CN110254435B - Driving style recognition method - Google Patents
Driving style recognition method Download PDFInfo
- Publication number
- CN110254435B CN110254435B CN201910580087.7A CN201910580087A CN110254435B CN 110254435 B CN110254435 B CN 110254435B CN 201910580087 A CN201910580087 A CN 201910580087A CN 110254435 B CN110254435 B CN 110254435B
- Authority
- CN
- China
- Prior art keywords
- driving style
- driving
- data
- road type
- track data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
- B60W40/09—Driving style or behaviour
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2720/00—Output or target parameters relating to overall vehicle dynamics
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Mathematical Physics (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a driving style identification method, which comprises the following steps: step S1: collecting driving track data by using a sensor; step S2: obtaining relative overspeed behavior by combining geographic information data based on the driving track data; step S3: calculating to obtain a driving style parameter based on the relative overspeed behavior; step S4: and identifying the driving style by using a clustering algorithm based on the driving style parameters. Compared with the prior art, the data accuracy requirement of driving style recognition is relaxed, and the data acquisition cost of driving style recognition is favorably reduced based on low-frequency data; the application of the low-frequency driving track data in the aspect of driving style exploration is expanded; the road traffic safety can be improved.
Description
Technical Field
The invention relates to the field of driving style recognition, in particular to a driving style recognition method.
Background
The driving style is an overall evaluation index representing the inherent driving mode of the driver. The driving style research mainly divides drivers from the aspects of energy-saving driving, traffic safety and the like. Existing research shows that the style of a driver is closely related to traffic safety: the driver with the aggressive driving style has the bad driving behaviors of frequent lane changing, rapid acceleration and deceleration, close-distance following and the like, and the accident occurrence probability is easy to increase. Through the research of the driving style, the bad driving style of the daily driving of the driver is detected or fed back, the monitoring and education of the driving behavior of the driver can be realized, and the driving safety is improved by taking corresponding measures in an auxiliary mode.
The data for the driving style research mainly comprises subjective questionnaire survey and objective driving data, and the driving style is evaluated from microscopic angles such as acceleration and jerk by taking high-precision driving behavior data as a main part, such as high-frequency GPS (global positioning system) driving track data with the collection frequency of 1Hz and 10 Hz. However, the vehicle track data collected by the vehicle service providing enterprises such as the current time-sharing rental mainly collects low-frequency vehicle track data (the GPS collection interval is 30 to 120s) with the aim of vehicle positioning. How to construct driving style parameters and develop driving style recognition based on the characteristics of low-frequency driving track data still lacks a reasonable scheme.
Disclosure of Invention
The present invention is directed to a driving style recognition method for overcoming the above-mentioned drawbacks of the prior art.
The purpose of the invention can be realized by the following technical scheme:
a driving style identification method, the method comprising the steps of:
step S1: collecting driving track data by using a sensor;
step S2: obtaining relative overspeed behavior by combining geographic information data based on the driving track data;
step S3: calculating to obtain a driving style parameter based on the relative overspeed behavior;
step S4: and identifying the driving style by using a clustering algorithm based on the driving style parameters.
The driving track data are GPS driving track data, and the GPS driving track data are low-frequency driving track data with the collection interval of 30-60 s.
The step S2 includes:
step S21: matching the driving track data with the geographic information data to obtain the road type of each track point;
step S22: and extracting the behavior of 99 quantiles of the road type running speed with the running speed higher than each track point to obtain the relative overspeed behavior.
The step S3 includes:
step S31: calculating to obtain a relative overspeed time proportion based on the relative overspeed behavior;
step S32: calculating to obtain an average relative overspeed time proportion based on the relative overspeed time proportion;
step S33: calculating to obtain a variation coefficient based on the average relative overspeed time proportion;
step S34: and calculating to obtain the driving style parameters based on the variation coefficient and the average relative overspeed time proportion.
For a trip, the relative overspeed time ratio on the road type i of the driver is compared with the overspeed time ratio ti_speeding%Comprises the following steps:
where i is the road type (i ═ 1,2, …, m), ti1For a running duration of 99 quantiles with a running speed higher than the i-road type running speed, ti2For a travel time period with an operating speed higher than the minimum speed for the level one service for that type of road.
Average relative overspeed time ratio x of i road type in each trip of j driverji1Comprises the following steps:
where p is the number of travel road types i, ti_speeding%_lIs the relative overspeed time proportion of the ith (i ═ 1,2, …, p) trip road type i.
Variation coefficient x of i road type in each trip of j driverji2Comprises the following steps:
j driver's driving style parameter xjComprises the following steps:
xj=(xj11,xj12,…,xji1,xji2,…,xjm1,xjm2)
the clustering algorithm is a K-Means clustering algorithm.
Compared with the prior art, the invention has the following advantages:
(1) the data accuracy requirement of driving style recognition is relaxed, and the data acquisition cost of driving style recognition is favorably reduced based on low-frequency data.
(2) The application of the low-frequency driving track data in the aspect of driving style exploration is expanded.
(3) For identified drivers with bad styles (such as driving style aggressive groups), risk management measures such as safety education projects or UBI vehicle insurance schemes can be formulated in a targeted manner, so that the road traffic safety can be improved, and the enterprise operation risk can be reduced.
(4) The calculation is carried out only by the vehicle data of the vehicle, namely the vehicle speed, so that the use amount of the sensors is reduced, and the data flow cost is reduced.
(5) The driving style parameters can accurately represent the driving style.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of a K-Means clustering algorithm;
fig. 3 shows example driving style clustering results.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Examples
The invention provides a driving style identification method based on low-frequency driving track data, which can identify groups with different driving styles without depending on high-precision data, find potential high-risk drivers and provide a basis for risk management measures such as a safety education project or a UBI (universal UBI vehicle insurance) scheme and the like.
The invention provides a driving style identification method based on low-frequency driving track data, which comprises the following steps of:
step S1: collecting driving track data by using a sensor;
step S2: obtaining relative overspeed behavior by combining geographic information data based on the driving track data;
step S3: calculating to obtain a driving style parameter based on the relative overspeed behavior;
step S4: and identifying the driving style by using a clustering algorithm based on the driving style parameters.
Further, the acquired trajectory data in step S1 is acquired by a vehicle device sensor, and the acquired trajectory data is low-frequency trajectory data, and the interval between the acquired trajectory data by the GPS is 30 to 60S.
Further, in step S2, the step of obtaining the relative overspeed behavior by combining the geographic information data specifically includes:
(2.1) matching the track data with the geographic information layer data: matching of track data and geographic information layer data is achieved based on OpenStreetMap (OSM) and Open Source Routing Engine (OSRM) technologies, and road and traffic management information of each track point is obtained, wherein the road and traffic management information includes but is not limited to road names, road types i (i is 1,2, …, m) and speed limits;
(2.2) obtaining relative overspeed behavior: based on the characteristics of low-frequency driving track data, relative overspeed behavior, namely the behavior that the running speed of a driver on a road type i is higher than 99 quantiles of the running speed of the road type, and the relative overspeed behavior is identified based on the road type i where each track point is located and the running speed.
Wherein, quantile means as follows: a group of data is sorted from small to large, corresponding cumulative percentiles are calculated, the value of the data corresponding to a certain percentile is called the quantile of the percentile, namely the value at the p% position is called the p quantile.
Further, in step S3, the driving style parameter is calculated based on the relative overspeed behavior, specifically:
(3.1) defining a relative overspeed time ratio based on the relative overspeed behavior: for a trip, the relative overspeed time ratio t of the driver on the road type ii_speeding%Comprises the following steps:
wherein, ti1For a 99-quantile driving duration, t, with a driving speed in the trip higher than the i-road type driving speedi2The travel time length of the travel is that the travel speed in the trip is higher than the typical speed (such as the minimum speed of the first-level service level) in the unobstructed traffic state;
(3.2) based on the relative overspeed time ratio, calculating the following driving style parameters:
average relative overspeed time ratio x of i road type in each trip of j driverji1Comprises the following steps:
where p is the number of travel road types i, ti_speeding%_lIs the relative overspeed time proportion of the ith (i ═ 1,2, …, p) trip road type i.
Variation coefficient x of i road type relative overspeed time proportion in each trip of j driverji2Comprises the following steps:
coefficient of variation xji2This behavior stability is characterized in that for a particular driver, relative speeding is a habitual or sporadic behavior.
The driving style parameter of the driver j is xj=(xj11,xj12,…,xji1,xji2,…,xjm1,xjm2)。
Further, the clustering algorithm in the step S4 is a K-Means clustering algorithm, and the driving style is identified and classified based on the K-Means clustering algorithm, specifically:
(4.1) evaluating the driving style based on a K-Means clustering algorithm, which comprises the following specific steps:
(4.1a) obtaining a set of drivers X ═ X1,x2,…,xj,…,xn)TAnd normalizing the data to obtain a processed matrix X '═ X'1,x'2,…,x'j,…,x'n)T;
(4.1b) initializing the cluster center number k ═ 2;
(4.1c) adopting K-Means clustering algorithm to pair x'1,x'2,…,x'j,…,x'nClassifying the driving style;
(4.1d) calculating the average contour coefficient s under k clusteringkThe calculation method comprises the following steps:
wherein a is the average value of the intra-class distances, and b is the average value of the distances to the nearest neighboring class;
(4.1e) comparison skAnd sk-1If s isk>sk-1If k is equal to k +1, repeat the stepsSteps (4.1c), (4.1d) and (4.1 e); if sk<sk-1The cycle is stopped;
(4.1f) outputting the optimal clustering number k-1 and the driving style classification result under the clustering number k-1;
and (4.2) identifying different driving styles based on the driving style classification result.
The following description is made in connection with a specific alternative embodiment:
GPS data of a vehicle is collected through a vehicle device sensor, the GPS collection interval is 30s, 235 drivers are obtained, and 19348km of urban expressway driving data are accumulated.
(II) extracting relative overspeed behaviors by combining geographic information data, and specifically comprising the following steps:
(1) matching the track data with the geographic information layer data: matching of track data and geographic information layer data is achieved based on OpenStreetMap (OSM) and Open Source Routing Engine (OSRM) technologies, and road and traffic management information of each track point is obtained, wherein the road and traffic management information comprises a road name, a road type i (i is 1,2, …, m) and a speed limit, the road type i and the speed limit are only included in the example, and the urban expressway i is 2;
(2) a relative overspeed behavior is obtained: based on the characteristics of low-frequency driving track data, relative overspeed behavior, namely the behavior that the running speed of a driver on a road type i is higher than the 99 quantile of the running speed of the road type, is identified based on the road type i where each track point is located and the running speed, and the 99 quantile v of the running speed21=90km/h。
(III) providing and calculating driving style parameters, which comprises the following specific steps:
(1) based on the relative overspeed behavior, the relative overspeed time ratio is defined as: proportion t of relative speeding behaviour of driver on road type i2_speeding%Comprises the following steps: 99 quantile v with running speed higher than running speed of urban expressway21As long as 90km/h21With the operating speed being higher than the minimum speed v of the first service level2273km/h driving time t22The ratio of (a) to (b), namely:
(2) based on the relative overspeed time ratio, the following driving style parameters are proposed and calculated:
j average relative overspeed time ratio x of driver on city expresswayj21Characterizing the driver's tendency to overspeed, expressed as:
j coefficient of variation x of driver on urban expresswayj22Comprises the following steps:
let the driving style parameter of driver j be xj=(xj21,xj22)。
(IV) driving style evaluation is carried out based on a K-Means clustering algorithm, and as shown in figure 2, the specific steps are as follows:
(1) obtaining a set of drivers X ═ X (X)1,x2,…,xj,…,x235)TAnd normalizing the data to obtain a processed matrix X '═ X'1,x'2,…,x'j,…,x'235)T;
(2) Initializing a clustering center number k to be 2;
(3) adopting K-Means clustering algorithm to pair x'1,x'2,…,x'j,…,x'235Classifying the driving style;
(4) calculating the average contour coefficient s under k clusteringkThe calculation method comprises the following steps:
wherein a is the average value of the intra-class distances, and b is the average value of the distances to the nearest neighboring class;
(5) comparison skAnd sk-1If s isk>sk-1If so, making k equal to k +1, and repeating the steps (3), (4) and (5); if sk<sk-1The cycle is stopped;
(6) outputting the optimal clustering number k-1 and the driving style classification result under the clustering number k-1, wherein the optimal clustering number is 3 and the average contour coefficient under the optimal clustering number is 0.43;
(8) based on the driving style classification result, identifying different driving styles: in this embodiment, the driving styles of the driver are divided into three types of cautious, mild and aggressive according to the variation coefficients of the relative overspeed time proportion and the relative overspeed time proportion of each trip, as shown in fig. 3, each style respectively accounts for 54.04%, 36.60% and 9.36% of the sample driver, and the characteristics are as follows:
1) prudent: the relative overspeed time proportion is the lowest, the variation coefficient of each trip relative to the overspeed time proportion is higher, namely the overspeed tendency is low, and the relative overspeed is accidental behavior;
2) and (2) mildness: the relative overspeed time proportion is lower, the variation coefficient of each trip relative to the overspeed time proportion is higher, namely the overspeed tendency is lower, and the relative overspeed is accidental behavior;
3) and (3) radical excitation: the relative overspeed time proportion is the highest, the variation coefficient of the relative overspeed time proportion of each trip is lower, namely the overspeed tendency is higher, and the relative overspeed appears more stably in each trip.
Through the steps, the driving style recognition based on the low-frequency data is completed, compared with the prior art that the driving style evaluation depends on high-precision track data to extract microscopic driving behavior characteristics, the steps relax the data precision requirement of the driving style evaluation, provide a reasonable scheme for driving style parameter and driving style recognition based on the low-frequency data, facilitate the reduction of the data acquisition cost of the driving style recognition, and expand the application of the low-frequency driving track data in the aspect of driving style exploration. For identified drivers with bad styles (such as driving style aggressive groups), risk management measures such as safety education projects or UBI vehicle insurance schemes can be formulated in a targeted manner, so that the road traffic safety can be improved, and the enterprise operation risk can be reduced.
Claims (5)
1. A driving style recognition method, characterized by comprising the steps of:
step S1: collecting driving track data by using a sensor;
step S2: obtaining relative overspeed behavior by combining geographic information data based on the driving track data;
step S3: calculating to obtain a driving style parameter based on the relative overspeed behavior;
step S4: identifying the driving style by using a clustering algorithm based on the driving style parameters;
the running track data is GPS running track data, and the GPS running track data is low-frequency running track data with an acquisition interval of 30-60 s;
the step S2 includes:
step S21: matching the driving track data with the geographic information data to obtain the road type of each track point;
step S22: extracting the behavior of 99 quantiles of the road type running speed with the running speed higher than each track point to obtain relative overspeed behavior;
the step S3 includes:
step S31: calculating to obtain a relative overspeed time proportion based on the relative overspeed behavior;
step S32: calculating to obtain an average relative overspeed time proportion based on the relative overspeed time proportion;
step S33: calculating to obtain a variation coefficient based on the average relative overspeed time proportion;
step S34: calculating to obtain driving style parameters based on the variation coefficient and the average relative overspeed time proportion;
for a trip, the relative overspeed time ratio t of the driver on the road type ii_speeding%Comprises the following steps:
where i is the road type, ti1For a running duration of 99 quantiles with a running speed higher than the i-road type running speed, ti2A travel time period in which the running speed is higher than the set speed for the road type is set.
2. The driving style recognition method of claim 1, wherein the average relative speeding time ratio x of i road type for j drivers in each tripji1Comprises the following steps:
where p is the number of travel road types i, ti_speeding%_lIs the relative overspeed time proportion of the ith trip i road type.
4. a driving style recognition method according to claim 3, characterized in that the driving style parameter x of j driversjComprises the following steps:
xj=(xj11,xj12,…,xji1,xji2,…,xjm1,xjm2)。
5. the driving style recognition method of claim 1, wherein the clustering algorithm is a K-Means clustering algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910580087.7A CN110254435B (en) | 2019-06-28 | 2019-06-28 | Driving style recognition method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910580087.7A CN110254435B (en) | 2019-06-28 | 2019-06-28 | Driving style recognition method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110254435A CN110254435A (en) | 2019-09-20 |
CN110254435B true CN110254435B (en) | 2021-06-04 |
Family
ID=67923212
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910580087.7A Active CN110254435B (en) | 2019-06-28 | 2019-06-28 | Driving style recognition method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110254435B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111259767B (en) * | 2020-01-13 | 2023-04-18 | 厦门大学 | Traffic illegal behavior identification method and system based on traffic data and street view data |
CN112721949B (en) * | 2021-01-12 | 2022-07-12 | 重庆大学 | Method for evaluating longitudinal driving personification degree of automatic driving vehicle |
CN113096379A (en) * | 2021-03-03 | 2021-07-09 | 东南大学 | Driving style identification method based on traffic conflict |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100019880A1 (en) * | 2008-07-24 | 2010-01-28 | Gm Global Technology Operations, Inc. | Adaptive vehicle control system with driving style recognition based on traffic sensing |
CN103914985B (en) * | 2014-04-25 | 2015-10-28 | 大连理工大学 | A kind of hybrid power passenger car following speed of a motor vehicle trajectory predictions method |
CN107153916A (en) * | 2017-04-30 | 2017-09-12 | 安徽中科美络信息技术有限公司 | A kind of driving behavior evaluation method clustered based on FCM with BP neural network |
CN108491859A (en) * | 2018-02-11 | 2018-09-04 | 郭静秋 | The recognition methods of driving behavior heterogeneity feature based on automatic coding machine |
CN109034181A (en) * | 2018-06-05 | 2018-12-18 | 广州杰赛科技股份有限公司 | The classification method and device of motion track, equipment, storage medium |
CN109523390A (en) * | 2018-09-28 | 2019-03-26 | 深圳市麦谷科技有限公司 | A method of air control model is established according to running car rule |
CN109559507A (en) * | 2018-11-16 | 2019-04-02 | 西南交通大学 | A kind of about vehicle excess speed event recognition methods of the net based on history GPS track data |
-
2019
- 2019-06-28 CN CN201910580087.7A patent/CN110254435B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN110254435A (en) | 2019-09-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110254435B (en) | Driving style recognition method | |
CN107492251B (en) | Driver identity recognition and driving state monitoring method based on machine learning and deep learning | |
EP2710572B1 (en) | Vehicle data analysis method and vehicle data analysis system | |
CN101965601B (en) | Driving support device and driving support method | |
Liu et al. | Customizing driving cycles to support vehicle purchase and use decisions: Fuel economy estimation for alternative fuel vehicle users | |
HUE029992T2 (en) | Driver profiling system and method | |
WO2006130146A1 (en) | Motor vehicle operating data collection and analysis | |
CN106529599A (en) | Driver ecologic driving behavior evaluation method facing event | |
CN110395245B (en) | Hybrid electric vehicle energy management system based on fixed route driving information | |
CN114299742B (en) | Speed limit information dynamic identification and update recommendation method for expressway | |
CN116628455B (en) | Urban traffic carbon emission monitoring and decision support method and system | |
CN114435138A (en) | Vehicle energy consumption prediction method and device, vehicle and storage medium | |
JP3981067B2 (en) | Travel speed data classification device, travel speed pattern estimation device, and hybrid vehicle drive control device | |
CN106940930B (en) | Motorway journeys time prediction system and prediction technique | |
CN113222385A (en) | Method for constructing and evaluating driving condition of electric automobile | |
Li et al. | Clustering pavement roughness based on the impacts on vehicle emissions and public health | |
CN113095387B (en) | Road risk identification method based on networking vehicle-mounted ADAS | |
CN112927511A (en) | Vehicle early warning method based on driver age and gender identification | |
CN110796374A (en) | Construction method of urban comprehensive driving working condition with typical lane changing and turning working conditions | |
CN116153082B (en) | Expressway road condition acquisition, analysis and processing system based on machine vision | |
CN110097074B (en) | Vehicle track compression method based on sequence similarity | |
CN115774942A (en) | Driving style identification model modeling and statistical method based on Internet of vehicles real vehicle data and SVM | |
CN114119256A (en) | UBI dangerous chemical vehicle driving behavior acquisition and analysis system and premium discount method | |
CN113147781A (en) | Intelligent driving information display system for automobile | |
CN113505955A (en) | User driving behavior scoring method based on TSP system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |