CN106557663A - Driving behavior methods of marking and device - Google Patents
Driving behavior methods of marking and device Download PDFInfo
- Publication number
- CN106557663A CN106557663A CN201611061545.9A CN201611061545A CN106557663A CN 106557663 A CN106557663 A CN 106557663A CN 201611061545 A CN201611061545 A CN 201611061545A CN 106557663 A CN106557663 A CN 106557663A
- Authority
- CN
- China
- Prior art keywords
- item
- index
- sample
- driving behavior
- scoring
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
Landscapes
- Traffic Control Systems (AREA)
Abstract
The application proposes a kind of driving behavior methods of marking and device, and the method includes the vehicle operation data for obtaining driving behavior to be scored, wherein, vehicle operation data includes:Multiple index item in vehicle traveling, and the corresponding index alert data of each index item;Vehicle operation data is counted, and in determining the stroke fragment that vehicle operation data is produced under driving behavior to be scored, the meansigma methodss of alarm times of each index item in unit distance;According to alarm times meansigma methodss and default characteristic model, the safety trend numerical value of each index item in stroke fragment is determined;The safety trend numerical value of the multiple index item during the corresponding weight of each index item is read from default weight table to stroke fragment is weighted the safety trend numerical value after summation as the appraisal result for treating scoring driving behavior.The reasonability and referring to property of scoring due to reference to more index item, making appraisal result more accurate, are improve, the generation of vehicle accident is reduced.
Description
Technical field
The application is related to vehicle drive behavioral analysis technology field, more particularly to a kind of driving behavior methods of marking and dress
Put.
Background technology
With expanding economy and the improvement of people's living standards, the quantity of motor vehicles is more and more, meanwhile, also cause
The frequent generation of vehicle accident, the life and property safety for giving people bring huge loss, how to avoid traffic accident
Generation, it has also become one of huge difficult problem of current transportation industry.
At present, the vehicle insurance product fixed a price based on driver's driving behavior is more and more, is gone by driving to driver
To be scored, and then corresponding vehicle insurance product is provided, the safe driving consumption view of driver, and the peace of driver can be improved
It is complete to drive consciousness, so as to reduce social road accident rate.
However, existing methods of marking, the accekeration being based only upon during the fatigue data of driver, vehicle are travelled etc. are limited
Several vehicle datas carry out driving behavior scoring, and appraisal result is not accurate, and reference value is low, poor user experience.
The content of the invention
The purpose of the application is intended at least to solve to a certain extent one of technical problem in correlation technique.
For this purpose, first purpose of the application is to propose a kind of driving behavior methods of marking, the method combines more
Index item, makes appraisal result more accurate, improves the reasonability and referring to property of scoring, reduces the generation of vehicle accident, change
It has been apt to Consumer's Experience.
Further object is to propose a kind of driving behavior scoring apparatus.
To reach above-mentioned purpose, the driving behavior methods of marking that the application first aspect embodiment is proposed, including:Acquisition is treated
The vehicle operation data of scoring driving behavior, wherein, the vehicle operation data includes:Multiple index item in vehicle traveling,
With the corresponding index alert data of each index item;The vehicle operation data is counted, and determination is waited to score described
Produce under driving behavior in the stroke fragment of the vehicle operation data, alarm times of each index item in unit distance
Meansigma methodss;According to the alarm times meansigma methodss and default characteristic model, each index item in the stroke fragment is determined
Safety trend numerical value;The corresponding weight of each index item described in reading from default weight table, and according to the corresponding power
Summation is weighted to the safety trend numerical value of the multiple index item in the stroke fragment again;By the safety after weighted sum
Tendentiousness numerical value is used as the appraisal result to the driving behavior to be scored.
The driving behavior methods of marking of the embodiment of the present application, obtains the vehicle operation data of driving behavior to be scored first,
Wherein, vehicle operation data includes:Multiple index item in vehicle traveling, and the corresponding index alert data of each index item,
Then vehicle operation data is counted, and determines the stroke fragment that vehicle operation data is produced under driving behavior to be scored
In, the meansigma methodss of alarm times of each index item in unit distance, then according to alarm times meansigma methodss and default feature
Model, determines the safety trend numerical value of each index item in stroke fragment, finally reads each from default weight table and refer to
The mark corresponding weight of item, and according to corresponding weight to stroke fragment in multiple index item safety trend numerical value carry out plus
Safety trend numerical value after power summation is used as the appraisal result for treating scoring driving behavior.Due to reference to more index item,
Make appraisal result more accurate, improve the reasonability and referring to property of scoring, reduce the generation of vehicle accident, improve user
Experience.
In addition, can also have following additional technology special according to the driving behavior methods of marking of the above embodiments of the present application
Levy:
In one embodiment of the invention, the default characteristic model is set up by following steps, including:
The a plurality of vehicle sample running data of multiple sample stroke fragments is obtained, wherein, a plurality of vehicle sample traveling
Every vehicle sample running data in data includes:Multiple sample index's items in vehicle traveling, and each sample index's item
Corresponding sample index's alert data;
To the plurality of sample stroke fragment according to producing whether collision accident divides, collision accident is not produced
First sample stroke fragment and produce collision accident the second sample stroke fragment;
Every vehicle sample running data of the first sample stroke fragment is counted, and determine it is described each the
In the same this stroke fragment, the first alarm times that each sample index's item produces alarming index item in unit distance are average
Value;
Every vehicle sample running data of the second sample stroke fragment is counted, and determine it is described each the
In two sample stroke fragments, the second alarm times that each sample index's item produces alarming index item in unit distance are average
Value;
Each sample index's item corresponding multiple first alarm times standard errors of the mean described in calculating, and by the mark
Quasi- difference is not used as producing corresponding first standard deviation of each sample index's item in the first sample stroke fragment of collision accident;
Each sample index's item corresponding multiple second alarm times standard errors of the mean described in calculating, and by the mark
Quasi- difference is used as corresponding second standard deviation of each sample index's item in the second sample stroke fragment of generation collision accident;
The plurality of first alarm times meansigma methodss are carried out being averaging with computing, and the numerical value that average calculating operation is obtained as
Corresponding first meansigma methodss of each sample index's item in the first sample stroke fragment of collision accident are not produced;
The plurality of second alarm times meansigma methodss are carried out being averaging with computing, and the numerical value that average calculating operation is obtained as
In producing the second sample stroke fragment of collision accident, corresponding second meansigma methodss of each sample index's item are described default to set up
Characteristic model.
In one embodiment of the invention, the default weight table is set up by following steps, including:
Each the sample index's item for calculating every vehicle sample running data predicts the outcome;
Whether the actual collision result of the every vehicle sample running data is obtained according to generation collision accident;
Corresponding confusion matrix, and root are generated with actual collision result according to predicting the outcome for each sample index's item
The prediction accuracy of each sample index's item is calculated according to the data in the corresponding confusion matrix;
Obtain the prediction accuracy of multiple sample index's items plus and value, and by the prediction of each sample index's item standard
Exactness with it is described plus and be worth ratio value as sample index's item each described weight;
The index item and corresponding weight are recorded to generate the default weight table.
In one embodiment of the invention, described each sample index for calculating every vehicle sample running data
Predict the outcome, including:
The first standard deviation in alarm times meansigma methodss, the default characteristic model according to each sample index's item
With the corresponding safe probability value of each sample index's item described in first mean value calculation;
The second standard deviation in alarm times meansigma methodss, the default characteristic model according to each sample index's item
With the corresponding dangerous probit of each sample index's item described in second mean value calculation;
When the corresponding safe probability value of described each sample index's item is more than dangerous probit, sample index's item
Predict the outcome not produce collision accident;
When the corresponding safe probability value of described each sample index's item is less than dangerous probit, sample index's item
Predict the outcome to produce collision accident.
In one embodiment of the invention, it is described according to the alarm times meansigma methodss and default characteristic model, it is determined that
The safety trend numerical value of each index item in the stroke fragment, including:
The first standard deviation in alarm times meansigma methodss, the default characteristic model and institute according to each index item
State the corresponding safe probability value of each index item described in the first mean value calculation;
The second standard deviation in alarm times meansigma methodss, the default characteristic model and institute according to each index item
State the corresponding dangerous probit of each index item described in the second mean value calculation;
The stroke piece is determined according to the corresponding safe probability value of described each index item and the dangerous probit
The safety trend numerical value of each index item in section.
In one embodiment of the invention, the safety trend of each index item in the determination stroke fragment
After numerical value, also include:
The safety trend numerical value and the first default value of each index item in the stroke fragment are made into product, and will
Scoring of each product value as each index item.
In one embodiment of the invention, in the safety trend numerical value using after weighted sum as treating to described
After the appraisal result of scoring driving behavior, also include;
Safety trend numerical value after the weighted sum is made into product with first default value, and product value is made
For the scoring of the driving behavior to be scored.
In one embodiment of the invention, described driving behavior methods of marking, also includes:
Whether the scoring of the scoring and/or driving behavior score of each index item described in judging is more than or equal to the
Two default values, obtain the first judged result;
The vehicle operation data of neural network model and Random Forest model to the driving behavior to be scored is respectively adopted
It is predicted, obtains the second judged result based on the neural network model and sentence based on the 3rd of the Random Forest model the
Disconnected result;
Using voting mechanism, first judged result, second judged result, and the 3rd judged result pair
The scoring of the scoring of each index item and/or the driving behavior to be scored is adjusted.
In one embodiment of the invention, the employing voting mechanism, first judged result, described second judge
And scoring of the 3rd judged result to the scoring and/or the driving behavior to be scored of each index item as a result,
It is adjusted, including:
In first judged result, second judged result, and the 3rd judged result, at least two sentence
Disconnected result is touched not produce collision accident/generation not produce collision accident/generation collision accident, and first judged result
When hitting event, scoring and/or the scoring of the driving behavior to be scored not to each index item are adjusted;
In first judged result, second judged result, and the 3rd judged result, at least two sentence
Disconnected result is not to produce collision accident/generations collision accident, and first judged result is generation collision accident/do not produce and touch
When hitting event, the scoring of scoring and/or the driving behavior to be scored to each index item is adjusted.
In one embodiment of the invention, it is described set up the default characteristic model and the default weight table it
Afterwards, also include:
The vehicle sample running data is gathered every predetermined period;
The default characteristic model and the default weight table are updated according to the vehicle sample running data.
To reach above-mentioned purpose, the driving behavior scoring apparatus that the application second aspect embodiment is proposed, including:First obtains
Delivery block, for obtaining the vehicle operation data of driving behavior to be scored, wherein, the vehicle operation data includes:Vehicle row
Multiple index item in sailing, and the corresponding index alert data of each index item;First processing module, for the vehicle row
Sail data to be counted, and in determining the stroke fragment for producing the vehicle operation data under the driving behavior to be scored,
The meansigma methodss of alarm times of each index item in unit distance;Determining module, for according to the alarm times meansigma methodss
And default characteristic model, determine the safety trend numerical value of each index item in the stroke fragment;Second processing module, uses
In the corresponding weight of each index item described in the reading from default weight table, and according to the corresponding weight to the stroke piece
The safety trend numerical value of the multiple index item in section is weighted summation;Grading module, for by the safety after weighted sum
Tendentiousness numerical value is used as the appraisal result to the driving behavior to be scored.
The driving behavior scoring of the embodiment of the present application, obtains the vehicle operation data of driving behavior to be scored first, wherein,
Vehicle operation data includes:Multiple index item in vehicle traveling, and the corresponding index alert data of each index item, it is then right
Vehicle operation data is counted, and in determining the stroke fragment that vehicle operation data is produced under driving behavior to be scored, often
The meansigma methodss of alarm times of the individual index item in unit distance, then according to alarm times meansigma methodss and default characteristic model,
Determine the safety trend numerical value of each index item in stroke fragment, finally each index item pair is read from default weight table
The weight answered, and according to corresponding weight to stroke fragment in the safety trend numerical value of multiple index item be weighted summation
Safety trend numerical value afterwards is used as the appraisal result for treating scoring driving behavior.Due to reference to more index item, making scoring
As a result more precisely, the reasonability and referring to property of scoring are improve, the generation of vehicle accident is reduced, is improved Consumer's Experience.
In addition, can also have following additional technology special according to the driving behavior scoring apparatus of the above embodiments of the present application
Levy:
In one embodiment of the invention, described driving behavior scoring apparatus, also include:
Second acquisition module, for obtaining a plurality of vehicle sample running data of multiple sample stroke fragments, wherein, it is described
Every vehicle sample running data in a plurality of vehicle sample running data includes:Multiple sample index's items in vehicle traveling,
With the corresponding sample index's alert data of each sample index's item;
Division module, obtains according to producing whether collision accident divides for the plurality of sample stroke fragment
The first sample stroke fragment for not producing collision accident and the second sample stroke fragment for producing collision accident;
3rd processing module, for uniting to every vehicle sample running data of the first sample stroke fragment
In meter, and each first sample stroke fragment described in determining, each sample index's item produces alarming index item in unit distance
The first alarm times meansigma methodss;
Fourth processing module, for uniting to every vehicle sample running data of the second sample stroke fragment
In meter, and each second sample stroke fragment described in determining, each sample index's item produces alarming index item in unit distance
The second alarm times meansigma methodss;
5th processing module, for calculating the corresponding multiple first alarm times meansigma methodss of described each sample index's item
Standard deviation, and as in the first sample stroke fragment for not producing collision accident, each sample index's item is corresponding using the standard deviation
The first standard deviation;
6th processing module, for calculating the corresponding multiple second alarm times meansigma methodss of described each sample index's item
Standard deviation, and the standard deviation is corresponding as each sample index's item in the second sample stroke fragment for producing collision accident
Second standard deviation;
7th processing module, for carrying out being averaging computing to the plurality of first alarm times meansigma methodss, and will be average
The numerical value that computing is obtained is not used as producing each sample index's item corresponding first in the first sample stroke fragment of collision accident
Meansigma methodss;
8th processing module, for carrying out being averaging computing to the plurality of second alarm times meansigma methodss, and will be average
The numerical value that computing is obtained is corresponding second flat as each sample index's item in the second sample stroke fragment of collision accident is produced
Average is setting up the default characteristic model.
In one embodiment of the invention, described driving behavior scoring apparatus, also include:
Computing module, for calculating predicting the outcome for each sample index's item of every vehicle sample running data;
3rd acquisition module, for the reality of every vehicle sample running data whether is obtained according to generation collision accident
Collide result in border;
First prediction module, for right with the generation of actual collision result according to predicting the outcome for each sample index's item
The confusion matrix answered, and the prediction of each sample index's item according to the data in the corresponding confusion matrix are calculated is accurate
Degree;
9th processing module, for obtaining adding and value for the prediction accuracy of multiple sample index's items, and by described in each
The prediction accuracy of sample index's item with it is described plus and be worth ratio value as sample index's item each described weight;
Generation module, for being recorded to generate the default weight table by the index item and corresponding weight.
In one embodiment of the invention, the computing module includes:
First computing unit, for the alarm times meansigma methodss according to each sample index's item, the default feature
The corresponding safe probability value of each sample index's item described in the first standard deviation and first mean value calculation in model;
Second computing unit, for the alarm times meansigma methodss according to each sample index's item, the default feature
The corresponding dangerous probit of each sample index's item described in the second standard deviation and second mean value calculation in model;
First comparing unit, for being more than dangerous probit in the corresponding safe probability value of described each sample index's item
When, sample index's item predicts the outcome not produce collision accident;
Second comparing unit, for being less than dangerous probit in the corresponding safe probability value of described each sample index's item
When, sample index's item predict the outcome for produce collision accident.
In one embodiment of the invention, the determining module includes:
3rd computing unit, for the alarm times meansigma methodss according to each index item, the default characteristic model
In the first standard deviation and the corresponding safe probability value of each index item described in first mean value calculation;
4th computing unit, for the alarm times meansigma methodss according to each index item, the default characteristic model
In the second standard deviation and the corresponding dangerous probit of each index item described in second mean value calculation;
Determining unit, for true according to the corresponding safe probability value of described each index item and the dangerous probit
The safety trend numerical value of each index item in the fixed stroke fragment.
In one embodiment of the invention, described driving behavior scoring apparatus, also include:
First product module, for will be the safety trend numerical value of each index item in the stroke fragment pre- with first
If numerical value makees product, and using each product value as each index item scoring.
In one embodiment of the invention, described driving behavior scoring apparatus, also include;
Second product module, for the safety trend numerical value after the weighted sum is made with first default value
Product, and using product value as the driving behavior to be scored scoring.
In one embodiment of the invention, described driving behavior scoring apparatus, also include:
First judge module, for judging the scoring of each index item and/or commenting for the driving behavior to be scored
Divide and whether be more than or equal to the second default value, obtain the first judged result;
To described, second prediction module, treats that scoring drives row for neural network model and Random Forest model is respectively adopted
For vehicle operation data be predicted, obtain the second judged result based on neural network model and based on the random forest
3rd judged result of model;
Adjusting module, for using voting mechanism, first judged result, second judged result and described
Scoring of 3rd judged result to the scoring and/or the driving behavior to be scored of each index item is adjusted.
In one embodiment of the invention, described driving behavior scoring apparatus, the adjusting module include:
First adjustment unit, in first judged result, second judged result, and the described 3rd judges
As a result, in, at least two judged results are not to produce collision accident/generation collision accident, and first judged result is not produce
During raw collision accident/generation collision accident, the not scoring to each index item and/or the driving behavior to be scored is commented
Divide and be adjusted;
Second adjustment unit, in first judged result, second judged result, and the described 3rd judges
As a result, in, at least two judged results are not to produce collision accident/generation collision accident, and first judged result is to produce
During collision accident/do not produce collision accident, the scoring of scoring and/or the driving behavior to be scored to each index item
It is adjusted.
In one embodiment of the invention, described driving behavior scoring apparatus, set up the default feature described
After model and the default weight table, also include:
Acquisition module, for gathering the vehicle sample running data every predetermined period;
Update module, for according to the vehicle sample running data to the default characteristic model and the default weight
Table is updated.
To reach above-mentioned purpose, the storage medium that the application fourth aspect embodiment is proposed, wherein, the storage medium is used
In storage application program, the application program is used to operationally perform the driving behavior described in first aspect present invention embodiment
Methods of marking.
To reach above-mentioned purpose, the application program that the 5th aspect embodiment of the application is proposed, wherein, the application program is used
In the driving behavior methods of marking described in operationally execution first aspect present invention embodiment.
The aspect and advantage that the application is added will be set forth in part in the description, and partly will become from the following description
Obtain substantially, or recognized by the practice of the application.
Description of the drawings
The above-mentioned and/or additional aspect of the application and advantage will become from the following description of the accompanying drawings of embodiments
It is substantially and easy to understand, wherein:
Fig. 1 is the flow chart of the driving behavior methods of marking according to the application one embodiment;
Fig. 2 is the flow chart of the driving behavior methods of marking according to the application another embodiment;
Fig. 3 is the structural representation of the driving behavior scoring apparatus according to the application one embodiment;
Fig. 4 is the structural representation of the driving behavior scoring apparatus according to the application another embodiment;
Fig. 5 is the structural representation of the driving behavior scoring apparatus according to the application another embodiment;
Fig. 6 is the structural representation of the driving behavior scoring apparatus according to the application a still further embodiment;
Fig. 7 is the structural representation of the driving behavior scoring apparatus according to the application further embodiment.
Specific embodiment
Embodiments herein is described below in detail, the example of the embodiment is shown in the drawings, wherein from start to finish
Same or similar label represents same or similar element or the element with same or like function.Below with reference to attached
The embodiment of figure description is exemplary, is only used for explaining the application, and it is not intended that restriction to the application.Conversely, this
The embodiment of application includes all changes fallen in the range of the spirit and intension of attached claims, modification and is equal to
Thing.
Below with reference to the accompanying drawings the driving behavior methods of marking and device of the embodiment of the present application are described.
In life, driving is a kind of technical ability.However, the driving technology of driver is uneven, driver is in vehicle
There are many dangerous or irrational operations in driving procedure, easily cause vehicle accident, but driver is likely to not anticipate
Know the dangerous and degree of danger size that oneself operation brings.
Generally, by being scored driver's driving behavior to be supplied to driver, to improve the safety of driver
Consciousness is driven, however, a limited number of vehicle data such as accekeration being based only upon in the fatigue data of driver, vehicle traveling enters
Row driving behavior is scored, and appraisal result is not accurate, and reference value is low, poor user experience.
In order to solve the above problems, a kind of driving behavior methods of marking of the application, during the method is by travelling to vehicle
Multiple index item, and each index item corresponding index alert data carries out statistical analysiss process, in determining stroke fragment
As the appraisal result for treating scoring driving behavior after the safety trend numerical value of each index item weighted sum.Thus, tie
More index item is closed, makes appraisal result more accurate, improve the reasonability and referring to property of scoring, reduce vehicle accident
Occur, improve Consumer's Experience.It is specific as follows:
Fig. 1 is the flow chart of the driving behavior methods of marking according to the application one embodiment.
As shown in figure 1, the driving behavior methods of marking of the embodiment of the present application is comprised the following steps:
Step 101, obtains the vehicle operation data of driving behavior to be scored, wherein, vehicle operation data includes:Vehicle row
Multiple index item in sailing, and the corresponding index alert data of each index item.
Specifically, mobile device, onboard diagnostic system (On-Board Diagnostic, abbreviation OBD) and car can be passed through
The vehicle operation data that the mode such as standby obtains driving behavior to be scored is installed before.
Wherein, vehicle operation data can be including the multiple index item in vehicle traveling and the corresponding index of each index item
Alert data.The type of index item has many kinds, can need to select according to practical application, be illustrated below:
For example, it may be collision warning, zig zag report to the police, it is anxious slow down report to the police, it is anxious accelerate to report to the police, anxious lane change alert, hypervelocity
Warning, fatigue driving warning, the warning of long-time idling, frequent lane change alert, the too high warning of rotating speed, water temperature over-high warning, neutral
Slide warning, machine oil (pressure, temperature) to report to the police, maintain warning and wheel tyre pressure warning etc..
For example, it is also possible to be vehicle single stroke distance travelled, main furnace building light switch, dipped beam lamp switch, side-marker lamp switch, mist
Lamp switch, left steering lamp switch, right turn lamp switch, dangerous lamp switch, door contact interrupter, door lock switch and car window switch etc..
For example, it is also possible to be the control of control unit of engine (Engine Control Module, abbreviation ECM)/electromotor
Module (Electronic Control Unit, abbreviation ECU), anti-blocking brake system (Antilock Brake System, letter
Claim ABS) and electronic security(ELSEC) air bag (Electronic Control of Safety Airbag, abbreviation SRS) etc..
Can also for example be that parking brake state, brake pedal, brake pedal relative position, gas pedal, gas pedal are relative
Position, clutch state, seat belt status, adaptive cruise (Adaptive Cruise Control, abbreviation ACC) signal, key
State, wiper status, air-conditioning switch gear, Engine Inlet Temperature and air-conditioning vehicle interior temperature etc..
Wherein, the type of the corresponding index alert data of each index item has many kinds, can be according to practical application needs
Select, for example may refer to mark the corresponding index alarm times of item, can also the corresponding index alert frequency of index item etc..
Step 102, counts to vehicle operation data, and determines that vehicle is produced under driving behavior to be scored travels number
According to stroke fragment in, the meansigma methodss of alarm times of each index item in unit distance.
Specifically, the vehicle operation data of acquisition is carried out counting the alarm times for obtaining each index item divided by whole row
The mileage number (kilometer, rice etc. can be made to be unit) of journey fragment, such that it is able to obtain report of each index item in unit distance
The meansigma methodss of alert number of times.
If it should be noted that multiple index item contain collision warning index item, index item code name can be made to be D, used " 0 "
Expression does not have collision warning, and " 1 " represents collision warning.It is concrete as shown in table 1:
Meansigma methodss of the table 1 for alarm times of each index item in unit distance (kilometer)
Step 103, according to alarm times meansigma methodss and default characteristic model, determines each index item in stroke fragment
Safety trend numerical value.
In particular it is required that setting up default characteristic model.In this example, set up the detailed process description of default characteristic model
It is as follows:
The a plurality of vehicle sample running data of multiple sample stroke fragments is obtained first, in a plurality of vehicle sample running data
Every vehicle sample running data include:Multiple sample index's items in vehicle traveling, and each sample index's item is corresponding
Sample index's alert data.
Wherein, the multiple sample index's items in vehicle traveling, and the corresponding sample index's warning number of each sample index's item
According to description refer in step 101 to the multiple index item in vehicle traveling, and the corresponding index warning number of each index item
According to description, no longer describe in detail herein.
Further, multiple sample stroke fragments are not produced and is touched according to producing whether collision accident divides
The first sample stroke fragment for hitting event and the second sample stroke fragment for producing collision accident.
Specifically, in order to further improve the accuracy of analysis driving behavior, multiple sample stroke fragments will be divided into
The first sample stroke fragment for not producing collision accident and the second sample stroke fragment for producing collision accident are processed.
Further, every vehicle sample running data of first sample stroke fragment is counted, and determines each
In first sample stroke fragment, the first alarm times that each sample index's item produces alarming index item in unit distance are average
Value.
Further, every vehicle sample running data of the second sample stroke fragment is counted, and determines each
In second sample stroke fragment, the second alarm times that each sample index's item produces alarming index item in unit distance are average
Value.
It should be noted that the first alarm times that each sample index's item produces alarming index item in unit distance are put down
Average and each sample index's item produce the second alarm times meansigma methodss of alarming index item in unit distance and refer to step
The description of the meansigma methodss of the alarm times in 102 to each index item in unit distance, is no longer described in detail herein.
Further, the corresponding multiple first alarm times standard errors of the mean of each sample index's item are calculated, and will
Standard deviation is not used as producing corresponding first standard deviation of each sample index's item in the first sample stroke fragment of collision accident.
Further, the corresponding multiple second alarm times standard errors of the mean of each sample index's item are calculated, and will
Standard deviation is used as corresponding second standard deviation of each sample index's item in the second sample stroke fragment of generation collision accident.
Further, multiple first alarm times meansigma methodss are carried out being averaging computing, and the number that average calculating operation is obtained
Value is not used as producing corresponding first meansigma methodss of each sample index's item in the first sample stroke fragment of collision accident.
Further, multiple second alarm times meansigma methodss are carried out being averaging computing, and the number that average calculating operation is obtained
Value is pre- to set up as corresponding second meansigma methodss of each sample index's item in the second sample stroke fragment of generation collision accident
If characteristic model.
Thus, more sample index's item is combined during setting up default characteristic model so that determine each index item
Safety trend numerical value it is more accurate, and then improve scoring reasonability and referring to property.
It should be noted that the first standard deviation of foregoing description, the second standard deviation, the first meansigma methodss and second meansigma methodss etc.
Change with the change of a plurality of vehicle sample running data of multiple sample stroke fragments, need periodicity or periodicity more
It is new ensureing precision of analysis.Need every predetermined period collection vehicle sample running data, and according to vehicle sample
This running data is updated to default characteristic model.
And then the meansigma methodss of alarm times, the first standard deviation, the second standard deviation, the first meansigma methodss and the second meansigma methodss are led
Enter preset formula or by preset algorithm model, determine the safety trend numerical value of each index item in stroke fragment.With
As a example by E1 in above-mentioned table one, the safe probability value in stroke fragment is calculated by two the formula of probability density function (1) and (2)
With dangerous probit, the safety trend numerical value of E1 is determined finally according to safe probability value and dangerous probit.Specifically describe such as
Under:
Wherein, E1 represents the alarm times meansigma methodss of index item E1.STDS(E1) represent the first standard deviation.AVGS(E1) table
Show the first meansigma methodss.STDD(E1) represent the second standard deviation.AVGD(E1) the second meansigma methodss are represented.
The safety trend numerical value of E1 can be obtained by formula (3):
In the same manner, the safety trend numerical value that E2, E3 wait until each index item of Ei can be obtained.Concrete calculating process can be joined
See E1.
Step 104, reads the corresponding weight of each index item from default weight table, and according to corresponding weight to stroke
The safety trend numerical value of the multiple index item in fragment is weighted summation.
Step 105, using the safety trend numerical value after weighted sum as the appraisal result for treating scoring driving behavior.
In particular it is required that building default weight table.In this example, the detailed process for building default weight table is described such as
Under:
First, each the sample index's item for calculating every vehicle sample running data predicts the outcome.
Specifically, each sample can be calculated by way of above-mentioned steps 103 calculate E1 safe probabilities and dangerous probability
The corresponding safe probability value of index item and dangerous probit.
The alarm times meansigma methodss according to each sample index's item, the first standard preset in characteristic model are obtained first
The second standard deviation and the second meansigma methodss in difference, the first meansigma methodss, default characteristic model passes through above-mentioned two probability density function
Formula (1) and (2) calculate the corresponding safe probability value of each sample index's item with dangerous probit.
Further, when the corresponding safe probability value of each sample index's item is more than dangerous probit, sample index's item it is pre-
Result is surveyed not produce collision accident;When the corresponding safe probability value of each sample index's item is less than dangerous probit, sample
Index item predict the outcome for produce collision accident.
Further, the actual collision result of every vehicle sample running data whether is obtained according to generation collision accident.
In order to those skilled in the art more understand said process, it is described as follows with reference to table 1:
Calculated 6 unit distances (kilometer) prediction knot is carried out using each index in table 1 as sample index's item
Fruit is as shown in table 2:
Table 2 predicts the outcome for the unit distance (kilometer) of each sample index's item
Further, predicting the outcome to generate with actual collision result and corresponding obscure square according to each sample index's item
Battle array, and the prediction accuracy of each sample index's item is calculated according to the data in corresponding confusion matrix.
Specifically, continue by taking E1 examples in above-mentioned table 2 as an example, being predicted the outcome, it is corresponding to generate with actual collision result
Confusion matrix, as shown in table 3:
Table 3 generate corresponding confusion matrix with actual collision result for predicting the outcome for index item E1
Further, obtain the prediction accuracy of multiple sample index's items plus and be worth, and by each sample index's item
Prediction accuracy with plus and be worth ratio value as each sample index's item weight.
Further, index item and corresponding weight are recorded to generate default weight table.
Specifically, continue, by taking above-mentioned E1 examples as an example, its prediction accuracy can be calculated by preset formula or algorithm.
For example, E1 prediction accuracies are calculated as (RsPs+RdPd)/(RsPs+RsPd+RdPs+RdPd) i.e. (2 with the data instance in table 3
+ 2)/(2+1+0+2)=0.8.
In the same manner, the prediction accuracy that E2, E3 wait until each index item of Ei can be obtained.Concrete calculating process may refer to
E1。
Further, the prediction accuracy of whole index item can be obtained and summation conduct is carried out and is added and value, each sample index
Prediction accuracy with plus and value ratio value as each sample index's item weight generating default weight table.
Thus, more sample index's item is combined during setting up default weight table, and combination predicts the outcome and actual
Collision result determines the prediction accuracy of each sample index's item, so that the power of sample index's item is determined finally according to prediction accuracy
Weight, further increases the reasonability and referring to property of scoring.
It should be noted that in order to ensure to analyze the accuracy of driver's driving behavior, needing to gather every predetermined period
Vehicle sample running data, and default weight table is updated according to vehicle sample running data.
Thus, it is possible to the corresponding weight of each index item is read from default weight table, according to its weight in step 103
The safety trend numerical value of each index item for obtaining is weighted, finally using the safety trend numerical value after weighted sum as
Treat the appraisal result of scoring driving behavior.
For example, the corresponding weight of vehicle operation data each index item that driving behavior to be scored is set be w1,
W2 ..., wi.Scope is [0,1].The appraisal result that driving behavior to be scored can be obtained is 100* [(w1* safety trend numbers
Value (E1))+(w2* safety trend numerical value (E2))+...+(wi* safety trend numerical value (Ei))].
In sum, itself asks the driving behavior methods of marking of embodiment, obtains the vehicle of driving behavior to be scored first
Running data, wherein, vehicle operation data includes:Multiple index item in vehicle traveling, and the corresponding index of each index item
Alert data, then counts to vehicle operation data, and determination produces vehicle operation data under driving behavior to be scored
Stroke fragment in, the meansigma methodss of alarm times of each index item in unit distance, then according to alarm times meansigma methodss
And default characteristic model, determine the safety trend numerical value of each index item in stroke fragment, finally from default weight table
Read the corresponding weight of each index item, and according to corresponding weight to stroke fragment in multiple index item safety trend
Numerical value is weighted the safety trend numerical value after summation as the appraisal result for treating scoring driving behavior.Due to reference to more
Index item, make appraisal result more accurate, improve scoring reasonability and referring to property, reduce the generation of vehicle accident,
Improve Consumer's Experience.
Fig. 2 is the flow chart of the driving behavior methods of marking according to the application another embodiment.
As shown in Fig. 2 the driving behavior methods of marking of the embodiment of the present application is comprised the following steps:
Step 201, obtains the vehicle operation data of driving behavior to be scored, wherein, vehicle operation data includes:Vehicle row
Multiple index item in sailing, and the corresponding index alert data of each index item.
Step 202, counts to vehicle operation data, and determines that vehicle is produced under driving behavior to be scored travels number
According to stroke fragment in, the meansigma methodss of alarm times of each index item in unit distance.
Step 203, according to alarm times meansigma methodss and default characteristic model, determines each index item in stroke fragment
Safety trend numerical value.
It should be noted that the description of step S201-S203 is corresponding with above-mentioned steps S101-S103, thus to step
The description of rapid S201-S203 will not be described here with reference to the description of above-mentioned steps S101-S103.
The safety trend numerical value and the first default value of each index item in stroke fragment are made product by step 204,
And using each product value as each index item scoring.
Specifically, the safety trend numerical value and the first default value of each index item can be made into product directly, and will
Each product value as each index item scoring as judged result.Wherein, the first default value can be according to practical application
Needs carry out selection setting.
Step 205, reads the corresponding weight of each index item from default weight table, and according to corresponding weight to stroke
The safety trend numerical value of the multiple index item in fragment is weighted summation.
It should be noted that the description of step S205 is corresponding with above-mentioned steps S104, thus to step S205 retouch
The description with reference to above-mentioned steps S104 is stated, be will not be described here.
Safety trend numerical value and the first default value after weighted sum is made product, and product value is made by step 206
For the scoring of driving behavior to be scored.
Whether step 207, judge the scoring of scoring and/or driving behavior to be scored of each index item more than or equal to the
Two default values, obtain the first judged result.
Specifically, the safety trend numerical value and the first default value after weighted sum can be made into product directly, and will
Scoring of the product value as driving behavior to be scored.By the scoring of each index item, or the scoring of driving behavior to be scored,
Or the scoring of the scoring and driving behavior to be scored of each index item comparison simultaneously with the second default value, and according to comparing
As a result obtain the first judged result.Wherein, the second default value can need to carry out selection setting according to practical application.
For example, the second default value is 50.The scoring of driving behavior to be scored is 80, then the first judged result is not for
Collision.The scoring of driving behavior to be scored is 30, then the first judged result is collision.
Step 208, is respectively adopted neural network model and Random Forest model treats the vehicle traveling of scoring driving behavior
Data are predicted, and obtain the second judged result based on neural network model and the 3rd judgement knot based on Random Forest model
Really.
Step 209, using voting mechanism, the first judged result, the second judged result, and the 3rd judged result to each
The scoring of the scoring of index item and/or driving behavior to be scored is adjusted.
Specifically, in order to further improve the accuracy of appraisal result, by neural network model and Random Forest model
Auxiliary judgment, the vehicle operation data for treating scoring driving behavior are predicted, and obtain sentencing based on the second of neural network model
Disconnected result and the 3rd judged result based on Random Forest model.
Wherein, the process that neutral net is predicted can be understood as initially setting up the neutral net that hidden layer is two-layer,
Every layer of neuronal quantity is i, and input layer quantity can subtract 1 for index item number amount in table 1, and output layer neuron number is
2.In the training stage, by data input neutral net in table 1, wherein collision index item data input output layer is used as classification results
(dependent variable), remaining index item input input layer (independent variable).Training can start to carry out run-length data to be assessed after terminating
Other indexs in addition to collision index are input into neural network input layer and obtain classification results as the second judgement knot by assessment
Really.
Wherein, the process that Random Forest model is predicted is appreciated that to be the tree number according to computing node performance to be adapted to
Amount, tree depth, barrelage amount set up random forest, in the training stage, by data input random forest in table 1, to collide index item
Used as classification results (dependent variable), remaining achievement data is independent variable to data, trains and can start to stroke to be assessed after terminating
Data are estimated, and other indexs in addition to collision index are input into random forest and classification results is obtained as the 3rd judgement knot
Really.
And then obtain three kinds of judged results, can by voting mechanism, by the first judged result and the second judged result, the
Three judged results collectively constitute 3 ballot papers, if wherein any two or whole ballot paper result are collision, final vote result
For collision, if wherein any two or whole ballot paper result are non-collision, final vote result is not to collide.
Further, by above-mentioned final vote result to the scoring and/or driving behavior to be scored of each index item
Scoring is adjusted.Concrete analysis is as follows:
The first example, at least two judged results are not produce collision accident/generation collision accident;First judged result
Not produce collision accident/generation collision accident.
Specifically, in the first judged result, the second judged result, and the 3rd judged result, at least two judge knot
Fruit is not to produce collision accident/generation collision accident, and the first judged result is not produce collision accident/generation collision accident
When, the not scoring of the scoring to each index item and/or the driving behavior to be scored is adjusted.
It is understood that when the first judged result is consistent with final vote result, it is not necessary to be adjusted.Represent mesh
Front appraisal result is reliable.
Second example, at least two judged results are not produce collision accident/generation collision accident;First judged result
To produce collision accident/do not produce collision accident.
Specifically, in the first judged result, the second judged result, and the 3rd judged result, at least two judge knot
Fruit is not to produce collision accident/generations collision accident, and the first judged result is generation collision accident/do not produce collision accident,
The scoring of scoring and/or the driving behavior to be scored to each index item is adjusted.
It is understood that in the first judged result and inconsistent final vote result, needing to be adjusted.Adjustment
Mode has many kinds.For example, at least two judged results are generation collision accident;First judged result is not produce collision thing
Part.Final total score in using R as stroke fragment, wherein R are the double precision random number more than 0 and less than 1.If D, D represent first
The difference of judged result and final total score.The corresponding weight of each index item that D is multiplied by step 104 obtains each index item
The scoring of each index item is deducted corresponding difference, to ensure the final scoring of each index item by the score value that scoring should be deducted
It is consistent with final total score after addition.
In sum, itself asks the driving behavior methods of marking of embodiment, further by neutral net mould is respectively adopted
Type and Random Forest model are treated the vehicle operation data of scoring driving behavior and are predicted, and the knot obtained using voting mechanism
Scoring of the fruit to the scoring and/or driving behavior to be scored of each index item is adjusted.By way of adjustment, tie scoring
Fruit more precisely, improves the reasonability and referring to property of scoring, and causes scoring knot more objective, be easy to analyze driver's weakness
Driving behavior.
In order to realize above-described embodiment, the application also proposed a kind of driving behavior scoring apparatus.
Fig. 3 is the structural representation of the driving behavior scoring apparatus according to the application one embodiment.
As shown in figure 3, the driving behavior scoring apparatus include:First acquisition module 31, first processing module 32, determine mould
Block 33, Second processing module 34 and grading module 35.
Wherein, the first acquisition module 31 is for obtaining the vehicle operation data of driving behavior to be scored, wherein, vehicle traveling
Data include:Multiple index item in vehicle traveling, and the corresponding index alert data of each index item.
First processing module 32 is for counting to vehicle operation data, and determines the generation under driving behavior to be scored
In the stroke fragment of vehicle operation data, the meansigma methodss of alarm times of each index item in unit distance.
Determining module 33 for according to alarm times meansigma methodss and default characteristic model, determine in stroke fragment each refer to
The safety trend numerical value of mark item.
Second processing module 34 for the corresponding weight of each index item is read from default weight table, and according to corresponding
Weight to stroke fragment in the safety trend numerical value of multiple index item be weighted summation.
Grading module 35 for using the safety trend numerical value after weighted sum as treat scoring driving behavior scoring
As a result.
Wherein, vehicle operation data can be including the multiple index item in vehicle traveling and the corresponding index of each index item
Alert data.The type of index item has many kinds, can need to select according to practical application.
Specifically, the vehicle operation data of acquisition is carried out counting the alarm times for obtaining each index item divided by whole row
The mileage number (kilometer, rice etc. can be made to be unit) of journey fragment, such that it is able to obtain report of each index item in unit distance
The meansigma methodss of alert number of times.
In particular it is required that setting up default characteristic model.As shown in figure 4, on the basis of Fig. 3, driving behavior scoring apparatus
Also include:Second acquisition module 36, division module 37, the 3rd processing module 38, fourth processing module 39, the 5th processing module
310th, the 6th processing module 311, the 7th processing module 312 and the 8th processing module 313.
Wherein, the second acquisition module 36 is used to obtain a plurality of vehicle sample running data of multiple sample stroke fragments, its
In, every vehicle sample running data in a plurality of vehicle sample running data includes:Multiple sample index in vehicle traveling
, and the corresponding sample index's alert data of each sample index's item.
Division module 37, is not produced according to producing whether collision accident divides for multiple sample stroke fragments
Second sample stroke fragment of the first sample stroke fragment and generation collision accident of raw collision accident.
3rd processing module 38 for counting to every vehicle sample running data of first sample stroke fragment, and
In determining each first sample stroke fragment, each sample index's item produces the first warning of alarming index item in unit distance
Number of times meansigma methodss.
Fourth processing module 39 for counting to every vehicle sample running data of the second sample stroke fragment, and
In determining each second sample stroke fragment, each sample index's item produces the second warning of alarming index item in unit distance
Number of times meansigma methodss.
5th processing module 310 is used to calculate the mark of the corresponding multiple first alarm times meansigma methodss of each sample index's item
It is accurate poor, and using standard deviation as in the first sample stroke fragment for not producing collision accident each sample index's item corresponding first
Standard deviation.
6th processing module 311 is used to calculate the mark of the corresponding multiple second alarm times meansigma methodss of each sample index's item
It is accurate poor, and using standard deviation as corresponding second mark of each sample index's item in the second sample stroke fragment for producing collision accident
It is accurate poor.
7th processing module 312 is for carrying out being averaging computing to multiple first alarm times meansigma methodss, and averagely will transport
The numerical value for obtaining is corresponding first flat as each sample index's item in the first sample stroke fragment of collision accident is not produced
Average.
8th processing module 313 is for carrying out being averaging computing to multiple second alarm times meansigma methodss, and averagely will transport
The numerical value for obtaining is corresponding second average as each sample index's item in the second sample stroke fragment of collision accident is produced
Value is setting up default characteristic model.
In particular it is required that building default weight table.As shown in figure 5, on the basis of Fig. 3, driving behavior scoring apparatus are also
Including:Computing module 314, the 3rd acquisition module 315, the first prediction module 316, the 9th processing module 317 and generation module
318。
Wherein, computing module 314 is used for the prediction knot of each the sample index's item for calculating every vehicle sample running data
Really.
3rd acquisition module 315 is for according to producing whether collision accident obtains the reality of every vehicle sample running data
Collision result.
First prediction module 316 is for corresponding with the generation of actual collision result according to predicting the outcome for each sample index's item
Confusion matrix, and the prediction accuracy of each sample index's item is calculated according to the data in corresponding confusion matrix.
9th processing module 317 be used to obtaining the prediction accuracy of multiple sample index's items plus and value, and by each institute
State the prediction accuracy of sample index's item with plus and value ratio value as each sample index's item weight.
Generation module 318 is for being recorded to generate default weight table by index item and corresponding weight.
In one embodiment of the application, computing module 314 includes:First computing unit 3141 and the second computing unit
3142nd, the first comparing unit 3143 and the second comparing unit 3144.
Wherein, the first computing unit 3141 is for the alarm times meansigma methodss according to each sample index's item, default feature
The corresponding safe probability value of the first standard deviation and the first mean value calculation each sample index's item in model.
Second computing unit 3142 is in the alarm times meansigma methodss according to each sample index's item, default characteristic model
The second standard deviation and the corresponding dangerous probit of each sample index's item of the second mean value calculation.
First comparing unit 3143 for when the corresponding safe probability value of each sample index's item is more than dangerous probit,
Sample index's item predicts the outcome not produce collision accident.
Second comparing unit 3144 for when the corresponding safe probability value of each sample index's item is less than dangerous probit,
Sample index's item predict the outcome for produce collision accident.
In one embodiment of the application, as shown in fig. 6, on the basis of Fig. 3, determining module 33 includes:3rd meter
Calculate unit 331, the 4th computing unit 332 and determining unit 333.
Wherein, the 3rd computing unit 331 is in the alarm times meansigma methodss according to each index item, default characteristic model
The first standard deviation and the corresponding safe probability value of each index item described in the first mean value calculation.
4th computing unit 332 for the alarm times meansigma methodss according to each index item, default characteristic model in the
Two standard deviations and the corresponding dangerous probit of each index item of the second mean value calculation.
Determining unit 333 is for determining stroke fragment according to the corresponding safe probability value of each index item and dangerous probit
In each index item safety trend numerical value.
Further the vehicle row of scoring driving behavior is treated by neural network model and Random Forest model is respectively adopted
Sail data to be predicted, and scoring and/or to be scored driving behavior of the result obtained using voting mechanism to each index item
Scoring be adjusted.By way of adjustment, the accuracy of the appraisal result of driving behavior is further improved so that scoring knot
It is more objective, it is easy to analyze driver's weakness driving behavior.As shown in fig. 7, on the basis of Fig. 3, driving behavior scoring apparatus
Also include:
First product module 319, the second product module 320, the first judge module 321, the second prediction module 322 and adjustment
Module 323.
Wherein, the first product module 319 is for by the safety trend numerical value of each index item in stroke fragment and
One default value makees product, and using each product value as each index item scoring.
Second product module 320 for the safety trend numerical value and the first default value after weighted sum is made product,
And using product value as driving behavior to be scored scoring.
Whether first judge module 321 is used for the scoring of the scoring and/or driving behavior to be scored for judging each index item
More than or equal to the second default value, the first judged result is obtained.
Second prediction module 322 is used to neural network model to be respectively adopted and Random Forest model treats scoring driving behavior
Vehicle operation data be predicted, obtain the second judged result based on neural network model and based on Random Forest model
3rd judged result.
Adjusting module 323 is used for using voting mechanism, the first judged result, the second judged result, and the 3rd judges knot
Scoring of the fruit to the scoring and/or driving behavior to be scored of each index item is adjusted.
In order to further improve the accuracy of analysis driving behavior, also including acquisition module 324 and update module 325.
Acquisition module 324 is for every predetermined period collection vehicle sample running data.
Update module 325 is for carrying out more to default characteristic model and default weight table according to vehicle sample running data
Newly.
In one embodiment of the application, adjusting module 323 includes:First adjustment unit 3231 and the second adjustment unit
3232。
First adjustment unit 3231 in the first judged result, the second judged result, and the 3rd judged result, extremely
Few two judged results are not to produce collision accident/generation collision accident, and the first judged result is not produce collision accident/product
During raw collision accident, the not scoring of the scoring to each index item and/or driving behavior to be scored is adjusted;
Second adjustment unit 3232 in the first judged result, the second judged result, and the 3rd judged result, extremely
Few two judged results are not to produce collision accident/generations collision accident, and the first judged result for generation collision accident/do not produce
During raw collision accident, the scoring of scoring and/or driving behavior to be scored to each index item is adjusted.
The driving behavior scoring that driving behavior scoring apparatus provided in an embodiment of the present invention are provided with above-mentioned several embodiments
Method is corresponding, therefore the embodiment in aforementioned driving behavior methods of marking is also applied for the driving behavior that the present embodiment is provided
Scoring apparatus, are not described in detail in the present embodiment.
In sum, itself asks the driving behavior scoring apparatus of embodiment, obtains the vehicle of driving behavior to be scored first
Running data, wherein, vehicle operation data includes:Multiple index item in vehicle traveling, and the corresponding index of each index item
Alert data, then counts to vehicle operation data, and determination produces vehicle operation data under driving behavior to be scored
Stroke fragment in, the meansigma methodss of alarm times of each index item in unit distance, then according to alarm times meansigma methodss
And default characteristic model, determine the safety trend numerical value of each index item in stroke fragment, finally from default weight table
Read the corresponding weight of each index item, and according to corresponding weight to stroke fragment in multiple index item safety trend
Numerical value is weighted the safety trend numerical value after summation as the appraisal result for treating scoring driving behavior.Due to reference to more
Index item, make appraisal result more accurate, improve scoring reasonability and referring to property, reduce the generation of vehicle accident,
Improve Consumer's Experience.
It should be noted that in the description of the present application, term " first ", " second " etc. are only used for describing purpose, and not
It is understood that to indicate or implying relative importance.Additionally, in the description of the present application, unless otherwise stated, the implication of " multiple "
It is two or more.
In flow chart or here any process described otherwise above or method description are construed as, expression includes
It is one or more for realizing specific logical function or process the step of the module of code of executable instruction, fragment or portion
Point, and the scope of the preferred implementation of the application includes other realization, wherein the suitable of shown or discussion can not be pressed
Sequence, including according to involved function by it is basic simultaneously in the way of or in the opposite order, carry out perform function, this should be by the application
Embodiment person of ordinary skill in the field understood.
It should be appreciated that each several part of the application can be realized with hardware, software, firmware or combinations thereof.Above-mentioned
In embodiment, the software that multiple steps or method can be performed in memory and by suitable instruction execution system with storage
Or firmware is realizing.For example, if realized with hardware, and in another embodiment, can be with well known in the art
Any one of row technology or their combination are realizing:With for the logic gates of logic function is realized to data signal
Discrete logic, the special IC with suitable combinational logic gate circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
Those skilled in the art are appreciated that to realize all or part of step that above-described embodiment method is carried
Suddenly the hardware that can be by program to instruct correlation is completed, and described program can be stored in a kind of computer-readable storage medium
In matter, the program upon execution, including one or a combination set of the step of embodiment of the method.
Additionally, each functional unit in the application each embodiment can be integrated in a processing module, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a module.Above-mentioned integrated mould
Block both can be realized in the form of hardware, it would however also be possible to employ the form of software function module is realized.The integrated module is such as
Fruit using in the form of software function module realize and as independent production marketing or use when, it is also possible to be stored in a computer
In read/write memory medium.
Storage medium mentioned above can be read only memory, disk or CD etc..
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
Example ", or the description of " some examples " etc. mean specific features with reference to the embodiment or example description, structure, material or spy
Point is contained at least one embodiment or example of the application.In this manual, to the schematic representation of above-mentioned term not
Identical embodiment or example are referred to necessarily.And, the specific features of description, structure, material or feature can be any
One or more embodiments or example in combine in an appropriate manner.
Although embodiments herein has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is impossible to the restriction to the application is interpreted as, one of ordinary skill in the art within the scope of application can be to above-mentioned
Embodiment is changed, changes, replacing and modification.
Claims (10)
1. a kind of driving behavior methods of marking, it is characterised in that comprise the following steps:
The vehicle operation data of driving behavior to be scored is obtained, wherein, the vehicle operation data includes:It is many in vehicle traveling
Individual index item, and the corresponding index alert data of each index item;
The vehicle operation data is counted, and determines that the vehicle is produced under the driving behavior to be scored travels number
According to stroke fragment in, the meansigma methodss of alarm times of each index item in unit distance;
According to the alarm times meansigma methodss and default characteristic model, the safety of each index item in the stroke fragment is determined
Tendentiousness numerical value;
The corresponding weight of each index item described in reading from default weight table, and according to the corresponding weight to the stroke
The safety trend numerical value of the multiple index item in fragment is weighted summation;
Using the safety trend numerical value after weighted sum as the appraisal result to the driving behavior to be scored.
2. driving behavior methods of marking as claimed in claim 1, it is characterised in that set up the default spy by following steps
Model is levied, including:
The a plurality of vehicle sample running data of multiple sample stroke fragments is obtained, wherein, a plurality of vehicle sample running data
In every vehicle sample running data include:Multiple sample index's items in vehicle traveling, it is corresponding with each sample index's item
Sample index's alert data;
To the plurality of sample stroke fragment according to producing whether collision accident divides, the of collision accident is not produced
Second sample stroke fragment of the same this stroke fragment and generation collision accident;
Every vehicle sample running data of the first sample stroke fragment is counted, and each first sample described in determining
In this stroke fragment, each sample index's item produces the first alarm times meansigma methodss of alarming index item in unit distance;
Every vehicle sample running data of the second sample stroke fragment is counted, and each second sample described in determining
In this stroke fragment, each sample index's item produces the second alarm times meansigma methodss of alarming index item in unit distance;
Each sample index's item corresponding multiple first alarm times standard errors of the mean described in calculating, and by the standard deviation
As not producing corresponding first standard deviation of each sample index's item in the first sample stroke fragment of collision accident;
Each sample index's item corresponding multiple second alarm times standard errors of the mean described in calculating, and by the standard deviation
As corresponding second standard deviation of each sample index's item in the second sample stroke fragment of generation collision accident;
The plurality of first alarm times meansigma methodss are carried out being averaging with computing, and the numerical value that average calculating operation is obtained is not used as producing
Corresponding first meansigma methodss of each sample index's item in the first sample stroke fragment of raw collision accident;
The plurality of second alarm times meansigma methodss are carried out being averaging with computing, and the numerical value that average calculating operation is obtained as generation
In second sample stroke fragment of collision accident, corresponding second meansigma methodss of each sample index's item are setting up the default feature
Model.
3. driving behavior methods of marking as claimed in claim 2, it is characterised in that set up the default power by following steps
Weight table, including:
Each the sample index's item for calculating every vehicle sample running data predicts the outcome;
Whether the actual collision result of the every vehicle sample running data is obtained according to generation collision accident;
Corresponding confusion matrix is generated with actual collision result according to predicting the outcome for each sample index's item, and according to institute
State the prediction accuracy that the data in corresponding confusion matrix calculate each sample index's item;
Obtain the prediction accuracy of multiple sample index's items plus and value, and by the prediction accuracy of each sample index's item
With it is described plus and be worth ratio value as sample index's item each described weight;
The index item and corresponding weight are recorded to generate the default weight table.
4. driving behavior methods of marking as claimed in claim 3, it is characterised in that the calculating every vehicle sample row
Each the sample index's item for sailing data predicts the outcome, including:
The first standard deviation in alarm times meansigma methodss, the default characteristic model and institute according to each sample index's item
State the corresponding safe probability value of each sample index's item described in the first mean value calculation;
The second standard deviation in alarm times meansigma methodss, the default characteristic model and institute according to each sample index's item
State the corresponding dangerous probit of each sample index's item described in the second mean value calculation;
When the corresponding safe probability value of described each sample index's item is more than dangerous probit, the prediction of sample index's item
As a result it is not produce collision accident;
When the corresponding safe probability value of described each sample index's item is less than dangerous probit, the prediction of sample index's item
As a result it is generation collision accident.
5. driving behavior methods of marking as claimed in claim 2, it is characterised in that described according to the alarm times meansigma methodss
And default characteristic model, determine the safety trend numerical value of each index item in the stroke fragment, including:
The first standard deviation in alarm times meansigma methodss, the default characteristic model according to each index item and described
The corresponding safe probability value of each index item described in one mean value calculation;
The second standard deviation in alarm times meansigma methodss, the default characteristic model according to each index item and described
The corresponding dangerous probit of each index item described in two mean value calculation;
Determined in the stroke fragment according to the corresponding safe probability value of described each index item and the dangerous probit
Each index item safety trend numerical value.
6. driving behavior methods of marking as claimed in claim 2, it is characterised in that every in the determination stroke fragment
After the safety trend numerical value of individual index item, also include:
The safety trend numerical value and the first default value of each index item in the stroke fragment are made into product, and by each
Scoring of the product value as each index item.
7. driving behavior methods of marking as claimed in claim 6, it is characterised in that incline in the safety by after weighted sum
After tropism numerical value is as the appraisal result to the driving behavior to be scored, also include;
Safety trend numerical value after the weighted sum is made into product with first default value, and using product value as institute
State the scoring of driving behavior to be scored.
8. driving behavior methods of marking as claimed in claims 6 or 7, it is characterised in that also include:
Whether the scoring of the scoring and/or the driving behavior to be scored of each index item described in judging is pre- more than or equal to second
If numerical value, the first judged result is obtained;
Neural network model and Random Forest model is respectively adopted to be carried out to the vehicle operation data of the driving behavior to be scored
Prediction, obtains the second judged result based on the neural network model and the 3rd judgement knot based on the Random Forest model
Really;
Using voting mechanism, first judged result, second judged result, and the 3rd judged result to described
The scoring of the scoring of each index item and/or the driving behavior to be scored is adjusted.
9. driving behavior methods of marking as claimed in claim 8, it is characterised in that the employing voting mechanism, described first
Judged result, second judged result, and the 3rd judged result is to the scoring of each index item and/or described
The scoring of driving behavior to be scored is adjusted, including:
In first judged result, second judged result, and the 3rd judged result, at least two judge knot
Fruit collides thing not produce collision accident/generation not produce collision accident/generation collision accident, and first judged result
During part, scoring and/or the scoring of the driving behavior to be scored not to each index item are adjusted;
In first judged result, second judged result, and the 3rd judged result, at least two judge knot
Fruit is not to produce collision accident/generations collision accident, and first judged result is generation collision accident/do not produce collision thing
During part, the scoring of scoring and/or the driving behavior to be scored to each index item is adjusted.
10. a kind of driving behavior scoring apparatus, it is characterised in that include:
First acquisition module, for obtaining the vehicle operation data of driving behavior to be scored, wherein, the vehicle operation data bag
Include:Multiple index item in vehicle traveling, and the corresponding index alert data of each index item;
First processing module, for counting to the vehicle operation data, and determines under the driving behavior to be scored
Produce in the stroke fragment of the vehicle operation data, the meansigma methodss of alarm times of each index item in unit distance;
Determining module, it is every in the stroke fragment for according to the alarm times meansigma methodss and default characteristic model, determining
The safety trend numerical value of individual index item;
Second processing module, for the corresponding weight of described each index item is read from default weight table, and according to described right
The weight answered is weighted summation to the safety trend numerical value of the multiple index item in the stroke fragment;
Grading module, for tying the safety trend numerical value after weighted sum as the scoring to the driving behavior to be scored
Really.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611061545.9A CN106557663A (en) | 2016-11-25 | 2016-11-25 | Driving behavior methods of marking and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611061545.9A CN106557663A (en) | 2016-11-25 | 2016-11-25 | Driving behavior methods of marking and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106557663A true CN106557663A (en) | 2017-04-05 |
Family
ID=58445069
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611061545.9A Pending CN106557663A (en) | 2016-11-25 | 2016-11-25 | Driving behavior methods of marking and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106557663A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109559245A (en) * | 2017-09-26 | 2019-04-02 | 北京国双科技有限公司 | A kind of method and device identifying specific user |
CN109649386A (en) * | 2018-12-26 | 2019-04-19 | 成都路行通信息技术有限公司 | A kind of vehicle collision detection method based on machine learning |
CN110288221A (en) * | 2019-06-19 | 2019-09-27 | 高新兴物联科技有限公司 | A kind of driving behavior points-scoring system and methods of marking |
CN111242484A (en) * | 2020-01-14 | 2020-06-05 | 北京车汇天下科技有限公司 | Vehicle risk comprehensive evaluation method based on transition probability |
CN111833480A (en) * | 2019-04-12 | 2020-10-27 | 比亚迪股份有限公司 | Driving behavior detection method and device and vehicle |
CN112183984A (en) * | 2020-09-21 | 2021-01-05 | 长城汽车股份有限公司 | Driving behavior processing method and device, storage medium and electronic equipment |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101937421A (en) * | 2009-07-03 | 2011-01-05 | 上海大潮电子技术有限公司 | Method for collecting real-time operation information of vehicle for operation security risk assessment |
CN102549628A (en) * | 2009-09-18 | 2012-07-04 | 丰田自动车株式会社 | Driving evaluation system, vehicle-mounted machine, and information processing center |
GB2489558A (en) * | 2011-03-30 | 2012-10-03 | Gm Global Tech Operations Inc | Evaluating driver behavior by classifying pairs of speed and acceleration into quality ranges |
CN102874188A (en) * | 2012-09-01 | 2013-01-16 | 北京车网互联科技股份有限公司 | Driving behavior warning method based on vehicle bus data |
CN103198685A (en) * | 2013-03-15 | 2013-07-10 | Tcl集团股份有限公司 | Method and system for achieving driving safety early warning |
CN103810904A (en) * | 2014-03-12 | 2014-05-21 | 哈尔滨工业大学 | VANET-based early-warning system and early-warning method for driving safety in expressway |
CN103871122A (en) * | 2014-03-11 | 2014-06-18 | 深圳市朗仁科技有限公司 | Driving behavior analysis method and driving behavior analysis system |
CN104092736A (en) * | 2014-06-25 | 2014-10-08 | 国信彩石(北京)科技股份有限公司 | Vehicle networking device, server and system, scoring method and data collection method |
CN104268701A (en) * | 2014-09-29 | 2015-01-07 | 清华大学 | Commercial vehicle driving safety evaluation system and method |
US20150112546A1 (en) * | 2013-10-23 | 2015-04-23 | Trimble Navigation Limited | Driver scorecard system and method |
-
2016
- 2016-11-25 CN CN201611061545.9A patent/CN106557663A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101937421A (en) * | 2009-07-03 | 2011-01-05 | 上海大潮电子技术有限公司 | Method for collecting real-time operation information of vehicle for operation security risk assessment |
CN102549628A (en) * | 2009-09-18 | 2012-07-04 | 丰田自动车株式会社 | Driving evaluation system, vehicle-mounted machine, and information processing center |
GB2489558A (en) * | 2011-03-30 | 2012-10-03 | Gm Global Tech Operations Inc | Evaluating driver behavior by classifying pairs of speed and acceleration into quality ranges |
CN102874188A (en) * | 2012-09-01 | 2013-01-16 | 北京车网互联科技股份有限公司 | Driving behavior warning method based on vehicle bus data |
CN103198685A (en) * | 2013-03-15 | 2013-07-10 | Tcl集团股份有限公司 | Method and system for achieving driving safety early warning |
US20150112546A1 (en) * | 2013-10-23 | 2015-04-23 | Trimble Navigation Limited | Driver scorecard system and method |
CN103871122A (en) * | 2014-03-11 | 2014-06-18 | 深圳市朗仁科技有限公司 | Driving behavior analysis method and driving behavior analysis system |
CN103810904A (en) * | 2014-03-12 | 2014-05-21 | 哈尔滨工业大学 | VANET-based early-warning system and early-warning method for driving safety in expressway |
CN104092736A (en) * | 2014-06-25 | 2014-10-08 | 国信彩石(北京)科技股份有限公司 | Vehicle networking device, server and system, scoring method and data collection method |
CN104268701A (en) * | 2014-09-29 | 2015-01-07 | 清华大学 | Commercial vehicle driving safety evaluation system and method |
Non-Patent Citations (2)
Title |
---|
LIANG JUN等: ""Multi-agent and driving behavior based rear-end collision alarm modeling and simulating"", 《SIMULATION MODELLING PRACTICE AND THEORY》 * |
咸化彩: ""次任务驾驶安全性评价指标及评价模型研究"", 《中国博士学位论文全文数据库 工程科技II辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109559245A (en) * | 2017-09-26 | 2019-04-02 | 北京国双科技有限公司 | A kind of method and device identifying specific user |
CN109559245B (en) * | 2017-09-26 | 2022-02-25 | 北京国双科技有限公司 | Method and device for identifying specific user |
CN109649386A (en) * | 2018-12-26 | 2019-04-19 | 成都路行通信息技术有限公司 | A kind of vehicle collision detection method based on machine learning |
CN111833480A (en) * | 2019-04-12 | 2020-10-27 | 比亚迪股份有限公司 | Driving behavior detection method and device and vehicle |
CN110288221A (en) * | 2019-06-19 | 2019-09-27 | 高新兴物联科技有限公司 | A kind of driving behavior points-scoring system and methods of marking |
CN111242484A (en) * | 2020-01-14 | 2020-06-05 | 北京车汇天下科技有限公司 | Vehicle risk comprehensive evaluation method based on transition probability |
CN111242484B (en) * | 2020-01-14 | 2023-09-19 | 沈阳平通高新技术有限公司 | Vehicle risk comprehensive evaluation method based on transition probability |
CN112183984A (en) * | 2020-09-21 | 2021-01-05 | 长城汽车股份有限公司 | Driving behavior processing method and device, storage medium and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106557663A (en) | Driving behavior methods of marking and device | |
CN106585635B (en) | Driving behavior methods of marking and device | |
CN106781454B (en) | The appraisal procedure and device of driving behavior | |
Feng et al. | Can vehicle longitudinal jerk be used to identify aggressive drivers? An examination using naturalistic driving data | |
CN106240571B (en) | Driving behavior analysis method and apparatus | |
Das et al. | Automated vehicle collisions in California: Applying Bayesian latent class model | |
CN103562978B (en) | Vehicle data analysis method and vehicle data analysis system | |
KR102353705B1 (en) | System and method for analyzing the energy efficiency of a motor vehicle, in particular of an apparatus of the motor vehicle | |
CN107066787B (en) | The methods of marking and device of vehicle travel | |
Zhao et al. | Identification of driver’s braking intention based on a hybrid model of GHMM and GGAP-RBFNN | |
CN106740864A (en) | A kind of driving behavior is intended to judge and Forecasting Methodology | |
Fugiglando et al. | Characterizing the" driver dna" through can bus data analysis | |
CN106777907A (en) | Driving behavior methods of marking and device | |
CN103531042A (en) | Rear-end collision pre-warning method based on driver types | |
Birrell et al. | Analysis of three independent real-world driving studies: A data driven and expert analysis approach to determining parameters affecting fuel economy | |
CN111785023A (en) | Vehicle collision risk early warning method and system | |
CN107291972A (en) | The Intelligent Vehicle Driving System efficiency evaluation method excavated based on multi-source data | |
CN103150677A (en) | Aggressive driving state identification method and system | |
Peng et al. | Intelligent method for identifying driving risk based on V2V multisource big data | |
KR20170115831A (en) | Method and apparatus for guiding automobile insurance using driver recognizing | |
Yang et al. | Driver2vec: Driver identification from automotive data | |
CN105329239A (en) | Aggressive driving behavior detection and processing | |
Yardy et al. | Detecting malicious driving with machine learning | |
Reddy | Driving behaviour classification: An eco-driving approach | |
CN115771506A (en) | Method and device for determining vehicle driving strategy based on passenger risk cognition |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170405 |
|
RJ01 | Rejection of invention patent application after publication |