CN110276370A - A kind of road traffic accident risk Factor Analysis method based on random forest - Google Patents
A kind of road traffic accident risk Factor Analysis method based on random forest Download PDFInfo
- Publication number
- CN110276370A CN110276370A CN201910366177.6A CN201910366177A CN110276370A CN 110276370 A CN110276370 A CN 110276370A CN 201910366177 A CN201910366177 A CN 201910366177A CN 110276370 A CN110276370 A CN 110276370A
- Authority
- CN
- China
- Prior art keywords
- random forest
- node
- accident
- severity
- risk factors
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Analytical Chemistry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The road traffic accident risk Factor Analysis method based on random forest that the invention discloses a kind of, it is by being related to the influence factor and corresponding severity of injuries data of four people, vehicle, road, environment aspects when extraction traffic accident generation, road traffic accident severity prediction model is established according to random forest predictive model algorithm, model training is carried out, model parameter is adjusted；Then the prediction model of foundation is subjected to variable importance measurement using Gini index method, importance ranking is carried out to associated risk factors, identifies the factor having a significant impact to severity of injuries.Method of the invention can many-sided influence factor for considering road traffic accident, depth excavates contacting between risk factors and severity of injuries, is conducive to road safety management.
Description
Technical field
The invention belongs to traffic safety research fields, and in particular to a kind of road traffic accident based on random forest
Risk Factor Analysis method.
Background technique
With economic rapid growth, China's private car's owning amount is increased with swift and violent speed, is reached within 2017
1.8695 hundred million, while road traffic accident plays number and remains high.According to statistics, total people that the whole world is got killed by traffic accident every year
Number is more than 1,000,000 people, and there are also 2000 to 50,000,000 people because road traffic accident is injured or disables.Road traffic accident is not only serious
The life and health that threaten people also creates huge economic loss, and the economic loss valuation of annual accumulative traffic injury is about
20000000000 dollars, reach the 1% to 3% of countries in the world gross national product.Therefore, effectively control and reduction traffic safety
Problem is the most important thing of economic society stable development.
China is usually by improving traffic legal system regulation, formulation traffic safety policy, reinforcing traffic management department's function etc.
Method ensures traffic safety.However, inducement of these traditional methods without fundamentally solving road traffic accident.It is counting greatly
According to the epoch, advanced analysis method and research model how are taken, historical traffic casualty data is excavated, finds road traffic accident
Potential influence factor is only the really effective basic method for improving traffic safety problem.
Summary of the invention
In order to solve the defects of prior art, the road traffic accident risk based on random forest that the present invention provides a kind of
Factor approach, it is characterised in that: include the following steps,
Step 1: acquisition leads to accident associated risk factors and the severity of injuries conduct of motor vehicle generation traffic accident
Sample data set, and the sample data set is divided into training set and test set to carry out data prediction；
Step 2: using training set as independent variable, it is pre- to establish random forest as dependent variable for corresponding severity of injuries
It surveys model and trains the model；
Step 3: test set being inputted in random forest prediction model, to judge the essence of random forest prediction model prediction
Degree, and the parameter for adjusting random forest prediction model makes its precision meet demand；
Step 4: variable importance degree being carried out to the random forest prediction model adjusted in step 3 with Gini index method
Amount, identifies the accident associated risk factors having a significant impact to severity of injuries.
It is further: the data prediction in step 1, including processing shortage of data situation and arrange accident relevant risk because
Element, using accident associated risk factors as correlated characteristic set V, corresponding severity of injuries is as prediction target y, with building
The data structure of sample data set D, sample data set D are { v1, v2... vn, y }, wherein vnIt is characterized, n is characterized dimension.
Further: accident associated risk factors include road conditions, driver condition, vehicle condition and ring in step 1
Border situation.
Further: according to training set in step 1: test set=7:3 ratio cut partition is to be pre-processed.
It is further: step 2 specific steps are as follows:
Step 2.1: the input parameter of setting random forest prediction model carries out model training, and input parameter includes: decision
The feature selecting number of a tree number t, the depth deep of every decision tree, accident associated risk factors dimension n and each node
F, t, deep and f are integer, square root that f is n or be logarithm that bottom takes n with 2；
Step 2.2: sampling with putting back to form t post-class processing using Bootstrap method from sample data set D
In self-service sample set X, since successively utilizing the node pair self-service sample set X corresponding with each tree in each tree root node
It is divided；
Step 2.3: nothing is put back to from the n dimension accident associated risk factors that training set has at each node of each tree
Ground randomly selects f dimension accident associated risk factors, and seeks classification effect according to Gini index from f dimension accident associated risk factors
The best kth of fruit ties up accident associated risk factors, and Gini exponential expression is
M is the possibility value number i.e. severity of injuries classification of traffic accident severity y, and i is sample point, piIt is i-th
The probability of the severity of injuries classification of a sample point, using the characteristic value of kth dimension accident associated risk factors as threshold value, to discontented
The present node of sufficient termination condition is divided, wherein the sample that kth dimensional feature in present node is less than threshold value is divided into a left side
Sample remaining in present node is divided into right node by node, and the value range of k is 1 to f；Working as termination condition will be met
Front nodal point is divided into leaf node, and termination condition is to stop division when the depth of tree reaches maximum designated depth, for classification
Problem stops division when node only has a kind of classification；
Step 2.4: all nodes of every decision tree of training establish random forest prediction model.
Further: parameter includes the quantity t and node diagnostic selection number f of decision tree in step 3.
It is further: to carry out variable importance measurement using Gini index method in step 4, be by measuring primary segmentation
The Gini index reduction summation of all nodes is come to feature v after one featurenIt is ranked up, feature vnTo the significance level of node,
Gini index degree variable quantity i.e. before and after node branch is calculated as
I=Gparent-Gsplit1-Gsplit2,
Wherein Gsplit1And Gsplit2Respectively indicate the Gini index of latter two child node of branch, the random forest prediction of foundation
There is t tree in model, by each feature vnTo being normalized after the t significance level set summation, correlated characteristic is obtained
All feature v in set V1—vnImportance sorting, identify the accident correlation wind having a significant impact to severity of injuries
Dangerous factor.
Compared with prior art, the present invention has the advantages that:
(1) influence of many factors to accident is considered, analysis is more fully；
(2) relative to traditional method analyzed accident risk factor, advanced machine learning algorithm is had chosen
The relationship between severity of injuries and accident risk factor is excavated, there is good anti-noise ability, avoided to a certain degree
Fitting phenomenon can excavate potentially risk factors influential on accident；
(3) accident risk factor different degree is ranked up using Gini index method, different risk factors can be explained to thing
Therefore influence.
Detailed description of the invention
Fig. 1 is the algorithm schematic diagram of random forest prediction model of the present invention；
Fig. 2 is the importance measures flow chart of accident associated risk factors of the present invention.
Specific embodiment
In the following with reference to the drawings and specific embodiments, the present invention is furture elucidated, it should be understood that these embodiments are merely to illustrate
It the present invention rather than limits the scope of the invention, after the present invention has been read, those skilled in the art are to of the invention each
The modification of kind equivalent form falls within the application range as defined in the appended claims.
The present invention is based on the road traffic accident risk Factor Analysis methods of random forest, by establishing road traffic accident
Severity prediction model analyzes the potential relationship between severity of injuries and associated risk factors, excavates related to accident
An important factor for, to provide improvement alternative to reduce traffic accident, following steps are specifically included,
Step 1: acquisition leads to road conditions, driver condition, vehicle condition and the environment of motor vehicle generation traffic accident
The accidents such as situation associated risk factors and severity of injuries press training as sample data set, and by the sample data set
Collection: for test set=7:3 ratio cut partition to carry out data prediction, data prediction includes processing shortage of data situation and arrangement
Accident associated risk factors, using accident associated risk factors as correlated characteristic set V, corresponding severity of injuries is as pre-
Target y is surveyed, to construct sample data set D, the data structure of sample data set D is { v1, v2... vn, y }, wherein vnIt is characterized, n
It is characterized dimension.
Step 2: using training set as independent variable, it is pre- to establish random forest as dependent variable for corresponding severity of injuries
It surveys model and trains the model；
Step 2.1: the input parameter of setting random forest prediction model carries out model training, and input parameter includes: decision
The feature selecting number of a tree number t, the depth deep of every decision tree, accident associated risk factors dimension n and each node
F, t, deep and f are integer, square root that f is n or be logarithm that bottom takes n with 2；
Step 2.2: sampling with putting back to form t post-class processing using Bootstrap method from sample data set D
In self-service sample set X, since successively utilizing the node pair self-service sample set X corresponding with each tree in each tree root node
It is divided；
Step 2.3: nothing is put back to from the n dimension accident associated risk factors that training set has at each node of each tree
Ground randomly selects f dimension accident associated risk factors, and seeks classification effect according to Gini index from f dimension accident associated risk factors
The best kth of fruit ties up accident associated risk factors, and Gini exponential expression is
M is the possibility value number i.e. severity of injuries classification of traffic accident severity y, and i is sample point, piIt is i-th
The probability of the severity of injuries classification of a sample point, using the characteristic value of kth dimension accident associated risk factors as threshold value, to discontented
The present node of sufficient termination condition is divided, wherein the sample that kth dimensional feature in present node is less than threshold value is divided into a left side
Sample remaining in present node is divided into right node by node, and the value range of k is 1 to f；Working as termination condition will be met
Front nodal point is divided into leaf node, and termination condition is to stop division when the depth of tree reaches maximum designated depth, for classification
Problem stops division when node only has a kind of classification；
Step 2.4: all nodes of every decision tree of training establish random forest prediction model.
Step 3: test set being inputted in random forest prediction model, to judge the essence of random forest prediction model prediction
Degree, and adjust the parameter of random forest prediction model (quantity t and node diagnostic the selection number f) of decision tree, keep its precision full
Sufficient demand；
Step 4: variable importance degree being carried out to the random forest prediction model adjusted in step 3 with Gini index method
Amount reduces summation particular by the Gini index of all nodes after measurement once one feature of segmentation and comes to feature vnIt is arranged
Sequence, feature vnTo the significance level of node, i.e., the Gini index degree variable quantity before and after node branch is calculated as
I=Gparent-Gsplit1-Gsplit2,
Wherein Gsplit1And Gsplit2Respectively indicate the Gini index of latter two child node of branch, the random forest prediction of foundation
There is t tree in model, by each feature vnTo being normalized after the t significance level set summation, correlated characteristic is obtained
All feature v in set V1—vnImportance sorting, identify the accident correlation wind having a significant impact to severity of injuries
Dangerous factor.
Embodiment 1
Step 1, data collection
The data of acquisition are road conditions when traffic accident occurs between motor vehicle on road, driver condition, vehicle
Condition and the relevant accident associated risk factors of ambient conditions, and corresponding severity of injuries is recorded, the relevant accident of road
Associated risk factors have road curvature, annual average daily traffic, road grade etc.；The relevant accident relevant risk of driver condition
Because be known as driver's gender, the age, whether drunk driving etc.；The relevant accident associated risk factors of vehicle condition have vehicle to use year
Age, type of vehicle relate to accident vehicle number etc.；The relevant accident associated risk factors of ambient conditions have visibility, weather conditions etc.；
Since severity of injuries is difficult to accurately divide the injured degree of personnel, death tolls, number of injured people are comprehensively considered
And three aspect of direct property loss, severity of injuries is divided into death, injury and only property loss three grades, accident
Data sample total amount at least needs 500, and the more result reliabilities of data volume are more preferable, rejects the sample for having missing data, right
Accident associated risk factors are arranged, and the correlated characteristic set V and sample data set D of accident associated risk factors are established, D's
Data structure is { v1, v2... vn, y }, vnIt is characterized, n is characterized dimension, and y is corresponding severity of injuries, and the value of y has
Sample set D is divided into training set and test set according to 7:3 by three classes.
Step 2 establishes random forest prediction model
Step 2.1: setting random forest parameter
The depth deep of decision tree number t and each tree is set, the feature selecting number of each node is f, and accident is related
Risk factors dimension n carries out model training；Wherein, t, deep and f are integer, square root that the value of f is n or with 2 are
Bottom takes the logarithm of n.
Step 2.2: generating the self-service sample set of every CART post-class processing
From have in the training sample set D using Bootstrap method sample with putting back to be formed t tree in self-service sample
Collect X, the self-service sample set X corresponding with each tree of the node pair in each tree successively is started with from root node to each tree
It is divided.
Step 2.3: generating decision tree using self-service sample set
Nothing randomly selects f Wei Te with putting back to from the n dimensional feature that training sample set has at each node that i-th is set
Sign, and the best kth dimensional feature of classifying quality is sought according to Gini index from f dimensional feature, for giving node t, Gini refers to
Number calculation formula is as follows:
Wherein, p (j | t) is the record proportion for belonging to class j in given node t, and M is the possibility value number i.e. accident of y
The classification of severity,
Using the characteristic value of kth dimensional feature as threshold value, the present node for being unsatisfactory for termination condition is divided, wherein will
The sample that kth dimensional feature is less than threshold value in present node is divided into left sibling, and sample remaining in present node is divided into the right side
Node, the value range of k are 1 to f, the present node for meeting termination condition are divided into leaf node, termination condition is tree
Depth reaches maximum designated depth or when node only has a kind of classification, stops division.
Step 2.4: generating random forest prediction model
The generation that step 2.3 completes t decision tree is repeated, random forest prediction model is obtained.
Step 3, adjustment random forest prediction model parameters
Test set is input in the Random Forest model of foundation, the precision of prediction of judgment models adjusts the number of decision tree
Amount t and node diagnostic selection number f make precision meet demand.
Step 4 judges accident associated risk factors significance level
Variable importance measurement is carried out using Gini index method, is by measuring primary all nodes after dividing a feature
Gini index reduce summation feature be ranked up, with feature vsFor the significance level of node m, before node m branch
Gini index variation amount afterwards calculates are as follows:
Wherein, DLAnd DRRespectively indicate two sub- node sets of branch's rear left and right, Gini (Dm) indicate that the Gini of node m refers to
Number, Gini (DL) and Gini (DR) the Gini index of node m branch two child nodes of rear left and right is respectively indicated,
If feature vsThe node m occurred in decision tree i is in set W, then vsIn the importance that i-th is set are as follows:
There is t tree in the Random Forest model of foundation, then feature vsThe significance level total to Random Forest model are as follows:
Significance level of the feature s to Random Forest model is normalized again:
The value of Δ G is bigger, represents that this feature is more important to Random Forest model, by the important journey after n feature normalization
Degree is ranked up, and identifies the accident associated risk factors having a significant impact to severity of injuries.
Claims (7)
1. a kind of road traffic accident risk Factor Analysis method based on random forest, it is characterised in that: include the following steps,
Step 1: the accident associated risk factors and severity of injuries that acquisition causes motor vehicle that traffic accident occurs are as sample
Data set, and the sample data set is divided into training set and test set to carry out data prediction；
Step 2: using the training set as independent variable, it is pre- to establish random forest as dependent variable for corresponding severity of injuries
It surveys model and trains the model；
Step 3: the test set being inputted in the random forest prediction model, to judge the prediction of random forest prediction model
Precision, and the parameter for adjusting random forest prediction model makes its precision meet demand；
Step 4: variable importance measurement being carried out to the random forest prediction model adjusted in step 3 with Gini index method, is known
It does not have the accident associated risk factors having a significant impact to severity of injuries.
2. the road traffic accident risk Factor Analysis method according to claim 1 based on random forest, feature exist
In: data prediction described in step 1, including processing shortage of data situation and arrangement accident associated risk factors, by accident
Associated risk factors are as correlated characteristic set V, and corresponding severity of injuries is as prediction target y, to construct sample data
Collect D, the data structure of the sample data set D is { v1, v2... vn, y }, wherein vnIt is characterized, n is characterized dimension.
3. the road traffic accident risk Factor Analysis method according to claim 1 based on random forest, feature exist
In: accident associated risk factors described in step 1 includes road conditions, driver condition, vehicle condition and ambient conditions.
4. the road traffic accident risk Factor Analysis method according to claim 1 based on random forest, feature exist
In: according to training set in step 1: test set=7:3 ratio cut partition is to be pre-processed.
5. the road traffic accident risk Factor Analysis method according to claim 2 based on random forest, feature exist
In: step 2 specific steps are as follows:
Step 2.1: the input parameter of setting random forest prediction model carries out model training, and the input parameter includes: decision
The feature selecting number of a tree number t, the depth deep of every decision tree, accident associated risk factors dimension n and each node
F, described t, deep and f are integer, square root that f is n or be logarithm that bottom takes n with 2；
Step 2.2: being formed in t post-class processing from being sampled with putting back in sample data set D using Bootstrap method
Self-service sample set X, since successively utilizing the self-service sample set X corresponding with each tree of the node pair in each tree to carry out root node
It divides；
Step 2.3: at each node of each tree from the n dimension accident associated risk factors that training set has without putting back to
Machine chooses f and ties up accident associated risk factors, and seeks classifying quality most according to Gini index from f dimension accident associated risk factors
Good kth ties up accident associated risk factors, and Gini exponential expression is
M is the possibility value number i.e. severity of injuries classification of traffic accident severity y, and i is sample point, piIt is i-th of sample
The probability of the severity of injuries classification of point, using the characteristic value of kth dimension accident associated risk factors as threshold value, to being unsatisfactory for terminating
The present node of condition is divided, wherein the sample that kth dimensional feature in present node is less than threshold value is divided into left sibling,
Sample remaining in present node is divided into right node, the value range of k is 1 to f；The present node of termination condition will be met
It is divided into leaf node, the termination condition is to stop division when the depth of tree reaches maximum designated depth, and classification is asked
Topic stops division when node only has a kind of classification；
Step 2.4: all nodes of every decision tree of training establish random forest prediction model.
6. the road traffic accident risk Factor Analysis method according to claim 1 based on random forest, feature exist
In: parameter described in step 3 includes the quantity t and node diagnostic selection number f of decision tree.
7. the road traffic accident risk Factor Analysis method according to claim 2 based on random forest, feature exist
In: variable importance measurement is carried out using Gini index method described in step 4, is by after measurement once one feature of segmentation
The Gini index of all nodes, which reduces summation, to be come to feature vnIt is ranked up, feature vnTo the significance level of node, i.e. node branch
The Gini index degree variable quantity of front and back is calculated as
I=Gparent-Gsplit1-Gsplit2,
Wherein Gsplit1And Gsplit2Respectively indicate the Gini index of latter two child node of branch, the random forest prediction model of foundation
In have t tree, by each feature vnTo being normalized after the t significance level set summation, correlated characteristic set V is obtained
Interior all feature v1—vnImportance sorting, identify accident relevant risk that severity of injuries is had a significant impact because
Element.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910366177.6A CN110276370A (en) | 2019-05-05 | 2019-05-05 | A kind of road traffic accident risk Factor Analysis method based on random forest |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910366177.6A CN110276370A (en) | 2019-05-05 | 2019-05-05 | A kind of road traffic accident risk Factor Analysis method based on random forest |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110276370A true CN110276370A (en) | 2019-09-24 |
Family
ID=67959630
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910366177.6A Pending CN110276370A (en) | 2019-05-05 | 2019-05-05 | A kind of road traffic accident risk Factor Analysis method based on random forest |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110276370A (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110491121A (en) * | 2019-07-26 | 2019-11-22 | 同济大学 | A kind of heterogeneity traffic accident causation analysis method and apparatus |
| CN110648537A (en) * | 2019-09-28 | 2020-01-03 | 安徽百诚慧通科技有限公司 | Traffic accident correlation analysis method based on Haen's law |
| CN111612233A (en) * | 2020-05-13 | 2020-09-01 | 云南电网有限责任公司电力科学研究院 | Method and device for obtaining importance scores of power distribution network line re-jump influence factors |
| CN111680897A (en) * | 2020-05-27 | 2020-09-18 | 公安部道路交通安全研究中心 | Service management and control method, device and system and computer readable storage medium |
| CN111680391A (en) * | 2020-04-23 | 2020-09-18 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Accident model generation method, device and equipment for man-machine loop coupling system |
| CN111784873A (en) * | 2020-07-01 | 2020-10-16 | 上海城市交通设计院有限公司 | Machine learning principle-based expressway safety vehicle-mounted intelligent system and working process |
| CN111951550A (en) * | 2020-08-06 | 2020-11-17 | 华南理工大学 | Traffic safety risk monitoring method and device, storage medium and computer equipment |
| CN112036709A (en) * | 2020-08-12 | 2020-12-04 | 东南大学 | Random forest based rainfall weather expressway secondary accident cause analysis method |
| CN112257914A (en) * | 2020-10-16 | 2021-01-22 | 中国人民解放军空军工程大学 | Aviation safety cause and effect prediction method based on random forest |
| CN112948749A (en) * | 2021-03-02 | 2021-06-11 | 北京交通大学 | System and method for identifying and predicting risk factors of full-chain logistics |
| CN113096405A (en) * | 2021-06-10 | 2021-07-09 | 天津所托瑞安汽车科技有限公司 | Construction method of prediction model, and vehicle accident prediction method and device |
| CN113160593A (en) * | 2021-01-18 | 2021-07-23 | 重庆交通大学 | Mountain road driving safety early warning method based on edge cloud cooperation |
| CN113392885A (en) * | 2021-05-31 | 2021-09-14 | 东南大学 | Traffic accident space-time hot spot distinguishing method based on random forest theory |
| CN114613130A (en) * | 2022-02-18 | 2022-06-10 | 北京理工大学 | Driving credibility analysis method in traffic and delivery system |
| CN114715146A (en) * | 2022-05-09 | 2022-07-08 | 吉林大学 | Method for predicting severity of potential collision accident |
| CN115640997A (en) * | 2022-10-15 | 2023-01-24 | 河北省交通规划设计研究院有限公司 | Traffic accident risk dynamic identification method and identification model construction method |
| CN115689040A (en) * | 2022-11-11 | 2023-02-03 | 浙江省交通运输科学研究院 | Traffic accident severity prediction method and system based on convolutional neural network |
| CN116151388A (en) * | 2022-12-01 | 2023-05-23 | 广州大学 | Federal random forest method based on localized differential privacy |
| CN116229740A (en) * | 2023-05-10 | 2023-06-06 | 湖北大学 | Traffic light timing control method and system |
| CN116777300A (en) * | 2023-07-24 | 2023-09-19 | 武汉大学 | Urban road traffic accident risk assessment method based on gradient lifting decision tree |
-
2019
- 2019-05-05 CN CN201910366177.6A patent/CN110276370A/en active Pending
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110491121A (en) * | 2019-07-26 | 2019-11-22 | 同济大学 | A kind of heterogeneity traffic accident causation analysis method and apparatus |
| CN110491121B (en) * | 2019-07-26 | 2022-04-05 | 同济大学 | Heterogeneous traffic accident cause analysis method and equipment |
| CN110648537A (en) * | 2019-09-28 | 2020-01-03 | 安徽百诚慧通科技有限公司 | Traffic accident correlation analysis method based on Haen's law |
| CN111680391A (en) * | 2020-04-23 | 2020-09-18 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Accident model generation method, device and equipment for man-machine loop coupling system |
| CN111680391B (en) * | 2020-04-23 | 2024-03-26 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Accident model generation method, device and equipment for man-machine loop coupling system |
| CN111612233A (en) * | 2020-05-13 | 2020-09-01 | 云南电网有限责任公司电力科学研究院 | Method and device for obtaining importance scores of power distribution network line re-jump influence factors |
| CN111680897A (en) * | 2020-05-27 | 2020-09-18 | 公安部道路交通安全研究中心 | Service management and control method, device and system and computer readable storage medium |
| CN111784873A (en) * | 2020-07-01 | 2020-10-16 | 上海城市交通设计院有限公司 | Machine learning principle-based expressway safety vehicle-mounted intelligent system and working process |
| CN111951550A (en) * | 2020-08-06 | 2020-11-17 | 华南理工大学 | Traffic safety risk monitoring method and device, storage medium and computer equipment |
| CN111951550B (en) * | 2020-08-06 | 2021-10-29 | 华南理工大学 | Traffic safety risk monitoring method and device, storage medium and computer equipment |
| CN112036709A (en) * | 2020-08-12 | 2020-12-04 | 东南大学 | Random forest based rainfall weather expressway secondary accident cause analysis method |
| CN112257914B (en) * | 2020-10-16 | 2023-06-06 | 中国人民解放军空军工程大学 | Aviation safety causal prediction method based on random forest |
| CN112257914A (en) * | 2020-10-16 | 2021-01-22 | 中国人民解放军空军工程大学 | Aviation safety cause and effect prediction method based on random forest |
| CN113160593A (en) * | 2021-01-18 | 2021-07-23 | 重庆交通大学 | Mountain road driving safety early warning method based on edge cloud cooperation |
| CN112948749A (en) * | 2021-03-02 | 2021-06-11 | 北京交通大学 | System and method for identifying and predicting risk factors of full-chain logistics |
| CN113392885A (en) * | 2021-05-31 | 2021-09-14 | 东南大学 | Traffic accident space-time hot spot distinguishing method based on random forest theory |
| CN113096405B (en) * | 2021-06-10 | 2021-09-03 | 天津所托瑞安汽车科技有限公司 | Construction method of prediction model, and vehicle accident prediction method and device |
| CN113096405A (en) * | 2021-06-10 | 2021-07-09 | 天津所托瑞安汽车科技有限公司 | Construction method of prediction model, and vehicle accident prediction method and device |
| CN114613130A (en) * | 2022-02-18 | 2022-06-10 | 北京理工大学 | Driving credibility analysis method in traffic and delivery system |
| CN114715146A (en) * | 2022-05-09 | 2022-07-08 | 吉林大学 | Method for predicting severity of potential collision accident |
| CN115640997A (en) * | 2022-10-15 | 2023-01-24 | 河北省交通规划设计研究院有限公司 | Traffic accident risk dynamic identification method and identification model construction method |
| CN115689040A (en) * | 2022-11-11 | 2023-02-03 | 浙江省交通运输科学研究院 | Traffic accident severity prediction method and system based on convolutional neural network |
| CN116151388A (en) * | 2022-12-01 | 2023-05-23 | 广州大学 | Federal random forest method based on localized differential privacy |
| CN116229740A (en) * | 2023-05-10 | 2023-06-06 | 湖北大学 | Traffic light timing control method and system |
| CN116229740B (en) * | 2023-05-10 | 2023-10-10 | 湖北大学 | Traffic light timing control method and system |
| CN116777300A (en) * | 2023-07-24 | 2023-09-19 | 武汉大学 | Urban road traffic accident risk assessment method based on gradient lifting decision tree |
| CN116777300B (en) * | 2023-07-24 | 2024-03-22 | 武汉大学 | Urban road traffic accident risk assessment method based on gradient lifting decision tree |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110276370A (en) | A kind of road traffic accident risk Factor Analysis method based on random forest | |
| CN109117987B (en) | Personalized traffic accident risk prediction recommendation method based on deep learning | |
| CN103198332B (en) | A kind of far infrared vehicle-mounted pedestrian detection method of real-time robust | |
| CN103996287B (en) | A kind of vehicle compulsory based on decision-tree model changes decision-making technique | |
| CN108596409B (en) | Method for improving accident risk prediction precision of traffic hazard personnel | |
| CN110458244A (en) | A kind of traffic accident Severity forecasting method applied to Regional Road Network | |
| CN110796859A (en) | Real-time traffic state identification and accident risk early warning method based on traffic flow | |
| CN111401414B (en) | Natural driving data-based dangerous scene extraction and classification method | |
| CN105868860A (en) | Clustering-based area trip will intensity analysis method, apparatus and addressing method | |
| CN112991685A (en) | Traffic system risk assessment and early warning method considering fatigue state influence of driver | |
| CN111784017A (en) | Road condition factor regression analysis-based road traffic accident quantity prediction method | |
| CN109191828B (en) | Traffic participant accident risk prediction method based on ensemble learning | |
| CN105654574A (en) | Vehicle equipment-based driving behavior evaluation method and vehicle equipment-based driving behavior evaluation device | |
| CN105930855A (en) | Vehicle detection method based on deep convolution neural network | |
| CN105809193A (en) | Illegal operation vehicle recognition method based on Kmeans algorithm | |
| CN109598931A (en) | Group based on traffic safety risk divides and difference analysis method and system | |
| CN112767684A (en) | Highway traffic jam detection method based on charging data | |
| CN111311093A (en) | Road intersection risk assessment and early warning method based on driver physiological data | |
| CN113160564A (en) | Traffic safety early warning analysis method and device and computer equipment | |
| CN111080158A (en) | Urban intersection traffic danger index evaluation method based on composite weight | |
| CN116028884A (en) | Prototype network-based vehicle lane change risk assessment method under small sample | |
| CN113962600A (en) | Traffic safety risk diagnosis method and device | |
| CN113581188A (en) | Commercial vehicle driver driving style identification method based on Internet of vehicles data | |
| CN117021959B (en) | Method for acquiring coasting recovery torque based on driving style identification | |
| DE102010027127A1 (en) | Methodology for identifying emerging problems based on a combined weighting and sensitivity of temporary trends |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190924 |
|
| WD01 | Invention patent application deemed withdrawn after publication |