A driving style classification method based on time series trajectory data under speed limit conditions

Abstract

The present invention provides a driving style classification method based on time series trajectory data under speed limit conditions, comprising: inputting a time series trajectory data set of a vehicle to be tested into a constructed driving style recognition model to obtain a driving style of a driver of the vehicle to be tested; the present invention solves the problem that traditional driving style classification uses cross-sectional traffic flow parameters, resulting in poor driving style classification characteristics, and also solves the problem that traditional driving style classification relies on vehicle-mounted data collection methods and driving simulation collection methods, resulting in limited data volume and insufficient driving style sample volume.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a driving style classification method based on time series trajectory data under speed limit conditions. Background Art

[0002] With the rapid development of digital technology and its popular application in all walks of life, the public's expectations and demands for intelligent and smart highway services are becoming more and more urgent. However, due to differences in drivers' personalities and driving habits, different drivers have personalized driving needs. How to use modern science and technology to provide personalized services for the personalized needs of highway users is an important direction for future highway services and technology upgrades.

[0003] With the development of traffic sensing technology, especially the cost of high-precision sensors such as millimeter-wave radar and lidar has gradually decreased, and its scope of application has gradually expanded to the transportation field. In particular, the application of traffic radar has been gradually promoted in recent years, and the full-time and space high-precision vehicle trajectory perception technology has gradually become possible, providing a reliable and feasible technical path for building a digital twin world of transportation where everything is connected. Digital twin refers to the virtual mapping of physical systems, which provides new solutions and high-precision analysis methods for high-reliability micro-traffic analysis, making high-precision micro-traffic analysis under complex road conditions possible. On the other hand, with the development of artificial intelligence technologies such as machine learning and deep learning, as well as the gradual improvement of computing power, data analysis has gradually developed in the direction of refinement. The previous driving style analysis methods were mostly based on cross-sectional traffic flow data and traffic flow characteristic parameter data. The driving style classification features were poor and the traditional driving style classification relied on the vehicle-mounted data collection method and the driving simulation collection method, resulting in limited data volume and insufficient driving style sample size. The traditional driving style recognition method is easily affected by the data form. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a driving style classification method based on time series trajectory data under speed limit conditions. The present invention solves the problem that the traditional driving style classification uses cross-sectional traffic flow parameters, resulting in poor driving style classification characteristics, and at the same time solves the problem that the traditional driving style classification relies on vehicle-mounted data collection methods and driving simulation collection methods, resulting in limited data volume and insufficient driving style sample volume.

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

[0006] A driving style classification method based on time series trajectory data under speed limit conditions, comprising:

[0007] Input the time series trajectory data set of the vehicle to be tested into the constructed driving style recognition model to obtain the driving style of the driver of the vehicle to be tested;

[0008] The method for constructing the driving style recognition model is as follows:

[0009] Millimeter-wave radars or laser radars are deployed on the expressway at preset intervals, and the millimeter-wave radars or laser radars are used to collect time-series trajectory datasets of training vehicles;

[0010] Lane-level variable speed limit plates are arranged at preset intervals on the expressway, and based on a preset control strategy, a preset variable information system is used to issue corresponding lane-level speed limit control instructions, adjust the speed limit values ​​of the lane-level variable speed limit plates, and obtain the speed limit values ​​of each lane-level variable speed limit plate;

[0011] Calculate the compliance degree of each driver of the tested vehicle;

[0012] Obtaining a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition according to the compliance degree of each driver of the vehicle to be tested, the speed limit value of each lane-level variable speed limit plate, and the time-series trajectory data of the vehicle to be tested;

[0013] Based on the Kshape clustering algorithm, cluster analysis is performed on the compliance time series rule data set of the vehicles to be tested under the various speed limit conditions to obtain the compliance time series data feature set of the vehicles to be tested under the various speed limit conditions;

[0014] Performing Hausdorff distance calculation on the time series matrices of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition to obtain an initial matrix of the time series matrices of each category;

[0015] Based on weighted calculation, a distance matrix is ​​obtained according to the initial matrix of the time series matrix of each category;

[0016] Clustering the distance matrix using a Kshape model to obtain a driver style database for the vehicle to be tested;

[0017] Based on the XGBoost model, the driving style recognition model is constructed according to the driver style type database of the vehicle to be tested and the time series trajectory data set.

[0018] Preferably, the collecting of the time series trajectory data set of the vehicle to be tested by using millimeter wave radar or laser radar includes:

[0019] A plurality of millimeter wave radars or laser radars are arranged at a preset distance on both sides of the expressway, wherein the preset distance is smaller than the scanning range of the two millimeter wave radars or laser radars;

[0020] The millimeter wave radar or laser radar is used to collect a time series trajectory data set of the vehicle to be tested.

[0021] Preferably, the calculation formula for the compliance degree of each driver of the vehicle to be tested is:

[0022]

[0023] Among them, Comprate i (t) is the compliance degree of vehicle i at time t, is the speed of vehicle i at 250 m upstream of the lane-level variable speed limit plate, VSL is the speed of vehicle i at 200 m downstream of the lane-level variable speed limit plate. i (t) is the variable speed limit issued to vehicle i at time t.

[0024] Preferably, the step of obtaining a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition according to the compliance degree of each driver of the vehicle to be tested, the speed limit value of each lane-level variable speed limit plate, and the time-series trajectory data of the vehicle to be tested comprises:

[0025] The compliance level of each driver of the vehicle to be tested is differentiated into compliance levels to obtain high compliance, medium compliance and low compliance;

[0026] According to the high compliance, medium compliance, low compliance, the speed limit values ​​of each lane-level variable speed limit plate and the time-series trajectory data of the vehicle to be tested, a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition is obtained.

[0027] Preferably, the Hausdorff distance calculation is performed on the time series matrices of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition to obtain the initial matrix of the time series matrices of each category, including:

[0028] Extracting the time series matrix of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition, wherein the time series matrix of each category includes: speed time series matrix, rule time series matrix, acceleration time series matrix and jerk time series matrix;

[0029] Hausdorff distance calculation is performed on the speed timing matrix, the rule timing matrix, the acceleration timing matrix and the jerk timing matrix to obtain the initial matrix of each category timing matrix.

[0030] Preferably, the calculation formula for obtaining the distance matrix is:

[0031] distancemat

[0032] =w0*distancemat_tra+w1*distancemat_speed+w2*

[0033] distancemat_acc+w3*distancemat_jerk;

[0034] Wherein, distancemat_tra is the bidirectional Hausdorff distance of PositionX and PositionY, distancemat_speed is the bidirectional Hausdorff distance of VelocityX and VelocityY, distancemat_acc is the bidirectional Hausdorff distance of AccX and AccY, and distancemat_jerk is the one-way Hausdorff distance of jerk, where PositionX is the longitudinal position of the vehicle, PositionY is the lateral position of the vehicle, VelocityX is the longitudinal speed of the vehicle, VelocityY is the lateral speed of the vehicle, AccX is the longitudinal acceleration of the vehicle, AccY is the lateral acceleration of the vehicle, and jerk is the longitudinal jerk value of the vehicle, where w ０ ,w1,w2,w3 are weighting coefficients, w0+w1+w2+w3=1.

[0035] Preferably, clustering the distance matrix using the Kshape model to obtain a driver style database for the vehicle to be tested includes:

[0036] According to the compliance degree of the driver of the vehicle to be tested, the driver is divided into aggressive high compliance, patient high compliance, cautious high compliance, aggressive medium compliance, patient medium compliance, cautious medium compliance, aggressive low compliance, patient low compliance and cautious low compliance;

[0037] A driver style type database of the vehicle to be tested is constructed according to the aggressive high compliance, patient high compliance, cautious high compliance, aggressive medium compliance, patient medium compliance, cautious medium compliance, aggressive low compliance, patient low compliance and cautious low compliance.

[0038] Preferably, the XGBoost model is used to construct the driving style recognition model according to the driver style database of the vehicle to be tested and the time series trajectory data set, including:

[0039] Performing dimensionality reduction processing on the time series trajectory data set to obtain a time series trajectory data set after dimensionality reduction;

[0040] The driving style recognition model is constructed according to the reduced-dimensional time series trajectory data set and the driver style type database of the vehicle to be tested.

[0041] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0042] The present invention provides a driving style classification method based on time series trajectory data under speed limit conditions. The present invention considers lane-level speed limits and uses time series trajectory data to construct a driving style recognition model, and uses the driving style recognition model to predict the driver's driving style, thereby improving the accuracy of style prediction classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A flow chart of a driving style classification method based on time series trajectory data under speed limit conditions provided by an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of a driving style classification method based on time series trajectory data under speed limit conditions provided by an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of a method for deploying a variable speed limit information release device in a driving style classification method based on time series trajectory data under speed limit conditions provided by an embodiment of the present invention;

[0047] Figure 4 A schematic diagram of a millimeter wave radar and a laser radar deployment method for a driving style classification method based on time series trajectory data under speed limit conditions provided by an embodiment of the present invention;

[0048] Figure 5 A schematic diagram of the Hausdorff distance calculation principle in a driving style classification method based on time series trajectory data under speed limit conditions provided by an embodiment of the present invention;

[0049] Figure 6 A schematic diagram of a method for determining trajectory characteristic parameters in a driving style classification method based on time series trajectory data under speed limit conditions provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0052] The terms "first", "second", "third" and "fourth" in the specification and claims of the present application and the drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a series of steps, processes, methods, etc. are not limited to the listed steps, but may optionally include steps that are not listed, or may optionally include other step elements inherent to these processes, methods, products or devices.

[0053] The purpose of the present invention is to provide a driving style classification method based on time series trajectory data under speed limit conditions. The present invention solves the problem that the traditional driving style classification uses cross-sectional traffic flow parameters, resulting in poor driving style classification characteristics, and at the same time solves the problem that the traditional driving style classification relies on vehicle-mounted data collection methods and driving simulation collection methods, resulting in limited data volume and insufficient driving style sample volume.

[0054] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] like Figure 1 As shown, the present invention provides a driving style classification method based on time series trajectory data under speed limit conditions, comprising:

[0056] Step 100: Inputting the time series trajectory data set of the vehicle to be tested into the constructed driving style recognition model to obtain the driving style of the driver of the vehicle to be tested;

[0057] The method for constructing the driving style recognition model is as follows:

[0058] like Figure 2 As shown, step 101: deploying millimeter wave radars or laser radars at preset intervals on the expressway, and using the millimeter wave radars or laser radars to collect a time series trajectory data set of a training vehicle;

[0059] Step 102: A series of lane-level variable speed limit information release devices are arranged at a certain interval on a free-flow highway section. The arrangement interval of the lane-level variable speed limit information release devices can be selected to be 600 meters, as shown in the attached figure. Figure 3As shown, based on the preset control strategy, the preset variable information system is used to issue and control the corresponding lane-level speed limit control command, adjust the speed limit value of the lane-level variable speed limit plate, and obtain the speed limit value of each lane-level variable speed limit plate;

[0060] Step 103: Calculate the compliance degree of each driver of the vehicle to be tested;

[0061] Step 104: obtaining a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition according to the compliance degree of each driver of the vehicle to be tested, the speed limit value of each lane-level variable speed limit plate, and the time-series trajectory data of the vehicle to be tested;

[0062] Step 105: Based on the Kshape clustering algorithm, cluster analysis is performed on the compliance time series rule data set of the vehicle to be tested under each speed limit condition to obtain a compliance time series data feature set of the vehicle to be tested under each speed limit condition;

[0063] Step 106: performing Hausdorff distance calculation on the time series matrices of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition to obtain the initial matrix of the time series matrices of each category;

[0064] Step 107: Based on weighted calculation, a distance matrix is ​​obtained according to the initial matrix of each category time series matrix;

[0065] Step 108: clustering the distance matrix using a Kshape model to obtain a driver style database for the vehicle to be tested;

[0066] Step 109: Based on the XGBoost model, the driving style recognition model is constructed according to the driver style type database of the vehicle to be tested and the time series trajectory data set.

[0067] Furthermore, the method of collecting a time series trajectory data set of the vehicle to be tested by using a millimeter wave radar or a laser radar includes:

[0068] Several millimeter wave radars or laser radars are arranged at a preset distance on both sides of the expressway. The preset distance is smaller than the scanning range of two millimeter wave radars or laser radars. The radar arrangement spacing needs to be smaller than the detection distance of two radars. In this embodiment, the ASR408-21 model millimeter wave radar is selected as the vehicle trajectory collection device. Its detection range is 250 meters, so the arrangement spacing is selected as 350 meters. The arrangement method is as follows: Figure 4 shown.

[0069] The millimeter wave radar or laser radar collects the time series trajectory data set of the vehicle to be tested, and realizes the collection and analysis of characteristic parameters such as the driver's time series trajectory data.

[0070] Specifically, the calculation formula for the compliance degree of each driver of the vehicle to be tested is:

[0071]

[0072]

[0073] Among them, Comprate i (t) is the compliance degree of vehicle i at time t, is the speed of vehicle i at 250 m upstream of the lane-level variable speed limit plate, VSL is the speed of vehicle i at 200 m downstream of the lane-level variable speed limit plate. i (t) is the variable speed limit value issued to vehicle i at time t. For drivers and vehicles, high, medium, and low compliance levels are defined with compliance levels of 1 and 0.8 as the boundaries. The compliance level is defined as:

[0074]

[0075] Furthermore, the compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition is obtained according to the compliance degree of each driver of the vehicle to be tested, the speed limit value of each lane-level variable speed limit plate and the time-series trajectory data of the vehicle to be tested, including:

[0076] The compliance level of each driver of the vehicle to be tested is differentiated into compliance levels to obtain high compliance, medium compliance and low compliance;

[0077] According to the high compliance, medium compliance, low compliance, the speed limit values ​​of each lane-level variable speed limit plate and the time-series trajectory data of the vehicle to be tested, a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition is obtained.

[0078] Furthermore, the Hausdorff distance calculation is performed on the time series matrices of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition to obtain the initial matrix of the time series matrices of each category, including:

[0079] Extracting the time series matrix of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition, wherein the time series matrix of each category includes: speed time series matrix, rule time series matrix, acceleration time series matrix and jerk time series matrix;

[0080] Hausdorff distance calculation is performed on the speed timing matrix, the rule timing matrix, the acceleration timing matrix and the jerk timing matrix to obtain the initial matrix of each category timing matrix.

[0081] According to the compliance level of drivers and vehicles, they are divided into three data sets, namely high compliance data set, medium compliance data set and low compliance data set. Kshape clustering algorithm is used to perform cluster analysis on the time series data of vehicles in different compliance data sets.

[0082] Among them, the characteristics of the time series trajectory data of drivers with different speed limit combinations and different compliance levels are analyzed, and specific lanes and vehicle models are selected to extract vehicle time series data, including: transverse and longitudinal position, transverse and longitudinal speed, transverse and longitudinal acceleration, jerk and other time series data. The Hausdorff distance is used to calculate the distance between different categories of time series data. The Hausdorff distance calculation principle is as follows Figure 5 shown.

[0083] Specifically, the calculation formula for obtaining the distance matrix is:

[0084] distancemat

[0085] =w ０ *distancemat_tra+w1*distancemat_speed+w2*

[0086] distancemat_acc+w3*distancemat_jerk;

[0087] Among them, distancemat_tra is the bidirectional Hausdorff distance of PositionX and PositionY, distancemat_speed is the bidirectional Hausdorff distance of VelocityX and VelocityY, distancemat_acc is the bidirectional Hausdorff distance of AccX and AccY, distancemat_jerk is the one-way Hausdorff distance of jerk, among which PositionX is the longitudinal position of the vehicle, PositionY is the lateral position of the vehicle, VelocityX is the longitudinal speed of the vehicle, VelocityY is the lateral speed of the vehicle, AccX is the longitudinal acceleration of the vehicle, AccY is the lateral acceleration of the vehicle, jerk is the longitudinal jerk value of the vehicle, among which w0, w1, w2, w3 are weighting coefficients, w0+w1+w2+w3=1.

[0088] The time series data is standardized using the TimeSeriesScalerMeanVariance method in tslearn, where the mean is 0.0 and the standard deviation is 1.0. The elbow rule is used to find the optimal number of clusters, and finally the cluster center is selected as 3 based on the elbow.

[0089] Furthermore, the distance matrix is ​​clustered using the Kshape model to obtain a driver style database for the vehicle to be tested, including:

[0090] According to the compliance degree of the driver of the vehicle to be tested, the driver is divided into aggressive high compliance, patient high compliance, cautious high compliance, aggressive medium compliance, patient medium compliance, cautious medium compliance, aggressive low compliance, patient low compliance and cautious low compliance;

[0091] A driver style type database of the vehicle to be tested is constructed according to the aggressive high compliance, patient high compliance, cautious high compliance, aggressive medium compliance, patient medium compliance, cautious medium compliance, aggressive low compliance, patient low compliance and cautious low compliance.

[0092] Specifically, the Kshape model is used to cluster the final distance matrix to obtain the driving style categories, and finally the driving style classification under different lane-level speed limit combinations is obtained. Among them, cluster1 has large driving behavior fluctuations and the highest average speed, which is defined as an aggressive driving style; cluster2 has small driving behavior fluctuations and the average speed is neither high nor low, which is defined as a patient driving style; cluster3 has small driving behavior fluctuations and the lowest average speed, which is defined as a cautious driving style.

[0093] The vehicle trajectory data is divided into 9 data sets according to the driver's driving style. The driver style type is used as the dependent variable, the trajectory feature parameter matrix is ​​used as the independent variable, and the training is based on the XGBoost model, in which the parameters in the trajectory feature parameter matrix are used as input variables and data standardization is performed. The driver style type database of the vehicle to be tested is used, and the importance ranking of the XGBoost model is used to reduce the data dimension to obtain the training data set.

[0094] The parameter selection rules of the trajectory characteristic parameter matrix are as follows: the extraction positions are 200m, 800m, 1400m, and 2000m upstream of the position where the vehicle receives the variable speed limit instruction, and 400m and 1000m downstream of the position where the vehicle receives the variable speed limit instruction. The extraction time granularity is 5 minutes, and the extraction time is 5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes, and 30 minutes before the vehicle receives the variable speed limit instruction. The characteristic parameters include lateral and longitudinal speed, lateral and longitudinal acceleration, jerk, and lateral offset value. The parameter selection rules are as follows: Figure 6 Shown

[0095] Furthermore, the XGBoost model is used to construct the driving style recognition model according to the driver style database of the vehicle to be tested and the time series trajectory data set, including:

[0096] Performing dimensionality reduction processing on the time series trajectory data set to obtain a time series trajectory data set after dimensionality reduction;

[0097] The driving style recognition model is constructed according to the reduced-dimensional time series trajectory data set and the driver style type database of the vehicle to be tested.

[0098] Specifically, based on the XGBoost model, the importance of independent variables is ranked and analyzed, and the 10 most important variables are selected as the variables finally selected by the prediction model.

[0099] Based on the 10 most important variables and driving styles, an XGBoost driving style recognition model is constructed, and finally driving style recognition based on cross-sectional data is realized.

[0100] Among them, Table 1 is a statistical table of lane-level speed limit information release in the embodiment, and Table 1 is as follows:

[0101] Table 1 Statistics of lane-level speed limit information release

[0102]

[0103]

[0104] Table 2 is a description table of variables related to trajectory characteristic parameters, as shown below:

[0105] Table 2 Description of variables related to trajectory characteristic parameters

[0106]

[0107]

[0108]

[0109] The beneficial effects of the present invention are as follows:

[0110] The method of the present invention takes into account the lane-level speed limit when classifying the driving style of the driver, and obtains the driving style distribution of the driver under the lane-level speed limit condition by using the time series trajectory data, thereby solving the problem that the traditional driving style classification uses cross-sectional traffic flow parameters and lacks the speed limit condition, resulting in poor driving style classification characteristics.

[0111] The present invention uses millimeter wave radar and laser radar deployed on the roadside as vehicle trajectory sensing means to obtain vehicle trajectory data, which solves the problem of limited data volume and insufficient driving style sample volume caused by the traditional driving style classification relying on the vehicle data collection method and the driving simulation collection method. This method provides a theoretical basis and technical support for the large-scale promotion of personalized management of drivers under lane-level variable speed limits in the future. At the same time, Kshape clustering has scaling and translation invariance, which can make driving behavior appear at any time, regardless of translation or scaling, and can be successfully identified, overcoming the defect that the traditional driving style recognition method is easily affected by the data form. The present invention has the characteristics of replicability and strong robustness.

[0112] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0113] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A driving style classification method based on time series trajectory data under speed limit conditions, characterized in that: include: Input the time series trajectory data set of the vehicle to be tested into the constructed driving style recognition model to obtain the driving style of the driver of the vehicle to be tested; The method for constructing the driving style recognition model is as follows: Millimeter-wave radars or laser radars are deployed on the expressway at preset intervals, and the millimeter-wave radars or laser radars are used to collect time-series trajectory datasets of training vehicles; Lane-level variable speed limit plates are arranged at preset intervals on the expressway, and based on a preset control strategy, a preset variable information system is used to issue corresponding lane-level speed limit control instructions, adjust the speed limit values ​​of the lane-level variable speed limit plates, and obtain the speed limit values ​​of each lane-level variable speed limit plate; Calculate the compliance degree of each driver of the tested vehicle; Obtaining a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition according to the compliance degree of each driver of the vehicle to be tested, the speed limit value of each lane-level variable speed limit plate, and the time-series trajectory data of the vehicle to be tested; Based on the Kshape clustering algorithm, cluster analysis is performed on the compliance time series rule data set of the vehicles to be tested under the various speed limit conditions to obtain the compliance time series data feature set of the vehicles to be tested under the various speed limit conditions; Performing Hausdorff distance calculation on the time series matrices of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition to obtain an initial matrix of the time series matrices of each category; Based on weighted calculation, a distance matrix is ​​obtained according to the initial matrix of the time series matrix of each category; Clustering the distance matrix using a Kshape model to obtain a driver style database for the vehicle to be tested; Based on the XGBoost model, the driving style recognition model is constructed according to the driver style type database of the vehicle to be tested and the time series trajectory data set.

2. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The method of collecting a time series trajectory data set of the vehicle to be tested by using a millimeter wave radar or a laser radar includes: A plurality of millimeter wave radars or laser radars are arranged at a preset distance on both sides of the expressway, wherein the preset distance is smaller than the scanning range of the two millimeter wave radars or laser radars; The millimeter wave radar or laser radar is used to collect a time series trajectory data set of the vehicle to be tested.

3. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The calculation formula of the compliance degree of each driver of the vehicle to be tested is: Among them, Comprate i (t) is the compliance degree of vehicle i at time t, is the speed of vehicle i at 250 m upstream of the lane-level variable speed limit plate, VSL is the speed of vehicle i at 200 m downstream of the lane-level variable speed limit plate. i (t) is the variable speed limit issued to vehicle i at time t.

4. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The step of obtaining a compliance time series trajectory data set of the vehicle to be tested under each speed limit condition according to the compliance degree of each driver of the vehicle to be tested, the speed limit value of each lane-level variable speed limit plate, and the time series trajectory data of the vehicle to be tested includes: The compliance level of each driver of the vehicle to be tested is differentiated into compliance levels to obtain high compliance, medium compliance and low compliance; According to the high compliance, medium compliance, low compliance, the speed limit values ​​of each lane-level variable speed limit plate and the time-series trajectory data of the vehicle to be tested, a compliance time-series trajectory data set of the vehicle to be tested under each speed limit condition is obtained.

5. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The Hausdorff distance calculation is performed on the time series matrix of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition to obtain the initial matrix of the time series matrix of each category, including: Extracting the time series matrix of each category in the time series data feature set of the compliance degree of the vehicle to be tested under each speed limit condition, wherein the time series matrix of each category includes: speed time series matrix, rule time series matrix, acceleration time series matrix and jerk time series matrix; Hausdorff distance calculation is performed on the speed timing matrix, the rule timing matrix, the acceleration timing matrix and the jerk timing matrix to obtain the initial matrix of each category timing matrix.

6. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The calculation formula for obtaining the distance matrix is: distancemat =w ０ *distancemat_tra+w1*distancemat_speed+w2* distancemat_acc+w3*distancemat_jerk; Among them, distancemat_tra is the bidirectional Hausdorff distance of PositionX and PositionY, distancemat_speed is the bidirectional Hausdorff distance of VelocityX and VelocityY, distancemat_acc is the bidirectional Hausdorff distance of AccX and AccY, distancemat_jerk is the one-way Hausdorff distance of jerk, among which PositionX is the longitudinal position of the vehicle, PositionY is the lateral position of the vehicle, VelocityX is the longitudinal speed of the vehicle, VelocityY is the lateral speed of the vehicle, AccX is the longitudinal acceleration of the vehicle, AccY is the lateral acceleration of the vehicle, jerk is the longitudinal jerk value of the vehicle, among which w0, w1, w2, w3 are weighting coefficients, w0+w1+w2+w3=1.

7. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The method of clustering the distance matrix using the Kshape model to obtain a driver style database for the vehicle to be tested includes: According to the compliance degree of the driver of the vehicle to be tested, the driver is divided into aggressive high compliance, patient high compliance, cautious high compliance, aggressive medium compliance, patient medium compliance, cautious medium compliance, aggressive low compliance, patient low compliance and cautious low compliance; A driver style type database of the vehicle to be tested is constructed according to the aggressive high compliance, patient high compliance, cautious high compliance, aggressive medium compliance, patient medium compliance, cautious medium compliance, aggressive low compliance, patient low compliance and cautious low compliance.

8. The driving style classification method based on time series trajectory data under speed limit conditions according to claim 1, characterized in that: The XGBoost model is based on which the driving style recognition model is constructed according to the driver style type database of the vehicle to be tested and the time series trajectory data set, including: Performing dimensionality reduction processing on the time series trajectory data set to obtain a time series trajectory data set after dimensionality reduction; The driving style recognition model is constructed according to the reduced-dimensional time series trajectory data set and the driver style type database of the vehicle to be tested.

Images