CN106448158B - Traffic accidents reason analysis method based on correspondence analysis model - Google Patents
Traffic accidents reason analysis method based on correspondence analysis model Download PDFInfo
- Publication number
- CN106448158B CN106448158B CN201610807054.8A CN201610807054A CN106448158B CN 106448158 B CN106448158 B CN 106448158B CN 201610807054 A CN201610807054 A CN 201610807054A CN 106448158 B CN106448158 B CN 106448158B
- Authority
- CN
- China
- Prior art keywords
- accident
- type
- driver
- regulations
- traffic
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 90
- 206010039203 Road traffic accident Diseases 0.000 title claims abstract description 85
- 230000000875 corresponding Effects 0.000 claims description 30
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000002265 prevention Effects 0.000 abstract description 3
- 239000000203 mixture Substances 0.000 description 7
- 238000000034 method Methods 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000005755 formation reaction Methods 0.000 description 3
- 230000037408 Distribution ratio Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007621 cluster analysis Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000003694 hair properties Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000000630 rising Effects 0.000 description 1
- 239000002965 rope Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Abstract
The present invention provides a kind of traffic accidents reason analysis method based on correspondence analysis model, comprising the following steps: input traffic accident data and break in traffic rules and regulations data, the two pass through driver identification information association;Correspondence analysis is carried out to driver's traffic accident type and its break in traffic rules and regulations type;To driver's traffic accident type and driver's driving age, gender, vehicle system Multiple correspondence analysis.Traffic accidents reason analysis method of this kind based on correspondence analysis model analyzes driver driving behavior and its influence factor by driver's history record violating the regulations, driving age, gender, vehicle brand and the type that traffic accident occurs;To realize the specific aim prevention of traffic accident, the safety of road is promoted.
Description
Technical field
The present invention relates to a kind of traffic accidents reason analysis methods based on correspondence analysis model.
Background technique
With gradually rising for domestic automobile ownership, traffic accident problem is also more prominent;Various types of traffic things
Therefore it emerges one after another;Various causes of accident are certainly existed in the behind of these accident patterns, it is based on driver, be based on
Road environment etc.;Therefore need by a large amount of data carry out investigative analysis, find out cause various accident pattern factors it
Between difference and connection;To targetedly prevent traffic accident, the safety of road is promoted.
Lack careful classification for the method for traffic accidents reason analysis at present, has focused largely on traffic accident in research
Number and its influence relationship;And the analysis of the accident pattern of the driving behavior analysis of driver, driver is lacked and is ground accordingly
Study carefully;It breaks rules and regulations to be the fuse cord for leading to finally to occur traffic accident in reality, how to break rules and regulations to record from the history of driver to explore
Driver's which kind of act of violating regulations in driving procedure will lead to what type of traffic accident;Secondly driver driving behavior again and
The vehicle that driver's driving age, gender drive has some relations;The solution of the above problem has the analysis of Traffic Accidents Reasons Analyzed positive
Effect.
Summary of the invention
The object of the present invention is to provide a kind of traffic accidents reason analysis methods based on correspondence analysis model, for difference
Traffic accident type will affect and the origin cause of formation of accident pattern occurs carry out various dimensions point using the method for correspondence analysis and cluster
Analysis is explored, and the above-mentioned problems in the prior art is solved.
The technical solution of the invention is as follows:
A kind of traffic accidents reason analysis method based on correspondence analysis model, includes the following steps,
S1, input break in traffic rules and regulations data and traffic accident data, and break in traffic rules and regulations data and traffic accident data are passed through
Driver identification information is associated;
S2, traffic accident type and break in traffic rules and regulations type correspondence analysis, first progress exploratory analysis, use pivot
Figure counts violation data, and the sample data violating the regulations of subsequent analysis is filtered out by threshold method;It is right under every kind of type of violation to find out
The accident pattern of the accident crowd and its generation that answer count accounting of the every kind of accident pattern in all accident patterns;Secondly, choosing
Representational type of fault and its accident pattern is taken to carry out simple correspondence analysis;
S3, traffic accident type and driver's driving age, gender and the Multiple Correspondence Analysis for driving vehicle;It is mentioned from casualty data
Take accident responsibility side's identity information and its corresponding accident pattern, four class of vehicle brand styles, driver's driving age and gender
Information;T class accident pattern, y class are driven type of vehicle, j class driving age classification, gender to correspond, generate x=(2*t* in total
Y*j) class combines;The corresponding number of every a kind of combination is counted, it is more using being obtained in statistics to the method for Multiple Correspondence Analysis
Weight corresponding relationship.
Further, in step S1, the feature vector of traffic accident data is CS=[C, K, V], and wherein C is accident
Type;K is personal information, including driver's gender, driving age, accident responsibility ownership;V is vehicle brand information;Break in traffic rules and regulations data
Feature vector be WS=[A, H], wherein A is personnel identity information violating the regulations, and H be the content violating the regulations for the personnel that break rules and regulations.
Further, in the step S2, exploratory analysis is carried out specifically:
S21, according to traffic accident type, accident driver is classified;
S22, according to accident driver information, the record violating the regulations of screening accidents happened driver, statistics from break in traffic rules and regulations tables of data
Each all kinds of numbers violating the regulations of accident pattern driver;
S23, according to all kinds of quantity violating the regulations after statistics, retain the type of violation that number violating the regulations is more than or equal to p times, wherein p
For screening sample threshold value of breaking rules and regulations;
S24, the corresponding various break in traffic rules and regulations numbers of each type traffic accident of calculating are total with each type break in traffic rules and regulations
The ratio of number;
S25, using the ratio of driver's break in traffic rules and regulations as ordinate, the type of driver's break in traffic rules and regulations is abscissa, will be all types of
The ratio for all kinds of breaks in traffic rules and regulations that the driver of traffic accident is occurred is drawn in this coordinate system, and all kinds of accident pattern drivers are observed
All kinds of ratios violating the regulations, desk study accident and relationship violating the regulations.
Further, the step S24 is specifically, calculation formula is as follows:
Traffic accident A={ a1, a2... ai... am, wherein aiFor traffic accident type, m is accident pattern order;
Accident number K={ k1, k2... k3..., km, wherein kiIt is a for traffic accident type occursiNumber;
Type of violation B={ b1, b2... bi..., bn, biFor break in traffic rules and regulations type, n is type of violation order:
Number violating the regulationskmbnIt is a for driver's traffic accident typem, occur violating the regulations
Type is bnNumber;
Accident pattern order is the total number of persons of the driver of m
Driver breaks rules and regulations total degree
Driver's type of violation scaling matrices
Further, it in the step S2, chooses representational type of fault and its accident pattern simply correspond to
Analysis, the standard of selection are as follows:
Wherein, δnFor distribution proportion and total accident pattern of the driver in every a kind of type of violation of various accident patterns
The difference that 2 times of number distribution proportion;λ is type of fault outlier threshold corresponding with classification violating the regulations;μ is every a kind of type of violation number
Threshold value.
The practical significance of selection standard is the distribution ratio for finding out the drivers of various accident patterns in every a kind of type of violation
The difference of example and total accident pattern number distribution proportion show that type of violation and accident pattern number distribution proportion are abnormal with this
Value;
Further, driver's driving age, gender, driving type of vehicle are more with generation traffic accident type in the step S3
First correspondence analysis the following steps are included:
S31, according to occur traffic accident classification of type accident responsibility side driver;
S32, simple statistics are carried out to the driving age of these drivers, obtains driver and change of the accident quantity on driver's driving age occurs
Change trend;
The driving age of accident responsibility side driver is carried out segmentation classification by S33, the variation tendency obtained according to step S32;
S34, according to vehicle brand type by vehicle classification;
Driving age of each driver under S35, the every class accident pattern of statistics, gender, driving type of vehicle, merge similar
Statistics is generated corresponding number under the combination of x=(2*t*y*j) class and every a kind of combination in total;
S36, the result that step S35 is obtained is obtained into multiple corresponding relationship using the method for correspondence analysis, in two-dimensional space
With the relationship between the reflection variable of the formal intuition of distance between points in coordinate.
The beneficial effects of the present invention are: traffic accidents reason analysis method of this kind based on correspondence analysis model, using pair
It answers analysis method to causing the different types of accident origin cause of formation to carry out various dimensions exploration, passes through driver's history record, driving age, property violating the regulations
Not, the type of vehicle brand and generation traffic accident, analyzes driver driving behavior and its influence factor, it is multiple right to obtain
It should be related to;To realize the specific aim prevention of traffic accident, the safety of road is promoted.
Detailed description of the invention
Fig. 1 is the flow diagram of traffic accidents reason analysis method of the embodiment of the present invention based on correspondence analysis model.
Fig. 2 is the schematic diagram of accident pattern obtained in embodiment and its corresponding type of violation.
Fig. 3 is accident pattern and its simply corresponding scatter plot of type of violation in embodiment.
Fig. 4 is the classification point joint figure of Multiple Correspondence Analysis in embodiment.
Specific embodiment
The preferred embodiment that the invention will now be described in detail with reference to the accompanying drawings.
Embodiment
Corresponding analysis method in embodiment Statistics Application carries out various dimensions spy to causing the different types of accident origin cause of formation
Rope.Embodiment explores driver's which kind of act of violating regulations in driving procedure will lead to which kind of type by driver's history record violating the regulations
Traffic accident;Secondly analysis obtains the relationship for the vehicle that the driving behavior of driver and driver's driving age, gender drive.
The traffic accidents reason analysis method based on correspondence analysis model of embodiment, comprising the following steps: input traffic
Casualty data and break in traffic rules and regulations data, the two pass through driver identification information association;Driver's traffic accident type and its traffic are disobeyed
Chapter type carries out correspondence analysis;To driver's traffic accident type and driver's driving age, gender, vehicle system Multiple correspondence analysis.
The traffic accidents reason analysis method based on correspondence analysis model of embodiment, by driver's history record violating the regulations,
Driving age, gender, vehicle brand and the type that traffic accident occurs, analyze driver driving behavior and its influence factor;From
And realize the specific aim prevention of traffic accident, promote the safety of road.
Break in traffic rules and regulations data include: the content violating the regulations of personnel identity information violating the regulations, personnel violating the regulations.
Traffic accident data include: accident occurrence type, driver's gender, attribution of liability, driving age, vehicle brand information.
Traffic accidents reason analysis method of this kind based on correspondence analysis model, such as Fig. 1 include the following steps,
S1, input break in traffic rules and regulations data and traffic accident data, and break in traffic rules and regulations data and traffic accident data are passed through
Driver identification information is associated;The feature vector of traffic accident data is CS=[C, K, V], and wherein C is accident occurrence type;
K is personal information, including driver's gender, driving age, accident responsibility ownership;V is vehicle brand information;The feature of break in traffic rules and regulations data
Vector is WS=[A, H], and wherein A is personnel identity information violating the regulations, and H is the content violating the regulations of personnel violating the regulations.
S2, traffic accident type and break in traffic rules and regulations type correspondence analysis, first progress exploratory analysis, use pivot
Figure counts violation data, and the sample data violating the regulations of subsequent analysis is filtered out by threshold method;It is right under every kind of type of violation to find out
The accident pattern of the accident crowd and its generation that answer count accounting of the every kind of accident pattern in all accident patterns;Secondly, choosing
Representational type of fault and its accident pattern is taken to carry out simple correspondence analysis.
In step S2, exploratory analysis is carried out specifically:
S21, according to traffic accident type, accident driver is classified;
S22, according to accident driver information, the record violating the regulations of screening accidents happened driver, statistics from break in traffic rules and regulations tables of data
Each all kinds of numbers violating the regulations of accident pattern driver;
S23, according to all kinds of quantity violating the regulations after statistics, retain the type of violation that number violating the regulations is more than or equal to p times, wherein p
For screening sample threshold value of breaking rules and regulations;
S24, the corresponding various break in traffic rules and regulations numbers of each type traffic accident of calculating are total with each type break in traffic rules and regulations
The ratio of number, calculation formula are as follows:
Traffic accident A={ a1, a2... ai... am, wherein aiFor traffic accident type, m is accident pattern order;
Accident number K={ k1, k2... k3..., km, wherein kiIt is a for traffic accident type occursiNumber;
Type of violation B={ b1, b2... bi..., bn, biFor break in traffic rules and regulations type, n is type of violation order;
Number violating the regulationskmbnIt is a for driver's traffic accident typem, occur violating the regulations
Type is bnNumber;
Accident pattern order is the total number of persons of the driver of m
Driver breaks rules and regulations total degree
Driver's type of violation scaling matrices
S25, using the ratio of driver's break in traffic rules and regulations as ordinate, the type of driver's break in traffic rules and regulations is abscissa, will be all types of
The ratio for all kinds of breaks in traffic rules and regulations that the driver of traffic accident is occurred is drawn in this coordinate system, and all kinds of accident pattern drivers are observed
All kinds of ratios violating the regulations, desk study accident and relationship violating the regulations.
S26, representational type of fault and its simple correspondence analysis of accident pattern progress, the standard of selection are chosen are as follows:
Wherein, δnFor distribution proportion and total accident pattern of the driver in every a kind of type of violation of various accident patterns
The difference that 2 times of number distribution proportion;λ is type of fault outlier threshold corresponding with classification violating the regulations;μ is every a kind of type of violation number
Threshold value.
The practical significance of selection standard is the distribution ratio for finding out the drivers of various accident patterns in every a kind of type of violation
The difference of example and total accident pattern number distribution proportion show that type of violation and accident pattern number distribution proportion are abnormal with this
Value;
S3, traffic accident type and driver's driving age, gender and the Multiple Correspondence Analysis for driving vehicle;It is mentioned from casualty data
Take accident responsibility side's identity information and its corresponding accident pattern, four class of vehicle brand styles, driver's driving age and gender
Information;T class accident pattern, y class are driven type of vehicle, j class driving age classification, gender to correspond, generate x=(2*t* in total
Y*j) class combines;The corresponding number of every a kind of combination is counted, it is more using being obtained in statistics to the method for Multiple Correspondence Analysis
Weight corresponding relationship.
Driver's driving age, gender, the Multiple Correspondence Analysis packet for driving type of vehicle and generation traffic accident type in step S3
Include following steps:
S31, according to occur traffic accident classification of type accident responsibility side driver;
S32, simple statistics are carried out to the driving age of these drivers, obtains driver and change of the accident quantity on driver's driving age occurs
Change trend;
The driving age of accident responsibility side driver is carried out segmentation classification by S33, the variation tendency obtained according to step S32;
S34, according to vehicle brand type by vehicle classification;
Driving age of each driver under S35, the every class accident pattern of statistics, gender, driving type of vehicle, merge similar
Statistics is generated corresponding number under the combination of x=(2*t*y*j) class and every a kind of combination in total;
S36, the result that step S35 is obtained is obtained into multiple corresponding relationship using the method for correspondence analysis, in two-dimensional space
With the relationship between the reflection variable of the formal intuition of distance between points in coordinate.
Experimental verification
Data Analysis Services done in embodiment method mainly use EXCEL and IBM SPSS Statistics
22.0 statistical analysis softwares;It is related in the based process means such as the screening, comparison, PivotTables of EXCEL and SPSS
Corresponding analysis method.
Such as Fig. 1, a kind of traffic accidents reason analysis method based on correspondence analysis model, comprising the following steps:
Input driver's violation data and casualty data, the two are corresponded to each other by driver identification information.
Case data contains certain city annual 56651 traffic accident data in 2015 and 2014-2016 2 years
61473 break in traffic rules and regulations data are related to more than 113302 people of personnel;Wherein, casualty data includes driver identification card number, accident class
Not, accountability information, driver's driving age, gender, vehicle brand;Violation data include driver identification card number, code name violating the regulations,
Content violating the regulations.
It is handled by Microsoft Excel, is that of a sort driver and its merging of type of violation data mention by accident occurrence type
It takes, obtains that its normal hair property type of violation situation of the driver of certain class accident occurs;Generation quantity according to certain type of violation is by 200
To 20 kinds, as a result as shown in Figure 2 kind violation data is screened.
Responsible driver and its situation discovery of breaking rules and regulations in two years in comparison accident, type of violation concentrates on a few
Class;And it was found that different accident patterns corresponds to different high-incidence types of breaking rules and regulations, that is, people's warp of certain a kind of traffic accident occurs
The a certain seed type of breaking rules and regulations of perseverance;Fig. 2 shows the corresponding generation high-incidence traffic of 20 classes of driver that 9 class traffic accident types occur
Ratio shared by type of violation;Such as occur rear-end collision people its occur " enter guided vehicle road, by prescribed direction travel
" this type of violation just will be high (accounting 22.29%) than other type of violation;Accident occurrence type is traffic signal violation
Driver, occur " operating motor vehicles on the road other than highway, city expressway not by provide lanes "
This accident pattern accounting is up to 60.66%;Therefore guess, possibility can be prejudged by driver's history act of violating regulations
Accident pattern, to there is driver targetedly to take precautions against traffic accident;
According to above-mentioned exploration thinking, chooses representational type of fault and its accident pattern carries out correspondence analysis;It chooses
δn> 0, cnThe type of violation as experiment of > 50 chooses accident pattern serial number 1-7 as experiment according to obtained matrix
Accident pattern;The practical significance of selection standard is the distribution for finding out the drivers of various accident patterns in every a kind of type of violation
The difference of ratio and total accident pattern number distribution proportion show that type of violation and accident pattern number distribution proportion are different with this
Normal value;Experimental subjects and result table 1
The type of fault and classification violating the regulations that table 1 is tested
Above-mentioned 7 class accident pattern and its corresponding 13 class are broken rules and regulations the quantity of high-incidence type, correspondence analysis is carried out.
The statistical abstract table of 2 correspondence analysis of table
A. freedom degree 72
Table 2 is that the statistical abstract table of correspondence analysis subtracts 1 according to type of fault (7 class) in this experiment, i.e., 6 dimensions
Degree;" inertia ratio " indicates characteristic value, is to measure the index for explaining data variation ability;The first dimension illustrates 42.8% in table
Variation, the second dimension illustrates remaining 28.4% variation;Two-dimensional standard deviation 0.22-0.30 illustrates point estimate ratio
Relatively accurate, the related coefficient -0.024 of the factor then illustrates that Factorization is highly stable;Therefore the experiment of this correspondence analysis is reliable
Property is higher.
Shown in Fig. 3 is the scatter plot of correspondence analysis, and obvious in Fig. 3 is that type of fault occur be " 1 " and classification
The driver of " 7 " is higher with type of violation " g ", " b ", the correlation of " d ";There is the driver and type of violation that type of fault is " 3 "
The correlation of " a " and " f " are higher;It is higher for the driver of " 2 " and the correlation of type of violation " e " to there is type of fault;There is thing
Therefore classification is higher for the driver of " 4 " and the correlation of type of violation " c ";Occur driver that type of fault is " 5 " and " 6 " its separated
Without apparent difference in chapter classification.
From the point of view of next is according to type of violation and corresponding quantity, there is the driver that accident pattern is " 1 ", " 3 ", " 6 ", " 7 ",
One kind can be approximately divided into;It thereby it is assumed that out, there are biggish similar for the driving behavior of this kind of accident drivers.
Accident responsibility side's identity information and its corresponding accident pattern, vehicle brand class are extracted from casualty data
Four category information of type, driver's driving age and gender, such as table 3;Accident pattern is divided into nine classes according to initial data, it is specific classify situation and
Weight frequency in table 3 as divided shown in table 1;Vehicle brand styles are divided into six classes according to vehicle system, specific situation and the weight of classifying
Frequency in table 3 as divided shown in table 2;Driving age classification is divided into 4 classes according to responsible party driver's driving age distribution situation;Specific classification situation and
Weight frequency in table 3 as divided shown in table 3;Accident responsibility side's Gender Classification and its weight frequency in table 3 as divided shown in table 4.
By the accident pattern of accident responsibility side's generation, driving type of vehicle, driving age classification, gender one in 29098 accidents
One is corresponding, amounts to and generates 356 kinds of efficient combination types, using in statistics to the method for Multiple Correspondence Analysis, in dots
The more intuitive relationship reacted by distance between two classified variables in two-dimensional space.
3 descriptive statistic summary sheet of table
The model abstract of processing result is shown: Cronbach's Alpha average value is equal to 0.572, in pilot study
The upper certainty value is in tolerance interval;Dimension 1 and dimension 2 amount up to 87.6% to the variable explanation degree of model, are considered as anti-
Most of feature of model is reflected;Therefore it is suitable that the statistical tool of Multiple correspondence analysis is selected in the case.
4 model of table abstract
A.Cronbach's Alpha average value is based on mean eigenvalue;
Shown in Fig. 4 is the scatter plot of Multiple Correspondence Analysis, i.e., shows the potential of classification and sample by way of figure
Relationship, column point nearlyr expression relationship at a distance from row point are closer;The result finally obtained is more intuitively:
Driving age is that the driver of 0-2 is easier to the accident pattern to knock into the back;Driving age is that the driver of 3-7 is easy to happen violation
The accident pattern of traffic signals;Driving age is that the driver of 8-12 is easier to occur because of change into lower gear non-when stopping, does not draw parking braking,
Lead to the accident pattern of vehicle sliding;12 years or more old drivers do not have particular incident occurrence type.
Male driver is easier to occur because of change into lower gear non-when stopping, does not draw parking braking, leads to the accident class of vehicle sliding
Type;Women driver is easier to the accident pattern for occurring not give way by regulation;Men and women driver is easy to happen retrograde accident pattern.
Data analysis shows that: non-change into lower gear when the car owner of domestic car is easier to occur because of parking does not draw parking braking, causes
The accident pattern of vehicle sliding;U.S.A is the accident pattern that the car owner of vehicle is easier to move backward;The car owner of Korea Spro system vehicle is easier to chase after
The accident pattern of tail;The accident pattern that the car owner of Japanese vehicle occurs is mostly the other types that should be born all the responsibilities in accordance with the law;Domestic car phase
To be easier to occur traffic accident and accident pattern it is more complicated;And it is safer for the relatively other kinds of vehicle of German vehicle.
In general: driving domestic car, the driving age is the male driver of 8-12, is easier to occur for classification because stopping
Che Shiwei change into lower gear does not draw parking braking, leads to the accident pattern of vehicle sliding.
Claims (5)
1. a kind of traffic accidents reason analysis method based on correspondence analysis model, it is characterised in that: include the following steps,
S1, input break in traffic rules and regulations data and traffic accident data, and break in traffic rules and regulations data and traffic accident data are passed through into driver
Identity information is associated;
S2, traffic accident type and break in traffic rules and regulations type correspondence analysis, first progress exploratory analysis, are united using pivot chart
Violation data is counted, and filters out the sample data violating the regulations of subsequent analysis by threshold method;It finds out corresponding under every kind of type of violation
The accident pattern of accident crowd and its generation counts accounting of the every kind of accident pattern in all accident patterns;Secondly, choosing has
Representative type of fault and its accident pattern carry out simple correspondence analysis;In step S2, exploratory analysis is carried out specifically:
S21, according to traffic accident type, accident driver is classified;
S22, according to accident driver information, the record violating the regulations of screening accidents happened driver, counts each from break in traffic rules and regulations tables of data
All kinds of numbers violating the regulations of accident pattern driver;
S23, according to all kinds of quantity violating the regulations after statistics, retain the type of violation that number violating the regulations is more than or equal to p times, wherein p is separated
Chapter screening sample threshold value;
S24, the total degree for calculating each type traffic accident corresponding various break in traffic rules and regulations numbers and each type break in traffic rules and regulations
Ratio;
S25, using the ratio of driver's break in traffic rules and regulations as ordinate, the type of driver's break in traffic rules and regulations is abscissa, by all types of traffic
The ratio for all kinds of breaks in traffic rules and regulations that the driver of accident is occurred is drawn in this coordinate system, and each of all kinds of accident pattern drivers is observed
Class ratio violating the regulations, desk study accident and relationship violating the regulations;
S3, traffic accident type and driver's driving age, gender and the Multiple Correspondence Analysis for driving vehicle;Thing is extracted from casualty data
Therefore responsible party's identity information and its corresponding accident pattern, vehicle brand styles, driver's driving age and four class of gender letter
Breath;T class accident pattern, y class are driven type of vehicle, j class driving age classification, gender to correspond, generate x=2*t*y*j in total
Class combination;The corresponding number of every a kind of combination is counted, it is multiple right using being obtained in statistics to the method for Multiple Correspondence Analysis
It should be related to.
2. as described in claim 1 based on the traffic accidents reason analysis method of correspondence analysis model, it is characterised in that: step
In S1, the feature vector of traffic accident data is CS=[C, K, V], and wherein C is accident occurrence type;K is personal information, including
Driver's gender, driving age, accident responsibility ownership;V is vehicle brand information;The feature vector of break in traffic rules and regulations data is WS=[A, H],
Wherein A is personnel identity information violating the regulations, and H is the content violating the regulations of personnel violating the regulations.
3. as described in claim 1 based on the traffic accidents reason analysis method of correspondence analysis model, it is characterised in that: described
Step S24 specifically:
Traffic accident A={ a1, a2... ai... am, wherein aiFor traffic accident type, m is accident pattern order;
Accident number K={ k1, k2... k3..., km, wherein kiIt is a for traffic accident type occursiNumber;
Type of violation B={ b1, b2... bi..., bn, biFor break in traffic rules and regulations type, n is type of violation order;
Number violating the regulationskmbnIt is a for driver's traffic accident typem, type of violation occurs
For bnNumber;
Accident pattern order is the total number of persons of the driver of m
Driver breaks rules and regulations total degree
Driver's type of violation scaling matrices
4. as described in claim 1 based on the traffic accidents reason analysis method of correspondence analysis model, it is characterised in that: described
In step S2, chooses representational type of fault and its accident pattern carries out simple correspondence analysis, the standard of selection are as follows:
Wherein, δnFor distribution proportion and total accident pattern number of the driver in every a kind of type of violation of various accident patterns
The difference that 2 times of distribution proportion;λ is type of fault outlier threshold corresponding with classification violating the regulations;μ is the threshold of every a kind of type of violation number
Value.
5. the traffic accidents reason analysis method according to any one of claims 1-4 based on correspondence analysis model, feature
It is: driver's driving age, gender, the Multiple Correspondence Analysis packet for driving type of vehicle and generation traffic accident type in the step S3
Include following steps:
S31, according to occur traffic accident classification of type accident responsibility side driver;
S32, simple statistics are carried out to driving age of these drivers, obtains driver variation of the accident quantity on driver's driving age occurs to become
Gesture;
The driving age of accident responsibility side driver is carried out segmentation classification by S33, the variation tendency obtained according to step S32;
S34, according to vehicle brand type by vehicle classification;
Driving age of each driver under S35, the every class accident pattern of statistics, gender, driving type of vehicle, merge similar statistics
Corresponding number under the combination of x=2*t*y*j class and every a kind of combination is generated in total;
S36, the result that step S35 is obtained is obtained into multiple corresponding relationship using the method for correspondence analysis, in 2-d spatial coordinate
In with the formal intuition of distance between points reflection variable between relationship.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610807054.8A CN106448158B (en) | 2016-09-06 | Traffic accidents reason analysis method based on correspondence analysis model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610807054.8A CN106448158B (en) | 2016-09-06 | Traffic accidents reason analysis method based on correspondence analysis model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106448158A CN106448158A (en) | 2017-02-22 |
CN106448158B true CN106448158B (en) | 2019-07-16 |
Family
ID=
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6662141B2 (en) * | 1995-01-13 | 2003-12-09 | Alan R. Kaub | Traffic safety prediction model |
CN1889144A (en) * | 2006-07-24 | 2007-01-03 | 王成利 | Electronic detecting device for carrying out management to motor vehicles and drivers |
CN102779430A (en) * | 2011-05-12 | 2012-11-14 | 德尔福技术有限公司 | Vision based night-time rear collision warning system, controller, and method of operating the same |
CN103914945A (en) * | 2014-03-25 | 2014-07-09 | 刘业兴 | Monitoring and managing system and method for driver of motor vehicle |
CN104331740A (en) * | 2013-07-22 | 2015-02-04 | 李娟� | Management device for motor vehicles and drivers |
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6662141B2 (en) * | 1995-01-13 | 2003-12-09 | Alan R. Kaub | Traffic safety prediction model |
CN1889144A (en) * | 2006-07-24 | 2007-01-03 | 王成利 | Electronic detecting device for carrying out management to motor vehicles and drivers |
CN102779430A (en) * | 2011-05-12 | 2012-11-14 | 德尔福技术有限公司 | Vision based night-time rear collision warning system, controller, and method of operating the same |
CN104331740A (en) * | 2013-07-22 | 2015-02-04 | 李娟� | Management device for motor vehicles and drivers |
CN103914945A (en) * | 2014-03-25 | 2014-07-09 | 刘业兴 | Monitoring and managing system and method for driver of motor vehicle |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Guo et al. | Individual driver risk assessment using naturalistic driving data | |
Lee et al. | A framework for evaluating aggressive driving behaviors based on in-vehicle driving records | |
Seraneeprakarn et al. | Occupant injury severities in hybrid-vehicle involved crashes: A random parameters approach with heterogeneity in means and variances | |
Arbabzadeh et al. | A data-driven approach for driving safety risk prediction using driver behavior and roadway information data | |
Wang et al. | Assessing the relationship between self-reported driving behaviors and driver risk using a naturalistic driving study | |
Bagdadi | Assessing safety critical braking events in naturalistic driving studies | |
Eluru et al. | A joint econometric analysis of seat belt use and crash-related injury severity | |
Bagdadi | Estimation of the severity of safety critical events | |
Wali et al. | Exploring microscopic driving volatility in naturalistic driving environment prior to involvement in safety critical events—Concept of event-based driving volatility | |
CN111401414B (en) | Natural driving data-based dangerous scene extraction and classification method | |
DE102006038151A1 (en) | Method and device for controlling personal protective equipment | |
Al-Bdairi et al. | Comparison of contributing factors for injury severity of large truck drivers in run-off-road crashes on rural and urban roadways: Accounting for unobserved heterogeneity | |
Wang et al. | Effect of daily car-following behaviors on urban roadway rear-end crashes and near-crashes: A naturalistic driving study | |
EP4067186A1 (en) | Method for determining a trajectory of an at least partially assisted motor vehicle, computer program and assistant system | |
Kong et al. | Mining patterns of near-crash events with and without secondary tasks | |
Mori et al. | Integrated modeling of driver gaze and vehicle operation behavior to estimate risk level during lane changes | |
Teimouri et al. | A real-time warning system for rear-end collision based on random forest classifier | |
Oh et al. | In-depth understanding of lane changing interactions for in-vehicle driving assistance systems | |
Zou et al. | Multivariate analysis of car-following behavior data using a coupled hidden Markov model | |
CN106448158B (en) | Traffic accidents reason analysis method based on correspondence analysis model | |
Toledo et al. | Alternative definitions of passing critical gaps | |
Krampe et al. | Injury severity for hazard & risk analyses: calculation of ISO 26262 S-parameter Values from Real-World Crash Data | |
Lee et al. | Detection of drowsy driving based on driving information | |
Guyonvarch et al. | Evaluation of safety critical event triggers in the UDrive data | |
Li et al. | Estimating driver crash risks based on the extended Bradley–Terry model: an induced exposure method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
CP02 | Change in the address of a patent holder |
Address after: 211100 No. 19 Suyuan Avenue, Jiangning Economic and Technological Development Zone, Nanjing City, Jiangsu Province Patentee after: JIANGSU ZHITONG TRAFFIC TECHNOLOGY Co.,Ltd. Address before: 210006, Qinhuai District, Jiangsu, Nanjing should be 388 days street, Chenguang 1865 Technology Creative Industry Park E10 building on the third floor Patentee before: JIANGSU ZHITONG TRAFFIC TECHNOLOGY Co.,Ltd. |