city code plus serial number, license plate color), vehicle type, vehicle category, special situation value, collection time, driving direction, vehicle load, etc.

[0040] Preferably, the data preprocessing of this embodiment further includes: elimination of duplicate data. For multiple pieces of vehicle passing data with exactly the same information, only keep 1 piece and eliminate the redundant data.

[0041] S2: Calculate the similarity distances between the target license plate number of the target vehicle on the trajectory to be extracted and each license plate number in the database except the target license plate number, and select the license plate numbers with similarity distances less than the similarity threshold as the similar license plate numbers of the target license plate number.

[0042] In S2, first extract the license plate number information of all vehicle passing data in the database, including the provincial abbreviation, city code plus sequence number, and license plate color, to determine each license plate number in the database except the target license plate number. On this basis, calculating the similarity distances between the target license plate number of the target vehicle on the trajectory to be extracted and each license plate number in the database except the target license plate number may include: for each license plate number in the database except the target license plate number, use the edit distance calculation formula to calculate the edit distance between the provincial abbreviation of the target license plate number and the provincial abbreviation of the license plate number to obtain the first distance; use the edit distance calculation formula to calculate the edit distance between the city code plus sequence number of the target license plate number and the city code plus sequence number of the license plate number to obtain the second distance; calculate the third distance between the license plate color of the target license plate number and the license plate color of the license plate number; perform weighted summation on the first distance, the second distance, and the third distance to obtain the similarity distance.

[0043] Among them, to calculate the edit distance of each element of the license plate number (the elements include the provincial abbreviation and the city code plus sequence number): by constructing a calculation loop, calculate the edit distance between the target license plate and other license plate numbers in the database. The used edit distance calculation formula is:

[0044]

[0045] Among them, dis a,b (i,j) is the edit distance between the i-th character of string a and the j-th character of string b; using the above formula, the total edit distance, that is, the first distance and the second distance, can be calculated through step-by-step iteration.

[0046] Calculate the edit distance of the license plate color: determine whether the license plate color of the target license plate number is the same as the license plate colors of other license plate numbers in the database; if the same, the third distance is 0; if different, the third distance is 1.

[0047] In this embodiment, the weights of the first distance, the second distance, and the third distance can be defined by itself. For example, the weights of the first distance, the second distance, and the third distance can be set to 2, 1, and 1 respectively. Let each distance be multiplied by its own weight and then summed to obtain the similarity distance.

[0048] Based on the above process, in this embodiment, the similarity distances between the target license plate number and each license plate number in the database except the target license plate number can be obtained, and the license plate numbers with similarity distances less than the similarity threshold are selected as the similar license plate numbers of the target license plate number. All the similar license plate numbers can be used to construct a similar license plate alternative library, so as to perform similar license plate identification on the target vehicle for which trajectory credibility extraction is required, and obtain the similar license plate numbers of the target license plate number. In this embodiment, the similarity threshold can be set to 2, that is, if the similarity distance between the target license plate and a license plate number in the database does not exceed 2, then this license plate number is regarded as the similar license plate number of the target license plate number and stored in the similar license plate alternative library.

[0049] S3: Select the vehicle passing data with the license plate number being the target license plate number from the database to form a first data set; for each of the similar license plate numbers, select the vehicle passing data with the license plate number being the similar license plate number from the database to form a second data set; perform selective fusion on the first data set and the second data set corresponding to each similar license plate number to obtain a fused data set;

[0050] Preferably, before performing selective fusion on the first data set and the second data set corresponding to each similar license plate number as shown Figure 4 The vehicle trajectory credibility extraction method of this embodiment may further include: performing abnormal data removal processing on the collected data of the target license plate number and the similar license plate numbers respectively, that is, performing abnormal data removal processing on the first data set and the second data set corresponding to each similar license plate number respectively to obtain a new first data set and a new second data set, and then performing selective fusion.

[0051] Among them, the abnormal data removal processing may include:

[0052] (1) Sort the vehicle passing data in the data set in the order of collection time; the data set is the first data set or the second data set corresponding to any similar license plate number;

[0053] (2) Compare two consecutive vehicle passing data in a loop and exclude them when the following situations occur: For any two consecutive vehicle passing data in the data set, if the two consecutive vehicle passing data are the acquisition data of the ETC gantry at the same entrance and exit, and the time interval (i.e., the difference in acquisition time) is less than the first time threshold t1, and the traveling directions are out and in respectively, it is considered that the vehicle has changed its travel intention and has not used the highway, and the two consecutive vehicle passing data are excluded; if the two consecutive vehicle passing data belong to the acquisition data of the same gantry or different directions of the same group of gantries, and the time interval is less than the second time threshold (which can be 5 minutes), it is considered duplicate acquisition, and the vehicle passing data with the earlier acquisition time in the two consecutive vehicle passing data is retained. Preferably, when considering duplicate acquisition, first determine whether there is data with a special situation value being empty in the two consecutive vehicle passing data. If there is, retain the data with the earlier acquisition time among the data with a special situation value being empty. If not, retain the data with the earlier acquisition time.

[0054] It should be noted that the same group of gantries refers to two gantries located at the same position and collecting data in different directions.

[0055] In this embodiment, by evaluating the similarity of the vehicle acquisition data of vehicles with similar license plates and performing reliable fusion on the vehicle acquisition data with high similarity, in S3, selective fusion is performed on the first data set and the second data set corresponding to each similar license plate number, and the fused data set obtained may include:

[0056] (1) Select the second data set corresponding to any similar license plate number as the third data set;

[0057] (2) Compare the data volumes of the first data set and the third data set, select the data set with the larger data volume as the reference data set, and select the data set with the smaller data volume as the evaluation data set;

[0058] The data volume in this embodiment refers to the number of vehicle passing data included in the data set. For example, if the first data set includes 5 vehicle passing data and the third data set includes 3 vehicle passing data, it is considered that the data volume of the third data set is small, and it is used as the reference data set.

[0059] (3) For each evaluation data in the evaluation data set, determine the reference data in the reference data set with the smallest time interval from the evaluation data according to the acquisition time, record the acquisition time, mileage, traveling direction of the reference data with the smallest time interval, and the time interval from this evaluation data, and use the reference data with the smallest time interval as the similar data of this evaluation data; both the evaluation data and the reference data are vehicle passing data;

[0060] (4) For each piece of evaluation data, calculate the time interval between the evaluation data and its similar data, and determine whether the minimum value of the time interval is less than the third time threshold (which can be 60 minutes) to obtain a first judgment result;

[0061] (5) If the first judgment result is yes, then for each piece of evaluation data, calculate the similarity between the evaluation data and its similar data; calculate the average value of the similarities corresponding to all the evaluation data, and determine whether the average value exceeds the average threshold (which can be 0.7) to obtain a second judgment result;

[0062] (6) If the second judgment result is yes, then consider the reference data set and the evaluation data set as similar data, replace the license plate numbers of the evaluation data set with those of the reference data set to obtain an updated set, and fuse the reference data set and the updated set to obtain a preliminary fused set; determine whether all the second data sets have been selected; if so, use the preliminary fused set as the fused data set; if not, randomly select an unselected second data set as the third data set for the next cycle, use the preliminary fused set as the first data set for the next cycle, and return to the step of "comparing the data volumes of the first data set and the third data set";

[0063] Among them, calculating the similarity between the evaluation data and its similar data may include:

[0064] (6.1) Calculate the collection point distance between the evaluation data and its similar data;

[0065] If the evaluation data and its similar data are collected at the same ETC gantry, or are collected at two consecutive ETC gantries within a specified time, then these two data are likely to come from the same vehicle. The method for calculating the collection point distance is as follows: if the evaluation data and the similar data are collected at the same ETC gantry or the same group of ETC gantries, the collection point distance is 0; if the evaluation data and the similar data are collected at two consecutive (groups) of ETC gantries, the collection point distance is 1; otherwise, the collection point distance is the sum of the number of ETC gantries passed by the evaluation data and the similar data and 1, and in this case, the collection point distance is greater than 1.

[0066] (6.2) If the collection point distance is 0, then determine whether the time interval between the evaluation data and the similar data of the evaluation data is less than the fourth time threshold (which can be 10 minutes). If so, it is regarded as data collected in the same time, and the similarity is 1; otherwise, the similarity is 0. If the collection point distance is 1, then determine whether the ratio of the distance difference (i.e., the difference in mileage) between the evaluation data and the similar data of the evaluation data to the time interval is greater than the first preset ratio s1. If so, the two pieces of data are regarded as continuous trajectories collected during the vehicle operation, and the similarity is 1; otherwise, the similarity is 0. If the collection point distance is greater than 1, then determine whether the ratio of the distance difference between the evaluation data and the similar data of the evaluation data to the time interval is greater than the second preset ratio s2. If so, the two pieces of data are regarded as trajectories collected during the vehicle operation, but there are missing trajectories in the middle, and the similarity is 1; otherwise, the similarity is 0.

[0067] (7) If the first judgment result is negative, or the second judgment result is negative, then it is regarded that there is no mutual interference between the trajectories, and it is considered that the reference data set and the evaluation data set are irrelevant vehicle data. Determine whether all the second data sets have been selected; if so, use the first data set of this cycle as the fused data set; if not, randomly select an unselected second data set as the third data set of the next cycle, use the first data set of this cycle as the first data set of the next cycle, and return to the step of "comparing the data volume sizes of the first data set and the third data set".

[0068] If the target license plate number has no similar license plate numbers, there is no need to perform the fusion step. Directly use the first data set as the fused data set and execute S4.

[0069] S4: Split the fused data set into single-trip data to obtain several single-trip trajectories of the target vehicle during the analysis period.

[0070] For the fused data set that has completed the acquisition of trustworthy data fusion, perform single-journey data splitting. S4 may include: sorting the data in the fused data set in the order of acquisition time; the data is vehicle passing data; determining the traveling direction of the (i - 1)-th data; if it is "out", then form a single trip trajectory with the starting point of this single trip and all the data within the (i - 1)-th data, use the i-th data as the starting point of the next single trip trajectory, and let the i-th data be the (i - 1)-th data in the next loop, and return to the step of "determining the traveling direction of the (i - 1)-th data" until all the data has been traversed; i > 1; if it is not "out", then determine whether the (i - 1)-th data and the i-th data need to be updated based on the traveling directions, mileage difference, and time interval between the i-th data and the (i - 1)-th data; if no update is needed, record the i-th data, and use the i-th data as the (i - 1)-th data in the next loop, and return to the step of "determining the traveling direction of the (i - 1)-th data"; if the i-th data needs to be updated, then update the i-th data, record the updated i-th data, and use the updated i-th data as the (i - 1)-th data in the next loop, and return to the step of "determining the traveling direction of the (i - 1)-th data"; if the (i - 1)-th data and the i-th data need to be updated, then update the (i - 1)-th data and the i-th data, record the updated (i - 1)-th data and the updated i-th data, and use the updated i-th data as the (i - 1)-th data in the next loop, and return to the step of "determining the traveling direction of the (i - 1)-th data"; until all the data has been traversed.

[0071] More specifically, S4 may include:

[0072] Step 1: If the amount of data in the fused data set is less than 2, it is considered that the target vehicle lacks front and rear trajectories, and the analysis ends; otherwise, sort the data in the fused data set in chronological order, record the first data, let the variable indicating whether it is the end point be panduan and assign it a value of 0, and the data extraction pointer number i = 2;

[0073] Step 2: Determine the traveling direction of the (i - 1)-th data. If the traveling direction is "out", or panduan = 1, then consider the (i - 1)-th data as the termination point of this trip, save the recorded data as single trip trajectory data, and record the i-th data as the starting point of a new trip, panduan = 0, i = i + 1; otherwise, make the following judgment:

[0074] (1) If the traveling direction of the (i - 1)-th data is "in":

[0075] If the traveling direction of the i-th data is upward, the mileage where the data is located is greater than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then the i-th data and the (i - 1)-th data are regarded as belonging to the same trip, record the i-th data, and i = i + 1;

[0076] If the traveling direction of the i-th data is downward, the mileage where the data is located is greater than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then the i-th data and the (i - 1)-th data are regarded as belonging to the same trip, change the traveling direction of the i-th data to upward, update the mileage of the i-th data to the mileage of the nearest gantry location, record the i-th data, and i = i + 1;

[0077] If the traveling direction of the i-th data is downward, the mileage where the data is located is less than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then the i-th data and the (i - 1)-th data are regarded as belonging to the same trip, record the i-th data, and i = i + 1;

[0078] If the traveling direction of the i-th data is upward, the mileage where the data is located is less than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then the i-th data and the (i - 1)-th data are regarded as belonging to the same trip, change the traveling direction of the i-th data to downward, update the mileage of the i-th data to the mileage of the nearest gantry location, record the i-th data, and i = i + 1;

[0079] If none of the above situations are met, it is regarded as the end point of the trip, panduan = 1, and return to step 2.

[0080] (2) If the traveling direction of the (i - 1)-th data is upward:

[0081] If the traveling direction of the i-th data is upward, the mileage where the data is located is greater than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then the i-th data and the (i - 1)-th data are regarded as belonging to the same trip, record the i-th data, and i = i + 1;

[0082] If the driving direction of the i-th data is downward, the mileage where the data is located is greater than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then it is considered that the i-th data and the (i - 1)-th data belong to the same trip. Change the driving direction of the i-th data to upward, update the mileage of the i-th data to the mileage of the nearest gantry location, record the i-th data, and i = i + 1;

[0083] If the driving direction of the i-th data is downward, the mileage where the data is located is less than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then it is considered that the i-th data and the (i - 1)-th data belong to the same trip. Update the driving direction of the (i - 1)-th data to downward, and update the mileage of the i-th data to the mileage point of the nearest ETC gantry in the same driving direction, record the i-th data, and i = i + 1;

[0084] If neither of the above situations is met, it is regarded as the end point of the trip, panduan = 1, and return to step 2.

[0085] (3) If the driving direction of the (i - 1)-th data is downward:

[0086] If the driving direction of the i-th data is downward, the mileage where the data is located is less than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then it is considered that the i-th data and the (i - 1)-th data belong to the same trip. Record the i-th data, and i = i + 1;

[0087] If the driving direction of the i-th data is upward, the mileage where the data is located is less than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then it is considered that the i-th data and the (i - 1)-th data belong to the same trip. Change the driving direction of the i-th data to downward, update the mileage of the i-th data to the mileage of the nearest gantry location, record the i-th data, and i = i + 1;

[0088] If the driving direction of the i-th data is upward, the mileage where the data is located is greater than the mileage of the (i - 1)-th data, and the vehicle speed or time from the mileage of the (i - 1)-th data to the mileage of the i-th data meets the specified threshold requirements s3 and t2, then it is considered that the i-th data and the (i - 1)-th data belong to the same trip. Update the driving direction of the (i - 1)-th data to upward, and update the mileage of the i-th data to the mileage point of the nearest ETC gantry in the same driving direction, record the i-th data, and i = i + 1;

[0089] If none of the above conditions are met, it is regarded as the end point of the trip, panduan = 1, and return to step 2.

[0090] Step 3: If all data have been traversed, record the license plate number, starting point, ending point, starting time, and ending time of the travel trajectory recognized as the same trip as a single trip trajectory, as shown in Figure 5 shown.

[0091] After obtaining several single trip trajectories of the target vehicle within the analysis period, the vehicle trajectory credibility extraction method of this embodiment further includes: performing a missing judgment on each single trip trajectory, and marking the single trip trajectories with missing data.

[0092] Specifically, for the single trip trajectory data extracted as belonging to the same trip, judge whether the starting and ending point acquisition data of the single trip are located at the toll station or the ETC gantry at the start and end of the traveling direction, judge whether there is data missing at the starting and ending points, judge whether there is a lack of ETC gantry points between two adjacent acquisition data in terms of time, judge whether there is an omission of ETC gantry perception data, and record the data missing situation at the starting and ending points and the passing ETC gantries, providing a basis for toll auditing, ETC operation status research and judgment, and further development and utilization of ETC data.

[0093] To overcome the problems of frequent occurrence of ETC gantry data issues and insufficient accuracy of vehicle trajectory identification on existing highways, this embodiment provides a method for reliably extracting vehicle trajectories based on ETC gantry data, including the following steps: 1) Basic data collection: Obtain the initial data collected by the ETC gantries at the entrance and exit toll stations, section ETC gantries, and branch ramp ETC gantries of a certain highway. 2) Data preprocessing: Calculate the mileage position where the initial data is located, mark the traveling direction of the collected data, standardize the license plate data, eliminate duplicate data, and build a database to obtain the database. 3) Similar license plate identification: For the selected target license plate, extract the license plate numbers of the passing data of other vehicles in the database, identify the similarity distance between each license plate number and the target license plate through the weighted edit distance, select the license plate numbers with a similarity distance lower than the similarity threshold as similar license plate numbers, and extract the passing data of the vehicles with similar license plate numbers as potential trajectory supplementary data. 4) Abnormal data elimination processing: Eliminate duplicate data such as changed travel intentions, repeated collection by ETC gantries, and collection by two-way ETC gantries. 5) Reliable fusion of similar data: Construct a spatio-temporal similarity scoring method, analyze the spatio-temporal similarity of the collected data with similar license plates, and fuse the collected data that meets the similarity evaluation criteria. 6) Single journey data splitting: By identifying the changes in the spatio-temporal characteristics of continuously collected data, split the perception data and record the single mileage situation. 7) Missing data annotation: Judge the missing data situation through the starting and ending traveling direction annotation and the situation of the ETC gantry points passed between two consecutive data, and annotate the missing data. Through data correction, fusion, and splitting, the reliable extraction of the vehicle trajectory of a single vehicle trip is realized, providing a data basis for highway toll auditing and further development and utilization of ETC data.

[0094] Embodiment 2:

[0095] This embodiment is used to provide a system for reliably extracting vehicle trajectories based on ETC gantry data, as Figure 6 shown. The vehicle trajectory reliable extraction system includes:

[0096] A construction module M1 for constructing a database; the database includes a number of vehicle passing data collected by each ETC gantry arranged on the highway during the analysis period; the vehicle passing data includes mileage, license plate number, collection time, and traveling direction; the license plate number includes a provincial abbreviation, a city code plus a sequence number, and a license plate color; the traveling direction includes out, in, up, and down;

[0097] A similar license plate number determination module M2 for respectively calculating the similarity distance between the target license plate number of the target vehicle of the trajectory to be extracted and each of the license plate numbers in the database other than the target license plate number, and selecting the license plate numbers with the similarity distance less than the similarity threshold as the similar license plate numbers of the target license plate number;

[0098] The data fusion module M3 is configured to select the vehicle passing data of the vehicle with the license plate number being the target license plate number from the database to form a first data set; for each of the similar license plate numbers, select the vehicle passing data of the vehicle with the license plate number being the similar license plate number from the database to form a second data set; perform selective fusion on the first data set and the second data set corresponding to each similar license plate number to obtain a fused data set;

[0099] The splitting module M4 is configured to split the fused data set into single-trip data to obtain several single-trip trajectories of the target vehicle during the analysis period.

[0100] In each embodiment of this specification, the key point is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple. For the relevant parts, refer to the description in the method section.

[0101] In this article, specific examples are used to elaborate on the principles and implementation manners of the present invention. The descriptions of the above embodiments are only used to help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for reliably extracting vehicle trajectories based on ETC gantry data, characterized in that, The vehicle trajectory reliable extraction method includes: Constructing a database; the database includes a number of vehicle passing data collected by each ETC gantry arranged on the highway during the analysis period; the vehicle passing data includes mileage, license plate number, collection time, and traveling direction; Calculating the similarity distance between the target license plate number of the target vehicle whose trajectory is to be extracted and each license plate number in the database except the target license plate number, and selecting the license plate numbers with the similarity distance less than the similarity threshold as the similar license plate numbers of the target license plate number; Selecting the vehicle passing data with the license plate number being the target license plate number from the database to form a first data set; for each of the similar license plate numbers, selecting the vehicle passing data with the license plate number being the similar license plate number from the database to form a second data set; performing selective fusion on the first data set and the second data sets corresponding to each of the similar license plate numbers to obtain a fused data set, specifically including: Selecting the second data set corresponding to any one of the similar license plate numbers as a third data set; Comparing the data volume sizes of the first data set and the third data set, selecting the data set with the larger data volume as the reference data set, and selecting the data set with the smaller data volume as the evaluation data set; For each piece of evaluation data in the evaluation data set, determining the reference data in the reference data set with the smallest time interval from the evaluation data according to the collection time, and using the reference data with the smallest time interval as the similar data of the evaluation data; both the evaluation data and the reference data are the vehicle passing data; For each piece of the evaluation data, calculating the time interval between the evaluation data and the similar data of the evaluation data, and judging whether the minimum value of the time interval is less than a third time threshold to obtain a first judgment result; If the first judgment result is yes, then for each piece of the evaluation data, calculating the similarity between the evaluation data and the similar data of the evaluation data; calculating the average value of the similarities corresponding to all the evaluation data, and judging whether the average value exceeds an average threshold to obtain a second judgment result; If the second judgment result is yes, then replacing the license plate number of the evaluation data set with the license plate number of the reference data set to obtain an updated set, and fusing the reference data set and the updated set to obtain a preliminary fused set; judging whether all the second data sets have been selected; if so, using the preliminary fused set as the fused data set; Performing single-trip data splitting on the fused data set to obtain a number of single-trip trajectories of the target vehicle during the analysis period.

2. The vehicle trajectory credible extraction method according to claim 1, wherein The constructing of the database specifically includes: Obtaining a number of initial data collected by each ETC gantry arranged on the highway during the analysis period; the initial data includes the ETC gantry stake number, the initial license plate number, the collection time, and the vehicle flow direction sensed by the ETC gantry; For each piece of the initial data, convert the ETC gantry stake number to mileage; perform data standardization processing on the initial license plate number to obtain the license plate number; determine the traveling direction according to the traffic flow direction sensed by the ETC gantry to obtain vehicle passing data; the license plate number includes a provincial abbreviation, a city code plus a sequence number, and a license plate color; the traveling direction includes out, in, up, and down; all the vehicle passing data forms a database.

3. The vehicle trajectory credible extraction method according to claim 1, characterized in that The specific calculation of the similarity distance between the target license plate number of the target vehicle for which the trajectory is to be extracted and each license plate number in the database other than the target license plate number includes: For each license plate number in the database other than the target license plate number, use the edit distance calculation formula to calculate the edit distance between the provincial abbreviation of the target license plate number and the provincial abbreviation of the license plate number to obtain a first distance; use the edit distance calculation formula to calculate the edit distance between the city code plus the sequence number of the target license plate number and the city code plus the sequence number of the license plate number to obtain a second distance; calculate the third distance between the license plate color of the target license plate number and the license plate color of the license plate number. Perform a weighted sum of the first distance, the second distance, and the third distance to obtain the similarity distance.

4. The vehicle trajectory reliable extraction method according to claim 3, characterized in that The specific calculation of the third distance between the license plate color of the target license plate number and the license plate color of the license plate number includes: Judge whether the license plate color of the target license plate number is the same as the license plate color of the license plate number. If they are the same, the third distance is 0; if they are different, the third distance is 1.

5. The vehicle trajectory reliable extraction method according to claim 1, wherein, Before selectively fusing the first data set and the second data sets corresponding to each similar license plate number, the vehicle trajectory credible extraction method further includes: respectively performing abnormal data rejection processing on the first data set and the second data sets corresponding to each similar license plate number to obtain a new first data set and a new second data set. Among them, the abnormal data rejection processing specifically includes: Sort the vehicle passing data in the data set in the order of the collection time; the data set is the first data set or the second data set corresponding to any similar license plate number. For any two consecutive vehicle passing data in the data set, if the two consecutive vehicle passing data are the collection data of the ETC gantry at the same entrance and exit, the time interval is less than the first time threshold, and the traveling directions are out and in respectively, then the two consecutive vehicle passing data are rejected; if the two consecutive vehicle passing data belong to the collection data of the same gantry or different directions of the same group of gantries, and the time interval is less than the second time threshold, then retain the vehicle passing data with the earlier collection time among the two consecutive vehicle passing data.

6. The vehicle trajectory credible extraction method according to claim 1, characterized in that Judge whether all the second data sets have been selected; if not, randomly select an unselected second data set as the third data set for the next cycle, use the preliminary fusion set as the first data set for the next cycle, and return to the step of "comparing the data volume sizes of the first data set and the third data set". If the first judgment result is negative, or the second judgment result is negative, then it is judged whether all the second data sets have been selected; if so, the first data set of this loop is used as the fused data set; If not, a non - selected second data set is randomly selected as the third data set of the next loop, the first data set of this loop is used as the first data set of the next loop, and the step of "comparing the data volumes of the first data set and the third data set" is returned.

7. The vehicle trajectory reliable extraction method according to claim 1, wherein The calculation of the similarity between the evaluation data and the similar data of the evaluation data specifically includes: Calculating the collection point distance between the evaluation data and the similar data of the evaluation data; If the collection point distance is 0, then it is judged whether the time interval between the evaluation data and the similar data of the evaluation data is less than the fourth time threshold; if so, the similarity is 1, otherwise, the similarity is 0; If the collection point distance is 1, then it is judged whether the ratio of the distance difference between the evaluation data and the similar data of the evaluation data to the time interval is greater than the first preset ratio; if so, the similarity is 1, otherwise, the similarity is 0; If the collection point distance is greater than 1, then it is judged whether the ratio of the distance difference between the evaluation data and the similar data of the evaluation data to the time interval is greater than the second preset ratio; if so, the similarity is 1, otherwise, the similarity is 0.

8. The vehicle trajectory reliable extraction method according to claim 1, characterized in that The splitting of the fused data set into several single - trip trajectories of the target vehicle within the analysis period specifically includes: Sorting the data in the fused data set in the order of collection time; the data is the vehicle passing data; Judging the traveling direction of the (i - 1) - th data; If it is "out", then the starting point of this single - trip trajectory and all the data within the (i - 1) - th data are formed into a single - trip trajectory, the i - th data is used as the starting point of the next single - trip trajectory, and the i - th data is set as the (i - 1) - th data of the next loop, and the step of "judging the traveling direction of the (i - 1) - th data" is returned until all the data have been traversed; i > 1; If it is not "out", then it is judged whether the (i - 1) - th data and the i - th data need to be updated according to the traveling directions, mileage differences and time intervals of the i - th data and the (i - 1) - th data; if no update is needed, the i - th data is recorded, and the i - th data is used as the (i - 1) - th data of the next loop, and the step of "judging the traveling direction of the (i - 1) - th data" is returned; if the i - th data needs to be updated, the i - th data is updated, the updated i - th data is recorded, and the updated i - th data is used as the (i - 1) - th data of the next loop, and the step of "judging the traveling direction of the (i - 1) - th data" is returned; if the (i - 1) - th data and the i - th data need to be updated, the (i - 1) - th data and the i - th data are updated, the updated (i - 1) - th data and the updated i - th data are recorded, and the updated i - th data is used as the (i - 1) - th data of the next loop, and the step of "judging the traveling direction of the (i - 1) - th data" is returned; until all the data have been traversed.

9. The vehicle trajectory reliable extraction method according to claim 1, wherein After obtaining several single-trip trajectories of the target vehicle during the analysis period, the vehicle trajectory credibility extraction method further includes: performing a missing judgment on each of the single-trip trajectories, and marking the single-trip trajectories with missing data.

10. A vehicle trajectory reliable extraction system based on ETC gantry data, characterized in that, The vehicle trajectory credibility extraction system includes: A construction module for constructing a database; the database includes a number of vehicle passing data collected by each ETC gantry arranged on the highway during the analysis period; the vehicle passing data includes mileage, license plate number, collection time, and traveling direction; A similar license plate number determination module for calculating the similarity distance between the target license plate number of the target vehicle of the trajectory to be extracted and each of the license plate numbers in the database other than the target license plate number, and selecting the license plate number with the similarity distance less than the similarity threshold as the similar license plate number of the target license plate number; A data fusion module for selecting the vehicle passing data with the license plate number being the target license plate number from the database to form a first data set; for each of the similar license plate numbers, selecting the vehicle passing data with the license plate number being the similar license plate number from the database to form a second data set; performing selective fusion on the first data set and the second data set corresponding to each of the similar license plate numbers to obtain a fused data set, specifically including: Selecting the second data set corresponding to any one of the similar license plate numbers as a third data set; Comparing the data volume sizes of the first data set and the third data set, selecting the data set with the larger data volume as the reference data set, and selecting the data set with the smaller data volume as the evaluation data set; For each piece of evaluation data in the evaluation data set, determining the reference data in the reference data set with the smallest time interval from the evaluation data according to the collection time, and using the reference data with the smallest time interval as the similar data of the evaluation data; both the evaluation data and the reference data are the vehicle passing data; For each piece of the evaluation data, calculating the time interval between the evaluation data and the similar data of the evaluation data, and judging whether the minimum value of the time interval is less than a third time threshold to obtain a first judgment result; If the first judgment result is yes, then for each piece of the evaluation data, calculating the similarity between the evaluation data and the similar data of the evaluation data; calculating the average value of the similarities corresponding to all the evaluation data, and judging whether the average value exceeds the average value threshold to obtain a second judgment result; If the second judgment result is yes, then replacing the license plate number of the evaluation data set with the license plate number of the reference data set to obtain an updated set, and fusing the reference data set and the updated set to obtain a preliminary fused set; judging whether all the second data sets have been selected; if so, using the preliminary fused set as the fused data set; A splitting module for splitting the fused data set into several single-trip trajectories of the target vehicle during the analysis period.

Images