CN109572706A - A kind of driving safety evaluation method and device - Google Patents
A kind of driving safety evaluation method and device Download PDFInfo
- Publication number
- CN109572706A CN109572706A CN201811518029.3A CN201811518029A CN109572706A CN 109572706 A CN109572706 A CN 109572706A CN 201811518029 A CN201811518029 A CN 201811518029A CN 109572706 A CN109572706 A CN 109572706A
- Authority
- CN
- China
- Prior art keywords
- layer
- network
- output
- data
- node
- 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.)
- Granted
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 59
- 238000013528 artificial neural network Methods 0.000 claims abstract description 57
- 238000012360 testing method Methods 0.000 claims abstract description 43
- 238000012549 training Methods 0.000 claims abstract description 39
- 239000011159 matrix material Substances 0.000 claims abstract description 38
- 238000000034 method Methods 0.000 claims abstract description 37
- 238000013210 evaluation model Methods 0.000 claims abstract description 10
- 230000006399 behavior Effects 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 6
- 210000004218 nerve net Anatomy 0.000 claims description 3
- 230000006870 function Effects 0.000 description 21
- 238000010586 diagram Methods 0.000 description 20
- 230000001133 acceleration Effects 0.000 description 17
- 230000008569 process Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 230000001537 neural effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000001105 regulatory effect Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000011157 data evaluation Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005194 fractionation Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
- B60W40/09—Driving style or behaviour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Transportation (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Mechanical Engineering (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Automation & Control Theory (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of driving safety evaluation method and devices, are related to road traffic technical field.It is imperfect that there are factors of evaluation solving existing driving safety evaluation method, and while evaluating is influenced by subjective factor, so as to cause the not high problem of evaluation result accuracy.This method comprises: constructing specific assessment indicator system and determining multiple indexs in assessment indicator system, the corresponding multi-group data of multiple indexs is obtained from vehicle, obtains the subordinated-degree matrix of index；Using the subordinated-degree matrix for the multiple indexs for including in training data group as the input layer of the training network of BP neural network, when determining that the output error of output layer of training network is less than convergency value, using the subordinated-degree matrix for the multiple indexs for including in test data set as the input layer of the test network of BP neural network, when determining that the accuracy of output layer of test network is greater than the set value, BP neural network is determined as driving safety evaluation model.
Description
Technical field
The present invention relates to road traffic technical field, a kind of driving safety evaluation method and device are more particularly related to.
Background technique
Status data when driving safety overall merit refers in conjunction with driving behavior information and automobilism, with quantitative
The method of analysis carries out comprehensive analysis and assessment to the driving behavior of driver.It is directed at present and driving safety overall merit is asked
Topic, main method are that multisensor is combined to carry out data acquisition, with analytic hierarchy process (AHP), Field Using Fuzzy Comprehensive Assessment, cluster and spy
The methods of value indicative analysis processing data simultaneously obtain evaluation result.
At present part of data acquisition mainly with technology there are CAN bus based data to acquire, based on video information
Data acquisition etc., these methods are there are data source detection project is single, the defects of higher cost, not can guarantee the complete of factor of evaluation
The accuracy of whole property and evaluation result；And data processing section mainly with technology have levels analytic approach and fuzzy overall evaluation
Comprehensive analysis, clustering that method combines etc., that there are models is relatively easy for these methods, subjective factor influences stronger etc. lack
It falls into, not can guarantee the science and accuracy of evaluation result.
In conclusion that there are factors of evaluation is imperfect for existing driving safety evaluation method, and by subjective factor when evaluating
Influence, so as to cause the not high problem of evaluation result accuracy.
Summary of the invention
The embodiment of the present invention provides a kind of driving safety evaluation method and device, to solve existing driving safety evaluation
That there are factors of evaluation is imperfect for method, and while evaluating is influenced by subjective factor, so as to cause evaluation result accuracy it is not high
Problem.
The embodiment of the present invention provides a kind of driving safety evaluation method, comprising:
According to global driving behavior safety evaluatio main indicator, with reference to assessment indicator system and retrievable data, structure
It builds specific assessment indicator system and determines the multiple indexs for including in the assessment indicator system；
The corresponding multi-group data of multiple indexs is obtained from vehicle, the multiple indexs that will include in every group of data
It is normalized and by degree of membership formula, obtains the subordinated-degree matrix for multiple indexs that every group of data include；
Using the subordinated-degree matrix for the multiple indexs for including in training data group as the training network of BP neural network
Input layer, determine the reality output and output error of the output layer of the corresponding trained network of the input layer of each trained network,
It is multiple described by include in test data set when determining that the output error of output layer of the trained network is less than convergency value
Input layer of the subordinated-degree matrix of index as the test network of BP neural network, it is correct when the output layer for determining test network
When rate is greater than the set value, the BP neural network is determined as driving safety evaluation model.
Preferably, described using the subordinated-degree matrix for the multiple indexs for including in training data group as BP neural network
Training network input layer before, comprising:
The hidden layer node number of the BP neural network is set according to following equation:
Wherein, h is hidden layer node number, and m is input layer number, and n output layer interstitial content, a is between 1 to 10
Regulating constant.
Preferably, the reality of the output layer of the corresponding trained network of input layer of each trained network is determined by following equation
Border output:
xj=f (Sj)
The output error of the output layer of the trained network is determined by following equation:
The hidden layer and the output layer BP neural network connection weight and threshold value are adjusted by following equation:
By following equation between the input layer and the hidden layer weight and threshold values be adjusted:
Wherein, xjFor the output valve of node j, f is activation primitive,wijIt is node i with node j's
Connection weight, bjFor the threshold values of node j, xiFor the output of upper layer node i, j-th of neuron of output layer is denoted as yj, djFor output
Layer it is all as a result, η1, η2For learning rate.
Preferably, the accuracy of the output layer of the test network is determined by following equation:
Wherein, y 'iFor the reality of the output layer of the corresponding test network of input layer of each test network of test data set
Output, yiFor source data result.
Preferably, the index is the multiple first class index for including in the assessment indicator system or the index is
The multiple two-level index for including in the assessment indicator system.
Preferably, the subordinated-degree matrix is as follows:
Wherein, Ri=(ri1,ri2,…,rij), R represents multifactor evaluation subordinated-degree matrix, RiSingle factor evaluation is represented to be subordinate to
It spends vector and i-th of evaluation index corresponds to the degree of membership of each opinion rating, rijI-th of index is represented to comment relative to jth layer
The degree of membership of valence grade.
The embodiment of the invention also provides a kind of driving safety evaluating apparatus, comprising:
First determination unit is used for according to global driving behavior safety evaluatio main indicator, with reference to assessment indicator system
With retrievable data, constructs specific assessment indicator system and determine the multiple indexs for including in the assessment indicator system；
Unit is obtained, for obtaining the corresponding multi-group data of multiple indexs from vehicle, will include in every group of data
Multiple indexs be normalized and by degree of membership formula, obtain multiple indexs that every group of data include
Subordinated-degree matrix；
Second determination unit, for using the subordinated-degree matrix for the multiple indexs for including in training data group as BP mind
The input layer of training network through network determines the reality of the output layer of the corresponding trained network of the input layer of each trained network
Output and output error, when determining that the output error of output layer of the trained network is less than convergency value, by test data set
In include multiple indexs test network of the subordinated-degree matrix as BP neural network input layer, when determining Test Network
When the accuracy of the output layer of network is greater than the set value, the BP neural network is determined as driving safety evaluation model.
Preferably, second determination unit is also used to:
The hidden layer node number of the BP neural network is set according to following equation:
Wherein, h is hidden layer node number, and m is input layer number, and n output layer interstitial content, a is between 1 to 10
Regulating constant.
Preferably, the reality of the output layer of the corresponding trained network of input layer of each trained network is determined by following equation
Border output:
xj=f (Sj)
The output error of the output layer of the trained network is determined by following equation:
The hidden layer and the output layer BP neural network connection weight and threshold value are adjusted by following equation:
By following equation between the input layer and the hidden layer weight and threshold values be adjusted:
Wherein, xjFor the output valve of node j, f is activation primitive,wijIt is node i with node j's
Connection weight, bjFor the threshold values of node j, xiFor the output of upper layer node i, j-th of neuron of output layer is denoted as yj, djFor output
Layer it is all as a result, η1, η2For learning rate.
Preferably, the index is the multiple first class index for including in the assessment indicator system or the index is
The multiple two-level index for including in the assessment indicator system.
The embodiment of the invention provides a kind of driving safety evaluation method and devices, this method comprises: being driven according to the whole world
Behavior safety evaluates main indicator and constructs specific assessment indicator system simultaneously with reference to assessment indicator system and retrievable data
Determine the multiple indexs for including in the assessment indicator system；The corresponding multi-group data of multiple indexs is obtained from vehicle,
The multiple indexs for including in every group of data are normalized and by degree of membership formula, obtain every group of data packet
The subordinated-degree matrix of the multiple indexs included；Using the subordinated-degree matrix for the multiple indexs for including in training data group as
The input layer of the training network of BP neural network determines the output layer of the corresponding trained network of the input layer of each trained network
Reality output and output error will test number when determining that the output error of output layer of the trained network is less than convergency value
The input layer of test network according to the subordinated-degree matrix for the multiple indexs for including in group as BP neural network is surveyed when determining
When the accuracy of the output layer of examination network is greater than the set value, the BP neural network is determined as driving safety evaluation model.It should
Connection weight is constantly adjusted with BP neural network algorithm in method, improves subjective determining weight in fuzzy overall evaluation algorithm
Mode, evaluation result is obtained based on quantitative analysis, improves the accuracy and science of evaluation result, furthermore, with data
Be increasing, by constantly adjusting the connection weight of neural BP neural network, can dynamic with New Appraisement process, make the algorithm
There can be wide applicability；This method remains the subordinating degree function method of de-fuzzy, utilizes neural BP nerve net
Network is trained adjustment connection weight, avoids the subjective impact in evaluation procedure, ensure that the science and accuracy of result.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is a kind of driving safety evaluation method flow diagram provided in an embodiment of the present invention；
Fig. 2 is specific evaluation criteria system structural schematic diagram provided in an embodiment of the present invention；
Fig. 3 is initial data schematic diagram provided in an embodiment of the present invention；
Fig. 4 is four achievement data schematic diagrames provided in an embodiment of the present invention；
Fig. 5 is subordinating degree function schematic diagram provided in an embodiment of the present invention；
Fig. 6 is BP neural network structural schematic diagram provided in an embodiment of the present invention；
Fig. 7 is a kind of driving safety evaluating apparatus structural schematic diagram provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Fig. 1 illustratively shows a kind of driving safety evaluation method flow diagram of inventive embodiments offer, such as Fig. 1
Shown, this method mainly comprises the steps that
Step 101, according to global driving behavior safety evaluatio main indicator, with reference to assessment indicator system and retrievable
Data construct specific assessment indicator system and determine the multiple indexs for including in the assessment indicator system；
Step 102, the corresponding multi-group data of multiple indexs is obtained from vehicle, it is multiple by include in every group of data
The index is normalized and by degree of membership formula, obtains being subordinate to for multiple indexs that every group of data include
Spend matrix；
Step 103, using the subordinated-degree matrix for the multiple indexs for including in training data group as BP neural network
The input layer of training network determines the reality output of the output layer of the corresponding trained network of the input layer of each trained network and defeated
Error out, when determine the output error of output layer of the trained network be less than convergency value when, will include in test data set
The input layer of test network of the subordinated-degree matrix of multiple indexs as BP neural network, when the output for determining test network
When the accuracy of layer is greater than the set value, the BP neural network is determined as driving safety evaluation model.
In practical applications, the selection of evaluation index have very strong subjectivity, at present in the world to drive safety into
The Primary Reference index of row evaluation is as shown in table 1.
The global driving behavior safety evaluatio main indicator of table 1
| Project | Country | Evaluation index |
| Metronome | The U.S. | Mileage travelled |
| G-Book | Japan | Mileage, vehicle condition, time |
| Drive Wise | The U.S. | Travel total kilometrage, vehicle deceleration, turning |
| Snapshot | The U.S. | Whether night running, mileage travelled, anxious acceleration/urgency deceleration number |
| Rate My Drive | Europe | Accelerate according to the judgement of cellphone GPS module, slow down, number of turns |
| State Farm | The U.S. | Running time, mileage, hypervelocity number, anxious acceleration/urgency deceleration number, corner |
| Mein Copilot | Europe | Running time, mileage, hypervelocity number, anxious acceleration/urgency deceleration number, zig zag |
| IccCube | Thailand | Driving range and driving behavior |
In embodiments of the present invention, according to the Primary Reference index that drive safety is evaluated in the world to evaluation because
Element is chosen, and assessment indicator system as shown in Table 2 is constructed.
Table 2, assessment indicator system
In a step 101, main indicator is evaluated based on the global driving behavior safety indexes in above-mentioned table 1, in table 2
Assessment indicator system and the by conventional method available data arrived, establish and specifically comment provided by the embodiment of the present invention
Valence index system, Fig. 2 is specific evaluation criteria system structural schematic diagram provided in an embodiment of the present invention, as shown in Fig. 2, this is specific
The index for including in evaluation criteria system mainly includes two first class index and multiple two-level index.
For example, it is first class index that furious driving and " three anxious ", which drive, and acceleration, overspeed time, anxious acceleration times and
Anxious deceleration number is two-level index.
It should be noted that it is mutually indepedent between each first class index, it is also mutually indepedent between each two-level index.In reality
The specific evaluation index applied in the evaluation procedure of border is every two-level index.Wherein, the unit of acceleration is m/s2, when hypervelocity
Between be a ratio, by be more than threshold value running time (unit s) with travel that (ratio of unit s) is determined as exceeding the speed limit total time
Time, it should be noted that threshold speed is a setting value, for example, in embodiments of the present invention, being by threshold speed
60km/h.In practical applications, threshold speed can be changed according to practical application request.Anxious acceleration times and anxious deceleration number are
One incremental count value, refers to the acceleration rate threshold for being previously set, and then counts when being more than the acceleration rate threshold and adds one, at this
In inventive embodiments, acceleration rate threshold can be set as 3m/s2With -3m/s2。
It in practical applications, can be directly from the OBDII of vehicle (English are as follows: the Second On-Board
Diagnostics) interface obtains car status information, after the segmentation of Arduino development board, extracting, being packaged, by distance and in real time
Speed is sent to driving safety overall evaluation system by ESP8266 module.
In application in fact, ESP8266 is the UART-WiFi transparent transmission module of a super low-power consumption, aims at mobile device and object
The physical equipment of user, can be connected on Wi-Fi wireless network by working application design, carry out internet or local area network communication,
Realize network savvy.ESP8266 supports three kinds of antennal interface forms: onboard PCB antenna, IPEX interface and stamp hole interface, plate
Carrying PCB antenna and IPEX interface antenna client can be used directly, without adding any match circuit.
For example, Fig. 3 is initial data schematic diagram provided in an embodiment of the present invention, is obtained from the OBDII interface of vehicle
After data, is handled through Arduino development board and the initial data that is sent by ESP8266 module is as shown in figure 3, need to illustrate
It is Fig. 3 partial datas only intercepted in initial data, wherein every a line in Fig. 3 represents one group of data, one group of data
Be made of respectively three elements: serial number, distance (km), speed (km/h) are separated with comma between element.
Further, with initial data calculate acceleration, overspeed time, anxious acceleration times, anxious deceleration number this four
A index.Fig. 4 is four achievement data schematic diagrames provided in an embodiment of the present invention, as shown in figure 4, acc is acceleration,
Speedtime is overspeed time, and accaddtime is anxious acceleration times, and accdectime is anxious deceleration number.
In a step 102, by the fuzzy quantization processor of construction, the degree of membership of each index is determined.It specifically includes, it will
Each of each group of data in the corresponding multi-group data of multiple indexs got index value, is normalized place first
Reason.
Then by following subordinating degree function formula, the degree of membership that each index corresponds to each grade is calculated, thus
Obtain the subordinated-degree matrix of each index.
Wherein, subordinating degree function formula is specific as follows shown:
1, outstanding membership function function:
2, good level subordinating degree function:
3, general membership function function:
4, inferior grade subordinating degree function:
5, very poor membership function function:
Wherein:C is constant, Ke YiquThe value of k is true by international limiting value
It is fixed, it can be taken as
Fig. 5 is subordinating degree function schematic diagram provided in an embodiment of the present invention, it is possible to further obtain person in servitude as follows
Category degree matrix:
Wherein, Ri=(ri1,ri2,…,rij), R represents multifactor evaluation subordinated-degree matrix, RiSingle factor evaluation is represented to be subordinate to
Spend the degree of membership that vector i.e. i-th of evaluation index corresponds to each opinion rating, rijI-th of index is represented to comment relative to jth layer
The degree of membership of valence grade.
It should be noted that in embodiments of the present invention, is constructed in specific assessment indicator system by step 101 and include
4 two-level index, since degree of membership formula includes 5 grades, for each grade, definitionX1=0.0, X2
=0.2, finally obtained subordinated-degree matrix is the matrix of 4*5, using the matrix as the input of neural BP neural network.
Fig. 6 be BP neural network structural schematic diagram provided in an embodiment of the present invention, as shown in fig. 6, input layer, hidden layer,
Output layer respectively has M, I, J neurons, and m-th of neuron of input layer is denoted as xm, i-th of neuron of hidden layer be denoted as ki, output layer
J-th of neuron is denoted as yj, wmiFor xmTo kiConnection weight, wijFor kiTo yjConnection weight.
The output of node j is as follows:
xj=f (Sj) (formula 8)
Wherein xjFor the output valve of node j, wijFor the connection weight of node i and node j, bjFor the threshold values of node j, xiFor
The output of upper layer node i, f are activation primitive, use Sigmoid function in embodiments of the present invention, i.e.,
F (x)=1/ (1+e-x) (formula 9)
Weight and threshold values subprocess are reversely adjusted according to error: setting all results of output layer as dj, then error function be
According to gradient descent method, by between hidden layer and output layer connection weight and threshold value update it is as follows:
Weight and threshold value update between input layer and hidden layer is as follows:
Wherein η1, η2For learning rate.By continuously adjusting weight and threshold values, control output is in given error range
Interior convergence, then training terminates.
In step 103, the link weight coefficients of BP neural network initiation parameter such as BP neural network and each are set first
Then the multi-group data got in step 102 is divided into two groups by neuron threshold values, one group of data is as sample for training BP
Neural network is labeled as training data group；Another group of data are used for test b P neural network, are labeled as test data set.
The hidden layer node number of BP neural network is set by following equation 15, and formula 15 is as follows:
Wherein, h is hidden layer node number, and m is input layer number, and n output layer interstitial content, a is between 1 to 10
Regulating constant.
It should be noted that if related to above-mentioned listed embodiment, the m in formula 15, which can be determined as 4, n, to be determined
It is 1, further, a is set as the arbitrary constant between 1-10, the use of mean value is normal distribution that 0 variance is 1 to each weight
Its initial value is set, and setting training precision is 1*10-5.
Further, it is arranged with lesser random number after link weight coefficients and each neuron threshold values of BP neural network,
Using the corresponding subordinated-degree matrix of training data group as the input layer of the training network of BP neural network, due in step 101
The quantity of determining two-level index has 4, and therefore, the input layer of the training network of BP neural network here includes 4, and 4
The output layer of the corresponding 1 trained network of the input layer of a trained network, the corresponding input layer of each training network is each
The membership vector of a factor of evaluation, training network output layer it is corresponding be this group of data evaluation result.
In practical applications, due to the process that the training of BP neural network is an iteration, in the process that first pass calculates
In, the reality output of the output layer of the corresponding trained network of input layer of training network can be determined by formula 8, pass through formula
10 determine the output error of the output layer of the corresponding trained network of input layer of training network, however, it is determined that the output layer of training network
Output error be greater than setting convergency value, then need through formula 10, formula 11, formula 12 and formula 13 are to BP neural network
Parameter is adjusted, and then continues to carry out second time calculating with BP neural network after adjustment, until the output of training network
The output error of layer is less than setting convergency value extremely, then stops iteration, and BP neural network training terminates.
Further, it needs to test the BP neural network that training terminates with test data set, there is ground, by test data set
Including multiple two-level index subordinated-degree matrix as BP neural network test network input layer, it is true according to the above method
The reality output for determining the output layer of the corresponding test network of input layer of test network, by the output of all determining test network
The reality output of layer is compared with source data by following equation, when determining that accuracy is greater than the set value, it is determined that
The BP neural network is driving safety evaluation model.
Wherein it is determined that the accuracy formula of the reality output of the output layer of test network is as follows:
Wherein, y 'iFor the reality of the output layer of the corresponding test network of input layer of each test network of test data set
Output, yiFor source data result.
It in practical applications, then can be with the BP mind after determining BP neural network is driving safety evaluation model
Driving safety evaluation is carried out through network, corresponding each group of input inputs BP nerve after fuzzy quantization processor de-fuzzy
Network obtains corresponding output i.e. evaluation result.
In conclusion using BP nerve net in this method the embodiment of the invention provides a kind of driving safety evaluation method
Network algorithm constantly adjusts connection weight, the subjective mode for determining weight in fuzzy overall evaluation algorithm is improved, based on quantitative point
Analysis obtains evaluation result, improves the accuracy and science of evaluation result, furthermore, with being increasing for data, by not
The disconnected connection weight for adjusting neural BP neural network can enable the algorithm to have wide dynamically with New Appraisement process
Applicability；This method remains the subordinating degree function method of de-fuzzy, is trained adjustment using neural BP neural network and connects
Weight is connect, the subjective impact in evaluation procedure is avoided, ensure that the science and accuracy of result.
Based on the same inventive concept, the embodiment of the invention provides a kind of driving safety evaluating apparatus, due to the device solution
Certainly the principle of technical problem is similar to a kind of driving safety evaluation method, therefore the implementation of the device may refer to the reality of method
It applies, overlaps will not be repeated.
Fig. 7 is a kind of driving safety evaluating apparatus structural schematic diagram provided in an embodiment of the present invention, as shown in fig. 7, the dress
It sets including the first determination unit 701, obtains unit 702 and the second determination unit 703.
First determination unit 701 is used for according to global driving behavior safety evaluatio main indicator, with reference to evaluation index body
System and retrievable data construct specific assessment indicator system and determine the multiple indexs for including in the assessment indicator system；
Unit 702 is obtained, for obtaining the corresponding multi-group data of multiple indexs from vehicle, will be wrapped in every group of data
The multiple indexs included are normalized and by degree of membership formula, obtain multiple fingers that every group of data include
Target subordinated-degree matrix；
Second determination unit 703, for using the subordinated-degree matrix for the multiple indexs for including in training data group as
The input layer of the training network of BP neural network determines the output layer of the corresponding trained network of the input layer of each trained network
Reality output and output error will test number when determining that the output error of output layer of the trained network is less than convergency value
The input layer of test network according to the subordinated-degree matrix for the multiple indexs for including in group as BP neural network is surveyed when determining
When the accuracy of the output layer of examination network is greater than the set value, the BP neural network is determined as driving safety evaluation model.
Preferably, second determination unit 703 is also used to:
The hidden layer node number of the BP neural network is set according to following equation:
Wherein, h is hidden layer node number, and m is input layer number, and n output layer interstitial content, a is between 1 to 10
Regulating constant.
Preferably, the reality of the output layer of the corresponding trained network of input layer of each trained network is determined by following equation
Border output:
xj=f (Sj)
The output error of the output layer of the trained network is determined by following equation:
The hidden layer and the output layer BP neural network connection weight and threshold value are adjusted by following equation:
By following equation between the input layer and the hidden layer weight and threshold values be adjusted:
Wherein, xjFor the output valve of node j, f is activation primitive,wijIt is node i with node j's
Connection weight, bjFor the threshold values of node j, xiFor the output of upper layer node i, j-th of neuron of output layer is denoted as yj, djFor output
Layer it is all as a result, η1, η2For learning rate.
Preferably, the index includes any one following or multiple combination: acceleration, overspeed time, anxious to accelerate time
Number, anxious deceleration number；
The acceleration and the corresponding first class index of the overspeed time are furious driving, urgency acceleration times and described
The corresponding first class index of anxious deceleration number is that " three is anxious " drives.
It should be appreciated that one of the above driving safety evaluating apparatus include unit only according to the apparatus realize function
The logical partitioning that can be carried out in practical application, can carry out the superposition or fractionation of said units.And the one of embodiment offer
The function and a kind of driving safety evaluation method provided by the above embodiment that kind driving safety evaluating apparatus is realized correspond,
For the more detailed process flow that the device is realized, it has been described in detail in above method embodiment one, herein not
It is described in detail again.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the present invention
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the present invention, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although preferred embodiments of the present invention have been described, it is created once a person skilled in the art knows basic
Property concept, then additional changes and modifications may be made to these embodiments.So it includes excellent that the following claims are intended to be interpreted as
It selects embodiment and falls into all change and modification of the scope of the invention.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (10)
1. a kind of driving safety evaluation method characterized by comprising
According to global driving behavior safety evaluatio main indicator, with reference to assessment indicator system and retrievable data, building tool
Body assessment indicator system simultaneously determines the multiple indexs for including in the assessment indicator system；
The corresponding multi-group data of multiple indexs is obtained from vehicle, and the multiple indexs for including in every group of data are carried out
Normalized and by degree of membership formula, obtains the subordinated-degree matrix for multiple indexs that every group of data include；
Using the subordinated-degree matrix for the multiple indexs for including in training data group as the defeated of the training network of BP neural network
Enter layer, the reality output and output error of the output layer of the corresponding trained network of the input layer of each trained network is determined, when true
When the output error of the output layer of the fixed trained network is less than convergency value, the multiple indexs that will include in test data set
Subordinated-degree matrix as BP neural network test network input layer, when the accuracy for the output layer for determining test network is big
When setting value, the BP neural network is determined as driving safety evaluation model.
2. the method as described in claim 1, which is characterized in that the multiple indexs that will include in training data group
Before input layer of the subordinated-degree matrix as the training network of BP neural network, comprising:
The hidden layer node number of the BP neural network is set according to following equation:
Wherein, h is hidden layer node number, and m is input layer number, and n output layer interstitial content, a is the tune between 1 to 10
Save constant.
3. method according to claim 2, which is characterized in that determine the input layer pair of each trained network by following equation
The reality output of the output layer for the training network answered:
xj=f (Sj)
The output error of the output layer of the trained network is determined by following equation:
The hidden layer and the output layer BP neural network connection weight and threshold value are adjusted by following equation:
By following equation between the input layer and the hidden layer weight and threshold values be adjusted:
Wherein, xjFor the output valve of node j, f is activation primitive,wijFor the connection weight of node i and node j
Value, bjFor the threshold values of node j, xiFor the output of upper layer node i, j-th of neuron of output layer is denoted as yj, djFor the institute of output layer
Have as a result, η1, η2For learning rate.
4. the method as described in claim 1, which is characterized in that determine the output layer of the test network by following equation
Accuracy:
Wherein, y 'iFor the reality output of the output layer of the corresponding test network of input layer of each test network of test data set,
yiFor source data result.
5. the method as described in claim 1, which is characterized in that the index is multiple for include in the assessment indicator system
First class index or the index are the multiple two-level index for including in the assessment indicator system.
6. the method as described in claim 1, which is characterized in that the subordinated-degree matrix is as follows:
Wherein, Ri=(ri1,ri2,…,rij), R represents multifactor evaluation subordinated-degree matrix, RiRepresent single factor evaluation degree of membership to
Amount and i-th of evaluation index correspond to the degree of membership of each opinion rating, rijI-th of index is represented relative to jth layer evaluation etc.
The degree of membership of grade.
7. a kind of driving safety evaluating apparatus characterized by comprising
First determination unit, for according to global driving behavior safety evaluatio main indicator, with reference to assessment indicator system and can
The data of acquisition construct specific assessment indicator system and determine the multiple indexs for including in the assessment indicator system；
Unit is obtained, it is more by include in every group of data for obtaining the corresponding multi-group data of multiple indexs from vehicle
A index is normalized and by degree of membership formula, obtains the person in servitude for multiple indexs that every group of data include
Category degree matrix；
Second determination unit, for using the subordinated-degree matrix for the multiple indexs for including in training data group as BP nerve net
The input layer of the training network of network determines the reality output of the output layer of the corresponding trained network of the input layer of each trained network
And output error will be wrapped when determining that the output error of output layer of the trained network is less than convergency value in test data set
The input layer of test network of the subordinated-degree matrix of the multiple indexs included as BP neural network, when determining test network
When the accuracy of output layer is greater than the set value, the BP neural network is determined as driving safety evaluation model.
8. device as claimed in claim 7, which is characterized in that second determination unit is also used to:
The hidden layer node number of the BP neural network is set according to following equation:
Wherein, h is hidden layer node number, and m is input layer number, and n output layer interstitial content, a is the tune between 1 to 10
Save constant.
9. device as claimed in claim 7, which is characterized in that determine the input layer pair of each trained network by following equation
The reality output of the output layer for the training network answered:
xj=f (Sj)
The output error of the output layer of the trained network is determined by following equation:
The hidden layer and the output layer BP neural network connection weight and threshold value are adjusted by following equation:
By following equation between the input layer and the hidden layer weight and threshold values be adjusted:
Wherein, xjFor the output valve of node j, f is activation primitive,wijFor the connection weight of node i and node j
Value, bjFor the threshold values of node j, xiFor the output of upper layer node i, j-th of neuron of output layer is denoted as yj, djFor the institute of output layer
Have as a result, η1, η2For learning rate.
10. device as claimed in claim 7, which is characterized in that the index is more for include in the assessment indicator system
A first class index or the index are the multiple two-level index for including in the assessment indicator system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811518029.3A CN109572706B (en) | 2018-12-12 | 2018-12-12 | Driving safety evaluation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811518029.3A CN109572706B (en) | 2018-12-12 | 2018-12-12 | Driving safety evaluation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109572706A true CN109572706A (en) | 2019-04-05 |
| CN109572706B CN109572706B (en) | 2020-12-08 |
Family
ID=65929433
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811518029.3A Expired - Fee Related CN109572706B (en) | 2018-12-12 | 2018-12-12 | Driving safety evaluation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109572706B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110069894A (en) * | 2019-05-09 | 2019-07-30 | 同济大学 | A kind of objective mapping test method for intelligent automobile traffic coordinating |
| CN111563555A (en) * | 2020-05-11 | 2020-08-21 | 广东广顺新能源动力科技有限公司 | Driver driving behavior analysis method and system |
| CN111835715A (en) * | 2020-06-03 | 2020-10-27 | 北京邮电大学 | Method and device for determining safety value of virtual network function |
| CN112101777A (en) * | 2020-09-11 | 2020-12-18 | 湖南科技大学 | Safety evaluation method of hazardous chemical substance road transportation facility based on improved AHP |
| CN113380048A (en) * | 2021-06-25 | 2021-09-10 | 中科路恒工程设计有限公司 | Neural network-based high-risk road section vehicle driving behavior identification method |
| CN113537002A (en) * | 2021-07-02 | 2021-10-22 | 安阳工学院 | Driving environment evaluation method and device based on dual-mode neural network model |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09305898A (en) * | 1996-05-10 | 1997-11-28 | Hitachi Ltd | Stroll driving discriminating method |
| CN101727741A (en) * | 2008-10-30 | 2010-06-09 | 爱信艾达株式会社 | Safe driving evaluation system and safe driving evaluation program |
| JP4995046B2 (en) * | 2007-11-21 | 2012-08-08 | 株式会社日立製作所 | Auto insurance premium setting system |
| CN104268701A (en) * | 2014-09-29 | 2015-01-07 | 清华大学 | Commercial vehicle driving safety evaluation system and method |
| WO2015002025A1 (en) * | 2013-07-02 | 2015-01-08 | Ｊｘ日鉱日石エネルギー株式会社 | Method for generating index for evaluating driving, information processing apparatus, vehicle-mounted device, and control method and control program therefor |
| CN106022561A (en) * | 2016-05-05 | 2016-10-12 | 广州星唯信息科技有限公司 | Driving comprehensive evaluation method |
| CN106874669A (en) * | 2017-02-14 | 2017-06-20 | 深圳市可可卓科科技有限公司 | Vehicle security drive evaluation method and device |
| CN108171641A (en) * | 2017-12-21 | 2018-06-15 | 东南大学 | A kind of rail traffic emergency preplan appraisal procedure |
-
2018
- 2018-12-12 CN CN201811518029.3A patent/CN109572706B/en not_active Expired - Fee Related
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09305898A (en) * | 1996-05-10 | 1997-11-28 | Hitachi Ltd | Stroll driving discriminating method |
| JP4995046B2 (en) * | 2007-11-21 | 2012-08-08 | 株式会社日立製作所 | Auto insurance premium setting system |
| CN101727741A (en) * | 2008-10-30 | 2010-06-09 | 爱信艾达株式会社 | Safe driving evaluation system and safe driving evaluation program |
| WO2015002025A1 (en) * | 2013-07-02 | 2015-01-08 | Ｊｘ日鉱日石エネルギー株式会社 | Method for generating index for evaluating driving, information processing apparatus, vehicle-mounted device, and control method and control program therefor |
| CN104268701A (en) * | 2014-09-29 | 2015-01-07 | 清华大学 | Commercial vehicle driving safety evaluation system and method |
| CN106022561A (en) * | 2016-05-05 | 2016-10-12 | 广州星唯信息科技有限公司 | Driving comprehensive evaluation method |
| CN106874669A (en) * | 2017-02-14 | 2017-06-20 | 深圳市可可卓科科技有限公司 | Vehicle security drive evaluation method and device |
| CN108171641A (en) * | 2017-12-21 | 2018-06-15 | 东南大学 | A kind of rail traffic emergency preplan appraisal procedure |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110069894A (en) * | 2019-05-09 | 2019-07-30 | 同济大学 | A kind of objective mapping test method for intelligent automobile traffic coordinating |
| CN111563555A (en) * | 2020-05-11 | 2020-08-21 | 广东广顺新能源动力科技有限公司 | Driver driving behavior analysis method and system |
| CN111835715A (en) * | 2020-06-03 | 2020-10-27 | 北京邮电大学 | Method and device for determining safety value of virtual network function |
| CN111835715B (en) * | 2020-06-03 | 2021-06-04 | 北京邮电大学 | Method and device for determining safety value of virtual network function |
| CN112101777A (en) * | 2020-09-11 | 2020-12-18 | 湖南科技大学 | Safety evaluation method of hazardous chemical substance road transportation facility based on improved AHP |
| CN112101777B (en) * | 2020-09-11 | 2022-07-12 | 湖南科技大学 | Safety evaluation method of hazardous chemical substance road transportation facility based on improved AHP |
| CN113380048A (en) * | 2021-06-25 | 2021-09-10 | 中科路恒工程设计有限公司 | Neural network-based high-risk road section vehicle driving behavior identification method |
| CN113380048B (en) * | 2021-06-25 | 2022-09-02 | 中科路恒工程设计有限公司 | Neural network-based high-risk road section vehicle driving behavior identification method |
| CN113537002A (en) * | 2021-07-02 | 2021-10-22 | 安阳工学院 | Driving environment evaluation method and device based on dual-mode neural network model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109572706B (en) | 2020-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109572706A (en) | A kind of driving safety evaluation method and device | |
| CN112896170B (en) | Automatic driving transverse control method under vehicle-road cooperative environment | |
| CN109840660B (en) | Vehicle characteristic data processing method and vehicle risk prediction model training method | |
| CN112037513B (en) | Real-time traffic safety index dynamic comprehensive evaluation system and construction method thereof | |
| Du et al. | A hierarchical framework for improving ride comfort of autonomous vehicles via deep reinforcement learning with external knowledge | |
| CN106095963A (en) | Vehicle drive behavior analysis big data public service platform under the Internet+epoch | |
| CN106127586A (en) | Vehicle insurance rate aid decision-making system under big data age | |
| CN108152059A (en) | High-speed train bogie fault detection method based on Fusion | |
| CN104200267A (en) | Vehicle driving economy evaluation system and vehicle driving economy evaluation method | |
| DE102016210453A1 (en) | Vehicle, system in communication with a communication module of the vehicle, and system in communication with a group of vehicles | |
| CN103247091A (en) | Driving evaluation system and driving evaluation method | |
| CN107239905A (en) | Onboard networks safety risk estimating method based on advanced AHP GCM | |
| DE102022100549A1 (en) | RANKED FAULT CONDITIONS | |
| CN103456163A (en) | City expressway interchange traffic capacity and running status discrimination method and system | |
| Čubranić-Dobrodolac et al. | A bee colony optimization (BCO) and type-2 fuzzy approach to measuring the impact of speed perception on motor vehicle crash involvement | |
| CN104331746B (en) | A kind of dynamic path optimization system and method for separate type | |
| CN105835854B (en) | A kind of emergency braking control system and its control method | |
| CN113988705A (en) | Traffic safety risk assessment method and device | |
| Zeng et al. | An LSTM-based driving operation suggestion method for riding comfort-oriented critical zone | |
| Ou et al. | Risk prediction model for drivers’ in-vehicle activities–Application of task analysis and back-propagation neural network | |
| Ferrarotti et al. | Autonomous and Human-Driven Vehicles Interacting in a Roundabout: A Quantitative and Qualitative Evaluation | |
| Pellegrino | Road context evaluated by means of fuzzy interval | |
| Ma et al. | Impact of screen size and structure of IVIS on driving distraction | |
| CN115771506B (en) | Method and device for determining vehicle driving strategy based on passenger risk cognition | |
| Ma | Effects of vehicles with different degrees of automation on traffic flow in urban areas |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20201208 |