A vehicle movement feature recognition method based on trajectory information

Abstract

The present invention relates to a method for identifying vehicle movement characteristics based on trajectory information. The method, based on a distance threshold D std On the basis of identifying all suspected parking trajectory points and driving trajectory points included in vehicle trajectory data, by classifying multiple adjacent suspected parking trajectory points into the same suspected parking trajectory segment K, and relying on the parking stability S of the vehicle corresponding to each suspected parking trajectory segment K k To further judge each suspected parking trajectory segment K, so as to identify long-time parking trajectory segments and short-time parking trajectory segments, which can effectively improve the identification efficiency and accuracy, thereby providing reliable data support for subsequent analysis of the vehicle operation state within a certain time period and the operation characteristics of freight vehicles in the area.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle running state recognition, and in particular to a vehicle movement feature recognition method based on trajectory information. Background Art

[0002] With the development of social economy, urban logistics industry has developed rapidly. The development of information technology has spawned a large amount of GPS trajectory data of freight vehicles. These massive vehicle trajectory data provide effective data support for data analysts. In the existing technology, the research on vehicle trajectory data mainly includes vehicle trajectory clustering, trajectory anomaly detection, trajectory classification, etc. Among them, trajectory classification is to abstract the trajectory model by statistically analyzing the spatiotemporal characteristics of different trajectories of trajectory data, and use it as a classifier to classify the trajectory of the target vehicle.

[0003] The trajectory data of a freight vehicle within a certain time period usually consists of a series of trajectory points arranged in time series. Each trajectory point corresponds to a trajectory data, which is used to describe the spatiotemporal motion state of the freight vehicle within a certain geographical space range and time period. For example, a general freight vehicle usually uploads a record every 30 seconds as the trajectory data of a trajectory point. The trajectory data of a trajectory point usually includes the vehicle identification (carID), the time when the trajectory point is recorded or occurs (t), and the latitude and longitude coordinates (P) of the trajectory point, so as to intuitively show the geographical location of a vehicle at a certain time.

[0004] Normally, the original vehicle GPS trajectory data usually does not contain parking information, so it is often impossible to directly and effectively identify the parking behavior of the vehicle during actual operation from the original vehicle GPS trajectory data. For this reason, a Chinese invention patent with publication number CN111340427B and titled "A method for identifying the running status of a truck based on trajectory data" discloses a method for identifying the running status of a truck based on truck trajectory data. In the stage of identifying trajectory points, this method mainly calculates the distance difference and the stay time difference between two adjacent trajectory points, and compares the distance difference and the stay time difference with the preset distance threshold and time threshold, respectively, to identify the stay trajectory points and driving trajectory points in the trajectory data. However, there is often a certain ambiguity in identifying trajectory points in this way, and the recognition accuracy is not high and the error is large. Summary of the invention

[0005] The purpose of the present invention is to provide a vehicle movement feature recognition method based on trajectory information, so as to solve the technical problem of low accuracy and large error in the recognition of vehicle trajectory points in the prior art.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A vehicle movement feature recognition method based on trajectory information comprises the following steps:

[0008] S1. Obtain the trajectory data of the vehicle within a certain time period. The trajectory data includes several trajectory points P arranged in time series. i , where i=1, 2, 3, ...;

[0009] S2. Set distance threshold D std , calculate the distance difference ΔD between two adjacent trajectory points in sequence i，i+1 , the distance difference ΔD between two adjacent trajectory points i，i+1 With distance threshold D std Compare to identify suspected parking trajectory points and driving trajectory points in the trajectory data, and mark the driving trajectory points as 0 and the suspected parking trajectory points as 1;

[0010] S3. Classifying multiple adjacent suspected parking trajectory points into the same suspected parking trajectory segment K, where K = 1, 2, 3...;

[0011] S4. Setting parking stability threshold S std , calculate the parking stability S of each suspected parking trajectory segment K k , and the parking stability S of each suspected parking trajectory segment K is calculated in turn. k and parking stability threshold S std Make comparisons;

[0012] If S k ≥S std , classify the suspected parking trajectory segment K as a long-term parking trajectory segment, and record all the suspected parking trajectory points contained in the suspected parking trajectory segment K as long-term parking trajectory points;

[0013] If S k <S std , the suspected parking trajectory segment K is classified as a short-term parking trajectory segment, and all the suspected parking trajectory points contained in the suspected parking trajectory segment K are recorded as short-term parking trajectory points.

[0014] Optionally, in step S4, the parking stability S of the suspected parking trajectory segment K k Calculated by the following formula:

[0015]

[0016] Among them, T k is the cumulative dwell time of the suspected parking trajectory segment K, D k is the cumulative driving distance of the suspected parking trajectory segment K.

[0017] Optionally, the following steps are also included:

[0018] S5. Set the intra-field transfer distance threshold Din std , calculate the driving distance D between two adjacent suspected parking trajectory segments K k , k+1 , the driving distance ΔD between two adjacent suspected parking trajectory segments K k，k+1 Distance threshold Din std Comparisons are made to identify intra-court transfer trajectory segments.

[0019] Furthermore, in step S5, the driving distance ΔD between two adjacent suspected parking trajectory segments K k，k+1 is the sum of the driving distances between all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K;

[0020] At this time, if ΔD k，k+1 ≥Din std , then all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K are classified as the same non-in-field transfer trajectory segment;

[0021] If ΔD k，k+1 <Din std , then all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K are classified as the same intra-field transfer trajectory segment.

[0022] Optionally, step S2 specifically includes:

[0023] S21. The first trajectory point P in the trajectory data 1 As the calculation starting point, it is recorded as the suspected parking trajectory point and marked as 1;

[0024] S22. Calculate two adjacent trajectory points P in sequence i and P i+1 The distance difference ΔD i，i+1 , if ΔD i，i+1 <D std , then the trajectory point P i+1 Recorded as a suspected parking trajectory point, if ΔD i，i+1 ≥D std , then the trajectory point P i+1 Recorded as driving trajectory point.

[0025] Furthermore, the distance difference ΔD between two adjacent trajectory points i，i+1 The calculation is done using the haversine formula, specifically:

[0026]

[0027] Where R is the radius of the Earth, is the longitude of the trajectory point, and λ is the latitude of the trajectory point.

[0028] Optionally, the distance threshold D std The setting range is: 0km<D std <1km.

[0029] Furthermore, the intra-field transfer distance threshold Din std The setting range is: 0km<Din std <2km.

[0030] The technical solution of the embodiment of the present invention has at least the following advantages and beneficial effects:

[0031] 1. The vehicle movement feature recognition method provided by the present invention is based on the distance threshold D std Based on identifying all suspected parking trajectory points and driving trajectory points contained in the vehicle trajectory data, multiple adjacent suspected parking trajectory points are classified into the same suspected parking trajectory segment K, and the parking stability S of the vehicle corresponding to each suspected parking trajectory segment K is used to calculate the parking stability S of the vehicle. k To further judge each suspected parking trajectory segment K to identify long-term parking trajectory segments and short-term parking trajectory segments, it can effectively improve the recognition efficiency and accuracy, thereby providing reliable data support for the subsequent analysis of the vehicle's operating status within a certain time period and the operating characteristics of freight vehicles in the area.

[0032] 2. The vehicle movement feature recognition method provided by the present invention calculates the driving distance D between two adjacent suspected parking trajectory segments K based on the recognition of the driving trajectory points, long-term parking trajectory points and short-term parking trajectory points in the vehicle trajectory data. k，k+1 , and the travel distance D k，k+1 Distance threshold Din std By making comparisons, it is possible to effectively and accurately identify the transfer trajectory segments within the venue, thereby providing reliable data support for determining whether the parking behavior of the vehicle is a parking behavior generated during the transfer within the venue. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flow chart of a vehicle movement feature recognition method provided in Example 1 of the present invention;

[0034] Figure 2 A schematic diagram of identifying suspected parking trajectory points provided in Example 1 of the present invention;

[0035] Figure 3 A flow chart for determining suspected parking trajectory segments provided in Example 1 of the present invention;

[0036] Figure 4 This is a flow chart of identifying intra-field transfer trajectory segments provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0037] Example 1

[0038] Please refer to Figures 1 to 3 , this embodiment provides a vehicle movement feature recognition method based on trajectory information, comprising the following steps:

[0039] S1. Obtain the trajectory data of the vehicle within a certain time period. The trajectory data includes several trajectory points P arranged in time series. i , where i=1, 2, 3... It can be understood that the trajectory data of a vehicle within a certain time period can be the trajectory data of the vehicle within 24 hours a day, or the trajectory data of the vehicle within a certain period of time. In this case, P 1 Represents the data of the first track point in the vehicle track data, P 2 Represents the data of the second trajectory point in the vehicle trajectory data, and so on.

[0040] It should be noted that, considering that the original trajectory data of the vehicle often contains certain duplicate data, abnormal data and missing data, before performing feature recognition on the vehicle trajectory data, a series of trajectory data of the vehicle within a certain period of time should be preprocessed to further improve the accuracy of feature recognition using the vehicle trajectory data. Preprocessing the vehicle trajectory data includes steps such as processing duplicate values, processing missing values ​​and processing abnormal values ​​on the vehicle trajectory data.

[0041] S2. After obtaining the trajectory data of the vehicle within a certain period of time, set the distance threshold D std , calculate the distance difference ΔD between two adjacent trajectory points in sequence i，i+1 , the distance difference ΔD between two adjacent trajectory points i，i+1 With distance threshold D std Compare and identify the suspected parking trajectory points and driving trajectory points in the trajectory data, and mark the driving trajectory points as 0 and the suspected parking trajectory points as 1.

[0042] It can be understood that, based on the different operating conditions of different vehicles, the distance threshold D in this embodiment may be std It is set as an adjustable parameter and can be set according to the actual situation in actual implementation. For example, the distance threshold D in this embodiment std The setting range is: 0km<D std <1km.

[0043] In this embodiment, combined with Figure 2 The schematic diagram of identifying suspected parking trajectory points and driving trajectory points shown in FIG. 1 is a schematic diagram of identifying suspected parking trajectory points and driving trajectory points shown in FIG. 1 . The distance difference ΔD between two adjacent trajectory points is calculated in the above step S2. i，i+1 , and through the distance difference ΔD i，i+1 The process of identifying suspected parking trajectory points and driving trajectory points in the trajectory data specifically includes:

[0044] S21. The first trajectory point P in the trajectory data 1 As the calculation starting point, it is recorded as the suspected parking trajectory point and marked as 1;

[0045] S22. Calculate two adjacent trajectory points P in sequence i and P i+1 The distance difference ΔD i，i+1 , if ΔD i，i+1 <D std , indicating that the vehicle moves from trajectory point P i To trajectory point P i+1 The distance traveled during this process is less than the distance threshold D std , then the trajectory point P i+1 is recorded as a suspected parking trajectory point and marked as 1; otherwise, if ΔD i，i+1 ≥D std , indicating that the vehicle moves from trajectory point P i To trajectory point P i+1 The distance traveled during this process is greater than the distance threshold D std , then the trajectory point P i+1 It is recorded as a driving trajectory point and marked as 0.

[0046] Repeat the above operation until all the track points in the vehicle track data are identified, so as to identify all the suspected parking track points and driving track points.

[0047] It should be noted that the distance difference ΔD between two adjacent trajectory points i，i+1 The half-haversine formula can be used for calculation, specifically:

[0048]

[0049] Where R is the radius of the Earth, is the longitude of the trajectory point, and λ is the latitude of the trajectory point.

[0050] S3. After all suspected parking trajectory points and all driving trajectory points included in the vehicle trajectory data are identified, in order to improve the efficiency of subsequent identification of long-term parking trajectory points and short-term parking trajectory points, multiple adjacent suspected parking trajectory points are classified into the same suspected parking trajectory segment K, where K = 1, 2, 3, ..., which is used to indicate the number of the suspected parking trajectory segment. For example, continue to refer to Figure 2 , if P 1 and P 2 are all suspected parking trajectory points, then P 1 and P 2 Classified as the same suspected parking trajectory segment K, K = 1, indicating the first suspected parking trajectory segment; if P 3 , P 4 and P 5 are all driving trajectory points, then P 3 , P 4 and P 5 are classified as the same driving trajectory segment; if P 6 , P 7 , P 8 and P 9 are all suspected parking trajectory points, then P 6 , P 7 , P 8 and P 9 The same suspected parking trajectory segment K is classified, and K=2 at this time, indicating the second suspected parking trajectory segment. The same process is repeated until all the suspected parking trajectory points included in the vehicle trajectory data are classified into multiple different suspected parking trajectory segments K.

[0051] S4. After all suspected parking trajectory segments K are identified, based on the fact that vehicles often park for a certain period of time and have a certain stability in time within the interval corresponding to the suspected parking trajectory segment K, the parking stability S of the vehicle corresponding to each suspected parking trajectory segment K is further analyzed. k , it can effectively determine whether the suspected parking trajectory segment K belongs to a long-term parking trajectory segment or a short-term parking trajectory segment. At the same time, by considering the parking stability S of the vehicle k As the basis for judging whether the suspected parking trajectory segment K belongs to a long-term parking trajectory segment or a short-term parking trajectory segment, it can effectively improve the accuracy of recognition, thereby more accurately identifying the long-term parking trajectory points and short-term parking trajectory points in the vehicle trajectory data.

[0052] Specifically, the parking stability threshold S is first set std , where the parking stability threshold S std It is also set as an adjustable parameter, which can be set according to the actual situation during the actual implementation. Figure 3 The flow chart of judging suspected parking trajectory segments shown in FIG. 1 calculates the parking stability S of each suspected parking trajectory segment K. k , and the parking stability S of each suspected parking trajectory segment K is calculated in turn. k and parking stability threshold S std Make comparisons;

[0053] At this time, if S k ≥S std , the suspected parking trajectory segment K is classified as a long-term parking trajectory segment, and all the suspected parking trajectory points contained in the suspected parking trajectory segment K are recorded as long-term parking trajectory points, indicating that the parking behavior of the vehicle at these trajectory points belongs to long-term parking behavior, for example, the parking behavior when the driver is resting or loading and unloading goods;

[0054] On the contrary, if S k <S std , the suspected parking trajectory segment K is classified as a short-term parking trajectory segment, and all the suspected parking trajectory points contained in the suspected parking trajectory segment K are recorded as short-term parking trajectory points, indicating that the parking behavior of the vehicle at these trajectory points belongs to a short-term parking behavior during driving, for example, the parking behavior when waiting for a traffic light or stopping for refueling.

[0055] Repeat the above operation until all suspected parking trajectory segments K are identified.

[0056] In this embodiment, the parking stability S of the suspected parking trajectory segment K in the above step S4 is k Calculated by the following formula:

[0057]

[0058] Among them, T k is the cumulative dwell time of the suspected parking trajectory segment K, D k is the cumulative driving distance of the suspected parking trajectory segment K. It can be understood that all the trajectory points in the vehicle trajectory data are arranged in time series. Therefore, by obtaining the time at which the first suspected parking trajectory point in the suspected parking trajectory segment K and the time at which the last suspected parking trajectory point is located, the cumulative stop time T of the vehicle corresponding to the suspected parking trajectory segment K can be calculated. k Similarly, by obtaining the driving distances between all the suspected parking trajectory points of the suspected parking trajectory segment K, the cumulative driving distance D of the vehicle corresponding to the suspected parking trajectory segment K can be calculated. k .

[0059] It can be seen that the vehicle movement feature recognition method provided in this embodiment is based on the distance threshold D std Based on identifying all suspected parking trajectory points and driving trajectory points contained in the vehicle trajectory data, multiple adjacent suspected parking trajectory points are classified into the same suspected parking trajectory segment K, and the parking stability S of the vehicle corresponding to each suspected parking trajectory segment K is used to calculate the parking stability S of the vehicle. kTo further judge each suspected parking trajectory segment K to identify long-term parking trajectory segments and short-term parking trajectory segments, it can effectively improve the recognition efficiency and accuracy, thereby providing reliable data support for the subsequent analysis of the vehicle's operating status within a certain time period and the operating characteristics of freight vehicles in the area.

[0060] Example 2

[0061] Considering that in general, when analyzing the running status of a vehicle, it is often necessary to analyze whether the vehicle is transferred within the site (i.e., moving within a specific operating area, for example, moving within the factory area), therefore, on the basis of identifying the driving trajectory points, long-term parking trajectory points, and short-term parking trajectory points contained in the vehicle trajectory data in Example 1, a step of determining whether the vehicle is transferred within the site is added, specifically including:

[0062] S5. Set the intra-field transfer distance threshold Din std , calculate the driving distance D between two adjacent suspected parking trajectory segments K k , k+1 , the driving distance ΔD between two adjacent suspected parking trajectory segments K k，k+1 Distance threshold Din std The intra-field transfer distance threshold Din in this embodiment is compared to identify the intra-field transfer trajectory segment. std It is also set as an adjustable parameter and can be set according to the actual situation in actual implementation. For example, when a general freight vehicle is transferred within the field, the distance it travels is usually not more than 2 km. Therefore, in this embodiment, the field transfer distance threshold Din is std The setting range is: 0km<Din std <2km.

[0063] Specifically, in the above step S5, the driving distance ΔD between two adjacent suspected parking trajectory segments K is k，k+1 is the sum of the driving distances between all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K;

[0064] At this time, if ΔD k，k+1 ≥Din std , all the suspected parking trajectory points and all the driving trajectory points between the two adjacent suspected parking trajectory segments K are classified as the same non-in-field transfer trajectory segment, indicating that the parking behaviors corresponding to all the suspected parking trajectory points contained in the current two adjacent suspected parking trajectory segments K are not parking behaviors during in-field transfer.

[0065] If ΔD k，k+1 <Din std, all the suspected parking trajectory points and all the driving trajectory points between two adjacent suspected parking trajectory segments K are classified as the same intra-field transfer trajectory segment, indicating that the parking behaviors corresponding to all the suspected parking trajectory points contained in the current two adjacent suspected parking trajectory segments K are parking behaviors during intra-field transfer.

[0066] It can be seen that the vehicle movement feature recognition method provided in this embodiment calculates the driving distance D between two adjacent suspected parking trajectory segments K on the basis of identifying the driving trajectory points, long-term parking trajectory points and short-term parking trajectory points in the vehicle trajectory data. k，k+1 , and travel distance D k，k+1 Distance threshold Din std By making comparisons, it is possible to effectively and accurately identify the transfer trajectory segments within the venue, thereby providing reliable data support for determining whether the parking behavior of the vehicle is a parking behavior generated during the transfer within the venue.

[0067] In addition, it should be noted that the vehicle movement feature recognition method provided in this embodiment is not only applicable to identifying the trajectory data of freight vehicles, but also to identifying the trajectory data of other similar vehicles. For example, the method provided in this embodiment can also be used to identify the trajectory data of passenger vehicles such as taxis or long-distance and short-distance passenger buses, so as to analyze the operating status of passenger vehicles within a certain time period and the operating characteristics of passenger vehicles in the area.

[0068] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A vehicle movement feature recognition method based on trajectory information, characterized in that: The following steps are involved: S1. Obtain the trajectory data of the vehicle within a certain time period. The trajectory data includes several trajectory points P arranged in time series. i , where i=1, 2, 3, ...; S2. Set distance threshold D std , calculate the distance difference ΔD between two adjacent trajectory points in sequence i，i+1 , the distance difference ΔD between two adjacent trajectory points i，i+1 With distance threshold D std Compare to identify suspected parking trajectory points and driving trajectory points in the trajectory data, and mark the driving trajectory points as 0 and the suspected parking trajectory points as 1; S3. Classifying multiple adjacent suspected parking trajectory points into the same suspected parking trajectory segment K, where K=1, 2, 3...; S4. Setting parking stability threshold S std , calculate the parking stability S of each suspected parking trajectory segment K k , and the parking stability S of each suspected parking trajectory segment K is calculated in turn. k and parking stability threshold S std Make comparisons; If S k ≥S std , classify the suspected parking trajectory segment K as a long-term parking trajectory segment, and record all the suspected parking trajectory points contained in the suspected parking trajectory segment K as long-term parking trajectory points; If S k <S std , classify the suspected parking trajectory segment K as a short-term parking trajectory segment, and record all the suspected parking trajectory points contained in the suspected parking trajectory segment K as short-term parking trajectory points; S5. Set the intra-field transfer distance threshold Din std , calculate the driving distance D between two adjacent suspected parking trajectory segments K k，k+1 , the driving distance ΔD between two adjacent suspected parking trajectory segments K k，k+1 Distance threshold Din std Comparison is made to identify intra-court transfer trajectory segments; In step S4, the parking stability S of the suspected parking trajectory segment K k Calculated by the following formula: Among them, T k is the cumulative dwell time of the suspected parking trajectory segment K, D k is the cumulative driving distance of the suspected parking trajectory segment K; In step S5, the driving distance ΔD between two adjacent suspected parking trajectory segments K k，k+1 is the sum of the driving distances between all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K; At this time, if ΔD k，k+1 ≥Din std , then all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K are classified as the same non-in-field transfer trajectory segment; If ΔD k，k+1 <Din std , then all suspected parking trajectory points and all driving trajectory points between two adjacent suspected parking trajectory segments K are classified as the same intra-field transfer trajectory segment.

2. The vehicle movement feature recognition method based on trajectory information according to claim 1 is characterized in that: Step S2 specifically includes: S21. The first trajectory point P1 in the trajectory data is used as the calculation starting point, and is recorded as a suspected parking trajectory point, marked as 1; S22. Calculate two adjacent trajectory points P in sequence i and P i+1 The distance difference ΔD i，i+1 , if ΔD i，i+1 <D std , then the trajectory point P i+1 Recorded as a suspected parking trajectory point, if ΔD i，i+1 ≥D std , then the trajectory point P i+1 Recorded as driving trajectory point.

3. The vehicle movement feature recognition method based on trajectory information according to claim 2 is characterized in that: The distance difference between two adjacent trajectory points ΔD i，i+1 The calculation is done using the haversine formula, specifically: Among them, R is the radius of the earth, φ is the longitude of the trajectory point, and λ is the latitude of the trajectory point.

4. The vehicle movement feature recognition method based on trajectory information according to claim 1 is characterized in that: Distance threshold D std The setting range is: 0 km <D std <1 km .

5. The vehicle movement feature recognition method based on trajectory information according to claim 1 is characterized in that: Intra-field transfer distance threshold Din std The setting range is: 0 km <Din std <2 km .

Images