CN117077042A - Rural level crossing safety early warning method and system - Google Patents
Rural level crossing safety early warning method and system Download PDFInfo
- Publication number
- CN117077042A CN117077042A CN202311337787.6A CN202311337787A CN117077042A CN 117077042 A CN117077042 A CN 117077042A CN 202311337787 A CN202311337787 A CN 202311337787A CN 117077042 A CN117077042 A CN 117077042A
- Authority
- CN
- China
- Prior art keywords
- track
- decision tree
- taking
- classification
- level crossing
- 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
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000003066 decision tree Methods 0.000 claims abstract description 229
- 238000012549 training Methods 0.000 claims description 58
- 238000007637 random forest analysis Methods 0.000 claims description 26
- 238000012360 testing method Methods 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 238000009825 accumulation Methods 0.000 claims description 14
- 239000000203 mixture Substances 0.000 claims description 14
- 239000006185 dispersion Substances 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 238000005259 measurement Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 206010039203 Road traffic accident Diseases 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 238000012806 monitoring device Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 241001048891 Jatropha curcas Species 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 244000144972 livestock Species 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000009928 pasteurization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B31/00—Predictive alarm systems characterised by extrapolation or other computation using updated historic data
-
- 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
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- 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
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- 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
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
- G08G1/0141—Measuring and analyzing of parameters relative to traffic conditions for specific applications for traffic information dissemination
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/16—Anti-collision systems
- G08G1/166—Anti-collision systems for active traffic, e.g. moving vehicles, pedestrians, bikes
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Analytical Chemistry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Chemical & Material Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to the technical field of traffic control, and provides a rural level crossing safety early warning method and system, comprising the following steps: acquiring a traffic information sequence; obtaining track distribution similarity according to the difference of samples in nodes at the same depth in the track decision tree; acquiring a driving track classification index according to the track distribution similarity; obtaining track classification similarity according to the prediction result of the same sample by the track decision tree; obtaining a track classification difference coefficient according to the sample difference of the track decision tree; acquiring track prediction weights based on the track classification similarity and the track classification difference coefficient; acquiring a track ordinal number pair according to the track prediction weight; and acquiring the probability of collision of each automobile according to the matching result of the track ordinal pairs and the track ordinal pairs in the accident database, and sending traffic early warning information to the automobile which is likely to collide by the safety early warning system. According to the method, the prediction accuracy of the collision probability of the automobile is improved by analyzing the driving track characteristics of the vehicles in different lanes at the level crossing.
Description
Technical Field
The invention relates to the technical field of traffic control, in particular to a rural level crossing safety early warning method and system.
Background
The level crossing refers to a crossing where roads of different directions meet in the same plane, and various vehicles and pedestrians on each crossing road meet at the level crossing. In rural areas, rural small roads, countries and provinces form a plurality of level crossing points, but in the level crossing points in vast rural areas, road traffic safety facilities are seriously deficient, people, livestock and past vehicles fight for roads and rob, and some newly built roads are better in condition, wide in vision, easy to paralyze, and extremely easy to cause vehicles to run at overspeed, so that the traffic accident occurrence rate at the rural level crossing points is high, therefore, the prediction algorithm is needed to predict the running track of the vehicles and prompt the vehicles in time so as to reduce the traffic accident.
The random forest algorithm is a supervised learning algorithm based on decision tree integration, has the advantages of high efficiency, stability and strong robustness, and is widely used for tasks such as classification, regression and the like. Because the training set of the random forest algorithm is randomly extracted with a put-back, when different decision trees are trained, the situation that part of decision trees use the same data or the decision trees are too refined can occur, and the risk of overfitting of a random forest model can be increased, the generalization capability of the model is reduced, and the accuracy of the model is further affected.
Disclosure of Invention
The invention provides a rural level crossing safety pre-warning method and system, which aim to solve the problem that error is easy to occur in prediction of the vehicle running track of a level crossing due to overfitting of a random forest model, and the adopted technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a rural level crossing safety precaution method, including the steps of:
acquiring a traffic information sequence of each automobile at a level crossing;
taking a set of traffic information sequences of a preset number of vehicles as a training set in a random forest algorithm, and acquiring a track decision tree in a random forest by using the random forest algorithm based on the training set; obtaining the track distribution similarity of each depth on each track decision tree according to the difference of the running tracks between the samples in the left node and the right node under the same depth in each track decision tree; acquiring a driving track classification index of each track decision tree according to sample distribution in a training set of each track decision tree and the track distribution similarity;
obtaining track classification similarity between two track decision trees according to the prediction result of the two track decision trees on the same sample; acquiring a track classification difference coefficient between the two track decision trees according to the difference between the corresponding running information of the samples in the training sets of the two track decision trees; obtaining track prediction weights of each track decision tree according to the track classification similarity and the track classification difference coefficient between the track decision trees;
and acquiring track ordinal pairs of each automobile on each lane of the level crossing and the probability of collision according to track prediction weights of all track decision trees, and sending traffic early warning information to the automobiles which are likely to collide by the safety early warning system.
Preferably, the method for obtaining the track distribution similarity of each depth on each track decision tree according to the difference of the running tracks between the left and right samples in the same depth in each track decision tree comprises the following steps:
respectively acquiring lane numbers corresponding to each sample in a left node and a right node of each depth on each decision tree by using traffic information sequences corresponding to all samples in a training set of each track decision tree;
and acquiring a measurement distance between a probability histogram constructed by corresponding lane numbers of all samples in a left node of each depth and a probability histogram constructed by corresponding lane numbers of all samples in a right node of each depth, and taking the reciprocal of the sum of the measurement distance and a preset parameter as the track distribution similarity of each depth on each track decision tree.
Preferably, the method for obtaining the driving track classification index of each track decision tree according to the sample distribution in the training set of each track decision tree and the track distribution similarity comprises the following steps:
acquiring the dispersion degree of the driving data of each track decision tree training set according to the distribution characteristics of the driving characteristic data in each track decision tree training set;
taking the sum of the average value of the track distribution similarity of each depth on each track decision tree and the preset parameter as a denominator;
and taking the ratio of the dispersion degree of the running data of each track decision tree training set to the denominator as the running track classification index of each track decision tree.
Preferably, the method for obtaining the dispersion degree of the running data of each track decision tree training set according to the distribution characteristics of the running characteristic data in each track decision tree training set comprises the following steps:
taking the sum of the product of the standard deviation and the range of each driving characteristic data in the training set of the decision tree according to each track and a preset parameter as a first accumulation factor;
and taking the accumulation of the first accumulation factors in each track decision tree training set as the driving data dispersity of each track decision tree training set.
Preferably, the method for obtaining the track classification similarity between two track decision trees according to the prediction results of the two track decision trees on the same sample comprises the following steps:
taking a preset number of samples extracted from a test set as random test samples, taking each random test sample as the input of each track decision tree, and taking a set formed by label values corresponding to the output of each track decision tree according to the extraction sequence as a test track distribution sequence of each track decision tree;
and taking the difference value between elements at the same position in the test track distribution sequence of the two track decision trees as the input of a relation function, and taking the average value of the function values of the relation function on the test track distribution sequence as the track classification similarity between the two track decision trees.
Preferably, the method for obtaining the track classification difference coefficient between the two track decision trees according to the difference between the corresponding running information of the samples in the training sets of the two track decision trees comprises the following steps:
acquiring an automobile driving characteristic difference value between two track decision trees according to the distribution difference of the driving characteristic data in the two track decision tree training sets; taking the sum of the difference value of the driving characteristics of the automobile and preset parameters as a second composition factor;
taking the sum of the measurement distance between probability histograms constructed by corresponding lane numbers of all samples in the training sets of the two track decision trees and preset parameters as a first composition factor;
the track classification difference coefficient between the two track decision trees consists of a first composition factor and a second composition factor, wherein the track classification difference coefficient is in a direct proportion relation with the first composition factor and the second composition factor.
Preferably, the method for obtaining the automobile driving characteristic difference value between the two track decision trees according to the distribution difference of the driving characteristic data in the two track decision tree training sets comprises the following steps:
taking the coding result of the corresponding running characteristic set of each track decision tree training set as the running characteristic coding set of each track decision tree;
the method comprises the steps of taking the edit distance between elements at the same position in a running feature coding set of two track decision trees as input of a relation function, and taking accumulation of function values of the relation function on all elements in the running feature coding set as an automobile running feature difference value between the two track decision trees.
Preferably, the method for obtaining the track prediction weight of each track decision tree according to the track classification similarity and the track classification difference coefficient between the track decision trees comprises the following steps:
taking the product of the track classification similarity and the track classification difference coefficient between each track decision tree and the rest of each track decision tree as a first accumulation factor;
taking the average value of the first accumulation factors on each track decision tree and all the rest track decision trees as a track distinguishing index of each track decision tree;
and taking a normalization result of the product of the track distinguishing index and the running track classifying index of each track decision tree as the track prediction weight of each track decision tree.
Preferably, the method for obtaining the track ordinal number pair and the probability of collision of each automobile on each lane of the level crossing according to the track prediction weights of all the track decision trees comprises the following steps:
taking the track prediction weight of each track decision tree as the weight when voting each track decision tree in a random forest, taking the traffic information sequence of each automobile on each lane of a level crossing as the input of the random forest, acquiring the predicted track label of each automobile by using the random forest, and taking an array pair formed by the driving-in lane number of each automobile level crossing and the lane number corresponding to the predicted track label as the track ordinal number pair of each automobile;
and constructing a track ordinal number pair matching table of the level crossing by using priori knowledge, substituting the track ordinal number pair of each automobile into the track ordinal number pair matching table to obtain a matching result, and enabling the safety early warning system to early warn that the matching result is the automobile which is likely to collide.
In a second aspect, the embodiment of the invention also provides a rural level crossing safety precaution system, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein the processor realizes the steps of any one of the methods when executing the computer program.
The beneficial effects of the invention are as follows: the method provided by the invention has the beneficial effects that the classification effect of the track decision tree on different depths and the distribution characteristics of the driving characteristics of each automobile on the track decision tree are analyzed, and the driving track classification capability is constructed; secondly, analyzing the distribution difference conditions of the training set of each track decision tree and the automobile driving characteristic set on different track decision trees, and constructing a track distinguishing index by combining the similarity of classification capability among the track decision trees; the comprehensive weight of each track decision tree is obtained based on the track classification capability and the track distinguishing index, and is used as the weight when the track decision tree in the random forest algorithm votes, so that the accuracy of the random forest algorithm model is improved, and the automobile track prediction result with higher accuracy is obtained.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for warning security at rural level crossings according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a driving track of an automobile at a level crossing according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a flowchart of a rural level crossing safety pre-warning method according to an embodiment of the invention is shown, and the method includes the following steps:
and S001, acquiring a traffic information sequence of each automobile at the level crossing.
Because the traffic signal lamp is not installed under the common condition of the rural level crossing, the driving safety of the automobile at the rural level crossing needs an additional early warning method, and traffic accidents are avoided to the greatest extent. The invention aims to predict whether vehicles at a level crossing collide or not by predicting the running track of the vehicles entering the level crossing, and early warning is carried out on the vehicles in advance.
The traffic information monitoring device is arranged at a preset position of the level crossing and used for acquiring driving data of the automobile on the level crossing, the traffic information monitoring device comprises a radar sensor and a monitoring camera, the weather information of the automobile passing the level crossing is acquired by utilizing the weather information of vehicle-mounted navigation, the driving data comprises driving-in lanes of the automobile entering the level crossing, driving-out speeds, driving-out acceleration and driving-out angles when the automobile drives out of a current lane, namely 5 driving data are acquired in the method, the number n of the acquired vehicles takes a checked value of 1000, and the acquired weather information comprises sunny days, rainy days, snowy days and haze days, wherein the driving data and the number of the vehicles can be set by an operator according to specific conditions of the rural level crossing.
Secondly, the lanes of the cross-shaped level crossing are artificially numbered, the cross-shaped level crossing has 8 lanes, so that the 8 lanes are sequentially numbered, the 8 lanes are respectively marked as 1-8 according to the anticlockwise sequence, the acquired weather information is digitally processed, 4 kinds of weather information on sunny days, rainy days, snowy days and haze days are respectively corresponding to quantization numbers 1, 2,3 and 4, and the larger the quantization number is, the worse the weather is, the larger the influence on the driving safety is. Secondly, respectively carrying out normalization processing on the obtained data corresponding to the driving-in lane position, the driving-out speed, the driving-out acceleration, the driving-out angle and the weather condition, wherein the normalization processing of the data is a known technology, and is not repeated, a traffic information sequence of each automobile is constructed by utilizing the normalization result, and the following is carried outThe traffic information sequence of the vehicle is marked +.>Wherein->、/>、/>、/>、/>Respectively represent +.>Normalization results corresponding to the driving-in lane, the driving-out speed, the driving-out acceleration, the driving-out angle and the weather information of the vehicle.
So far, the traffic information sequence of each automobile at the level crossing is obtained and is used for generating subsequent early warning information.
Step S002, obtaining track distribution similarity based on the difference of the running tracks between samples at the same depth in the track decision tree, and obtaining the running track classification index based on the sample distribution and the track distribution similarity in the track decision tree training set.
The method for acquiring the early warning information is to predict the driving-out lanes of the automobiles driving into the level crossing by using a random forest algorithm, and predict whether collision occurs according to the driving-in lanes and the driving-out lanes between the automobiles. In the random forest algorithm, the stronger the classification capability of each decision tree and the smaller the correlation between any two decision trees are, the better the classification effect of the algorithm is. According to the method, the classification capability of different decision trees is evaluated, the decision tree with stronger data classification is given larger weight, and the decision tree with weak data classification capability is given smaller weight, so that the self-adaptive weight setting is carried out on the different decision trees.
Specifically, the number of the entering lane of each automobile is used as the label of each automobile, and the label is selectedTraffic information sequence of vehicle is used as training set B for training random forest, wherein, training is carried outTraining set->Is +.>Taking the experience value as 800, and marking the set consisting of the traffic information sequences of the rest automobiles as a test set +.>. Number of decision tree->The number of classification characteristics of each decision tree>The empirical values of 100 and 3 are respectively taken, the selection standard of the node classification characteristic is a base index, and when each decision tree is trained, the node classification characteristic is randomly replaced from the training set>Middle draw->Taking the automobile sample as a sample in the training set of each decision tree, and taking the training set of the h decision tree +.>The corresponding driving feature set is marked +.>The driving feature set comprises +.>Literal record information of the driving characteristics of the inner sample, wherein the driving characteristics refer to the driving-in lane position, driving-out speed and the like of the automobile, and the depth of each decision tree in a random forest algorithm is +.>Taking an empirical value of 30, and adopting a random forest algorithm as a known technology, and implementing the specific processAnd not described in detail, each trained decision tree is marked as a track decision tree.
The sample distribution in the left and right nodes at the same depth in each track decision tree reflects the classification capability of the track decision tree on automobiles with different driving characteristics, so that the accuracy of each track decision tree is evaluated by considering the difference between samples in the nodes at the same depth in each track decision tree.
Specifically, by the firstFor example, the corresponding training set is +.>Second, obtain->Track distribution histogram of sample points in left and right nodes of each track decision tree at each depth in the decision tree>、/>Wherein bin of the track distribution histogram is a serial label of the driving-in lane corresponding to the sample, the bin value is the frequency of occurrence of each label in the sample in the corresponding node, and further, the +_ is calculated based on the track distribution histogram>Depth in the individual track decision tree is +.>Node trace distribution similarity->：
In the method, in the process of the invention,、/>respectively represent +.>The depth of the individual track decision tree is +.>A track distribution histogram of the left and right nodes at the position; />Representing histogram->、/>The pasteurization distance between the two layers is a known technology, and the specific process is not repeated; />Representing parameter tuning constants, aiming at preventing denominator from being 0 ++>The magnitude of (2) takes the empirical value of 1.
Wherein, the firstThe depth of the individual track decision tree is +.>Track distribution histogram of left and right nodes at +.>、/>The more similar the distribution of->The smaller the value of +.>The larger the value of (2), the more the track decision tree is at depth +.>The worse the classification effect.
Further, based on the analysis, a driving track classification index is constructed here for representing classification accuracy of different track decision trees on different driving tracks, and the first is calculatedTravel track classification index of individual track decision tree +.>：
In the method, in the process of the invention,indicate->Training set corresponding to each track decision tree>The running data dispersion of the inner sample, M is the running feature set +.>The number of categories of the driving characteristics in the middle, M has a value of 5 +.>、/>Respectively represent the driving characteristic set->The%>The extreme differences and standard deviations on the individual travel characteristic data; />Indicating parameter tuning constant, prevent->The value of (2) is 0, (-)>Taking an empirical value of 1;
is->The depth of the individual track decision tree is +.>Similarity of node track distribution of depth h in each track decision tree, < ->Is a parameter regulating factor, and is a herb of Jatropha curcas>The function of (2) is to prevent the denominator from being 0, < >>The magnitude of (2) takes the empirical value of 1.
Wherein the training setEach sample in->Personal driving characteristicsThe greater the distribution range and degree of dispersion of the data on, i.e. +.>、/>The larger the value of (2), the first accumulation factor +.>The larger the value of (2), the training set +.>The more dispersed the values of the driving data of the samples are, +.>The greater the value of (2); the smaller the probability of excessively refined classification in the g-th track decision tree, the better the classification capability of the track decision tree is represented by +.>The larger the value of (2), the greater the weight of the classification result of the trajectory decision tree in the voting mechanism.
So far, the running track classification index of each track decision tree is obtained and used for weight calculation when each subsequent decision tree votes.
Step S003, obtaining a track classification difference coefficient according to sample differences in the track decision tree training set, and obtaining track prediction weights of each track decision tree according to track classification similarity and the track classification difference coefficient.
According to the steps, the evaluation result of the classification precision of each track decision tree on the automobiles with different driving tracks is obtained, and further, the similarity degree among different track decision trees is evaluated through the samples in the test set.
From a test setIs optionally extracted->A number of samples of the sample were taken,the size of k is taken as an empirical value 20, classification results of all track decision trees on k detection samples are obtained, for any track decision tree, a sequence formed by labels of the obtained k detection sample classification results according to the extraction sequence of k samples is taken as a test track distribution sequence of each track decision tree, and the test track distribution sequence of the g track decision tree is taken as a' per track>Track classification similarity between track decision trees is calculated based on test track distribution sequences>Calculate +.>、/>Track classification similarity between track decision trees +.>：
Wherein:indicate->Root and->Track classification similarity between track decision trees, < ->、/>Respectively +.>Root and->Test track distribution sequence of track decision tree, </i >>、/>Respectively express the sequence->、/>Middle->Element values for the individual locations; k represents the number of detection samples; />As a relation function, if the value input in the relation function is 0, then +.>Otherwise->Wherein->Indicate->The individual element is at->Root and->The decision tree of the track has the same classification result, otherwise the decision tree of the track represents the +.>The individual element is at->Root and->The trace decision tree has different classification results.
Wherein, the firstRoot and->The closer the classification result of k detection samples is to the track decision tree, the test track distribution sequence +.>、/>The smaller the difference between the element values of the same position in +.>It may be possible for the value of (c) to be 0,the larger the value of (c), the more similar the trajectory classification capability between two trajectory decision trees.
Further, obtaining the track classification difference coefficient between the track decision treesSpecifically, the travel feature set for each track decision tree +.>The data in the track decision tree are respectively subjected to UTF-8 coding, and the corresponding running characteristic coding set of each track decision tree training set is +.>Wherein UTF-8 coding is a known technique and will not be described in detail. Secondly, obtaining a track distribution histogram of sample points in a training set of each track decision tree>。
Based on the analysis, a track classification difference coefficient dc is constructed here for characterizing the similarity between different track decision trees, and the first is calculated、/>Track classification difference coefficient between track decision trees +.>：
In the method, in the process of the invention,respectively represent +.>First->The difference value of the running characteristics of the automobile between the track decision trees; />Indicate->The first part of the driving characteristic code set of the track decision tree>Individual vehicle driving characteristicsEncoding; />Indicate->The first part of the driving characteristic code set of the track decision tree>Encoding the driving characteristics of the individual automobiles; k represents the number of the automobile running feature codes in the automobile running feature code set; />Coding for indicating the driving characteristics of a motor vehicle>、/>ED editing distance between the two is a known technology and is not repeated; />As a relational function, if the value input by the relational function is 0, thenOtherwise->Wherein->Representing->I.e. +.>、/>The two automobile driving feature codes are identical;
indicate->Root and->Track classification difference coefficients between track decision trees; />Indicate->Root and->The Babbitt distance between the track distribution histograms corresponding to the track decision trees; />、/>Is used for regulating parameter and preventing +.>The value of (2) is 0, (-)>、/>The magnitude of (2) is respectively taken as an empirical value of 1.
Wherein, the firstFirst->The more the number of the same automobile driving characteristics in the driving characteristic set corresponding to the track decision tree is, namely +.>The greater the number of 0 +.>First->The more consistent the corresponding running feature set of the track decision tree is,/->The smaller the value of (2) the second composition factor +.>The smaller the value of (2); first->First->The smaller the track distribution difference of sample points in the track decision tree training set is, the more the track distribution difference is +.>The smaller the value of (a) is, the first composition factorThe smaller the value of +.>First->The more consistent the classification characteristics of the track decision trees, the closer the track classification results of the two track decision trees to the same sample are, the +.>First->The more similar the track classification capability between the track decision trees, i.e. +.>The smaller the value of (2).
Further, the similarity of the track classifications among the decision trees is determined according to each trackAnd the track classification difference coefficient->Obtaining the track distinguishing index of each track decision tree, and calculating the track distinguishing index of the g-th track decision tree>：
Wherein:、/>respectively represent +.>Root and->Track classification similarity and track classification difference coefficient between track decision trees>Representing the number of trajectory decision trees.
Wherein, the firstDecision tree of the track and->The more similar the classification capability between the trajectory decision trees, the +.>The greater the value of (2), the more->Decision tree of the track and->The greater the difference between samples in the training set of the trajectory decision tree, the more +.>The larger the value of (2), the first accumulation factor +.>The greater the value of +.>The track decision tree has a good classification effect in the whole sample set and has a large weight; and if->The difference between the decision tree of the trace and the training set of the decision tree of the rest trace is small, i.e./the decision tree of the trace is>The value of (2) is smaller, indicating +.>The classification characteristics of the trace decision tree and the rest trace decision tree are almost identical, which indicates +.>The reason why the classification results of the trace decision tree and the rest of the trace decision tree have higher similarity is that the similarity of training data is caused instead of the classification capability of the decision tree, so the method is>The trajectory decision tree should have a small weight.
Further, classifying indexes according to the driving track of each track decision treeAnd track differentiation index->Obtaining track prediction weight of each track decision tree>Calculate +.>Track prediction weight of track decision tree>：
In the above formula:、/>respectively represent +.>Travel track classification index, track differentiation index of track decision tree, < >>Is a normalization function.
So far, the track prediction weight of each track decision tree is obtained and is used for obtaining the track prediction result of each automobile passing through the level crossing in the subsequent random forest.
Step S004, track ordinal number pairs of each automobile on each lane of the level crossing and collision probability are obtained according to the track prediction weights, and the safety early warning system sends traffic early warning information to the automobile which is likely to collide.
Taking the track prediction weight of each track decision tree as each track decisionAnd obtaining the track prediction result of the random forest on the input sample by the weight of the tree during voting. Acquiring corresponding traffic information sequences of each automobile entering each lane in cross-type level crossingAs the input of random forest, obtaining the predicted track label of each car, using the data pair composed of the corresponding entering lane number of the car and the corresponding exiting lane number of the predicted track label as the track number pair of each car, and marking the track number pair of the jth car as (>, />) Wherein->Is the label of the driving lane of the j-th automobile driving into the level crossing,/for the vehicle>Is a predictive label of an outgoing lane of a j-th automobile which is driven out of and into a level crossing.
For the crossroad type level crossing, the level crossing consists of four lanes in different directions and a conflict area in the center of the crossing, each direction is two-way four lanes, and the lane schematic diagram of the crossroad type level crossing is shown in fig. 2. Each vehicle that is about to drive into the conflict area has three selectable driving maneuvers, left, straight and right. Wherein the vehicle turns left by the inner lane and the vehicle running on the outer lane can only turn straight and right. In the collision area, collision points exist between the left-turning vehicle and the opposite straight-going vehicle, and between the left-turning vehicle and the crossed straight-going vehicle and the left-turning vehicle; the straight-going vehicle and the crossed straight-going vehicle have collision points; the right-turn vehicle and the vehicles in other directions have no collision points. Therefore, the total of 16 collision points are included in the intersection of the cross-type level road. Therefore, according to the entering direction and the exiting direction of each lane in the cross-shaped level crossing, 12 vehicle track ordinal pairs are obtained, the vehicle track ordinal pairs with track collision are recorded, and a vehicle track interaction table is constructed, as shown in table 1, each row and column in the table has 12 vehicle track ordinal pairs, and the vehicle track ordinal pairs are respectively: (2,3), (2,5), (2,7), (4,1), (4,5), (4,7), (6,1), (6,3), (6,7), (8,1), (8,3), (8,5). The track ordinal pair with position 1 in the table is considered as the track ordinal pair with possible collision, and the track ordinal pair with position 0 in the table is considered as the track ordinal pair without collision.
TABLE 1 vehicle track interaction list
Further, the track ordinal number pair of each automobile is matched with the track ordinal numbers obtained by the other lane vehicles and the track ordinal number pairs in the vehicle track interaction table, and if the matching result is 1, the current predicted vehicle existence probability is considered to generate vehicle collision; if the matching result is 0, the current predicted vehicle absence probability is considered to generate vehicle collision, the matching result of the track ordinal number pair of each vehicle is uploaded to a safety early warning system, the safety early warning system utilizes a voice broadcasting device installed at each intersection to carry out voice reminding on the vehicle, early warning information is generated on the vehicle with the collision probability to remind a driver to carry out track adjustment, then the safety early warning system records running data before and after the early warning information is received by the early warning vehicle, the running data is stored in a database of the safety early warning system, and then early warning information regulations can be formed based on a large amount of running data and are sent to each vehicle.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. The foregoing description of the preferred embodiments of the present invention is not intended to be limiting, but rather, any modifications, equivalents, improvements, etc. that fall within the principles of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A rural level crossing safety pre-warning method is characterized by comprising the following steps:
acquiring a traffic information sequence of each automobile at a level crossing;
taking a set of traffic information sequences of a preset number of vehicles as a training set in a random forest algorithm, and acquiring a track decision tree in a random forest by using the random forest algorithm based on the training set; obtaining the track distribution similarity of each depth on each track decision tree according to the difference of the running tracks between the samples in the left node and the right node under the same depth in each track decision tree; acquiring a driving track classification index of each track decision tree according to sample distribution in a training set of each track decision tree and the track distribution similarity;
obtaining track classification similarity between two track decision trees according to the prediction result of the two track decision trees on the same sample; acquiring a track classification difference coefficient between the two track decision trees according to the difference between the corresponding running information of the samples in the training sets of the two track decision trees; obtaining track prediction weights of each track decision tree according to the track classification similarity and the track classification difference coefficient between the track decision trees;
and acquiring track ordinal pairs of each automobile on each lane of the level crossing and the probability of collision according to track prediction weights of all track decision trees, and sending traffic early warning information to the automobiles which are likely to collide by the safety early warning system.
2. The rural level crossing safety precaution method according to claim 1, wherein the method for obtaining the track distribution similarity of each depth on each track decision tree according to the difference of the running tracks between the samples in the left node and the right node under the same depth in each track decision tree is as follows:
respectively acquiring lane numbers corresponding to each sample in a left node and a right node of each depth on each decision tree by using traffic information sequences corresponding to all samples in a training set of each track decision tree;
and acquiring a measurement distance between a probability histogram constructed by corresponding lane numbers of all samples in a left node of each depth and a probability histogram constructed by corresponding lane numbers of all samples in a right node of each depth, and taking the reciprocal of the sum of the measurement distance and a preset parameter as the track distribution similarity of each depth on each track decision tree.
3. The rural level crossing safety precaution method according to claim 1, wherein the method for obtaining the driving track classification index of each track decision tree according to the sample distribution in the training set of each track decision tree and the track distribution similarity is as follows:
acquiring the dispersion degree of the driving data of each track decision tree training set according to the distribution characteristics of the driving characteristic data in each track decision tree training set;
taking the sum of the average value of the track distribution similarity of each depth on each track decision tree and the preset parameter as a denominator;
and taking the ratio of the dispersion degree of the running data of each track decision tree training set to the denominator as the running track classification index of each track decision tree.
4. The rural level crossing safety precaution method according to claim 3, wherein the method for acquiring the driving data dispersion degree of each track decision tree training set according to the distribution characteristics of the driving characteristic data in each track decision tree training set comprises the following steps:
taking the sum of the product of the standard deviation and the range of each driving characteristic data in the training set of the decision tree according to each track and a preset parameter as a first accumulation factor;
and taking the accumulation of the first accumulation factors in each track decision tree training set as the driving data dispersity of each track decision tree training set.
5. The rural level crossing safety precaution method according to claim 1, wherein the method for obtaining the track classification similarity between two track decision trees according to the prediction results of the two track decision trees on the same sample is as follows:
taking a preset number of samples extracted from a test set as random test samples, taking each random test sample as the input of each track decision tree, and taking a set formed by label values corresponding to the output of each track decision tree according to the extraction sequence as a test track distribution sequence of each track decision tree;
and taking the difference value between elements at the same position in the test track distribution sequence of the two track decision trees as the input of a relation function, and taking the average value of the function values of the relation function on the test track distribution sequence as the track classification similarity between the two track decision trees.
6. The rural level crossing safety precaution method according to claim 1, wherein the method for obtaining the track classification difference coefficient between the two track decision trees according to the difference between the corresponding running information of the samples in the training sets of the two track decision trees is as follows:
acquiring an automobile driving characteristic difference value between two track decision trees according to the distribution difference of the driving characteristic data in the two track decision tree training sets; taking the sum of the difference value of the driving characteristics of the automobile and preset parameters as a second composition factor;
taking the sum of the measurement distance between probability histograms constructed by corresponding lane numbers of all samples in the training sets of the two track decision trees and preset parameters as a first composition factor;
the track classification difference coefficient between the two track decision trees consists of a first composition factor and a second composition factor, wherein the track classification difference coefficient is in a direct proportion relation with the first composition factor and the second composition factor.
7. The rural level crossing safety precaution method according to claim 6, wherein the method for obtaining the automobile driving characteristic difference value between two track decision trees according to the distribution difference of the driving characteristic data in the two track decision tree training sets is as follows:
taking the coding result of the corresponding running characteristic set of each track decision tree training set as the running characteristic coding set of each track decision tree;
the method comprises the steps of taking edit distances between elements in a running feature coding set of two track decision trees as input of a relation function, and taking accumulation of function values of the relation function on all elements in the running feature coding set as an automobile running feature difference value between the two track decision trees.
8. The rural level crossing safety pre-warning method according to claim 1, wherein the method for obtaining the track prediction weight of each track decision tree according to the track classification similarity and the track classification difference coefficient between the track decision trees is as follows:
taking the product of the track classification similarity and the track classification difference coefficient between each track decision tree and the rest of each track decision tree as a first accumulation factor;
taking the average value of the first accumulation factors on each track decision tree and all the rest track decision trees as a track distinguishing index of each track decision tree;
and taking a normalization result of the product of the track distinguishing index and the running track classifying index of each track decision tree as the track prediction weight of each track decision tree.
9. The method for obtaining the track ordinal number pair of each automobile on each lane of the level crossing and the probability of collision according to the track prediction weights of all track decision trees according to the safety precaution method for the rural level crossing according to claim 1 is characterized in that:
taking the track prediction weight of each track decision tree as the weight when voting each track decision tree in a random forest, taking the traffic information sequence of each automobile on each lane of a level crossing as the input of the random forest, acquiring the predicted track label of each automobile by using the random forest, and taking an array pair formed by the driving-in lane number of each automobile level crossing and the lane number corresponding to the predicted track label as the track ordinal number pair of each automobile;
and constructing a track ordinal number pair matching table of the level crossing by using priori knowledge, substituting the track ordinal number pair of each automobile into the track ordinal number pair matching table to obtain a matching result, and enabling the safety early warning system to early warn that the matching result is the automobile which is likely to collide.
10. A rural level crossing safety precaution system comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1-9 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311337787.6A CN117077042B (en) | 2023-10-17 | 2023-10-17 | Rural level crossing safety early warning method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311337787.6A CN117077042B (en) | 2023-10-17 | 2023-10-17 | Rural level crossing safety early warning method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117077042A true CN117077042A (en) | 2023-11-17 |
| CN117077042B CN117077042B (en) | 2024-01-09 |
Family
ID=88715599
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311337787.6A Active CN117077042B (en) | 2023-10-17 | 2023-10-17 | Rural level crossing safety early warning method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117077042B (en) |
Citations (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150332097A1 (en) * | 2014-05-15 | 2015-11-19 | Xerox Corporation | Short-time stopping detection from red light camera videos |
| US20160101779A1 (en) * | 2013-05-31 | 2016-04-14 | Toyota Jidosha Kabushiki Kaisha | Movement trajectory predicting device and movement trajectory predicting method |
| CN108961758A (en) * | 2018-08-03 | 2018-12-07 | 东华理工大学 | A kind of crossing broadening lane detection method promoting decision tree based on gradient |
| CN109017799A (en) * | 2018-04-03 | 2018-12-18 | 张锐明 | A kind of new-energy automobile driving behavior prediction technique |
| CN109496288A (en) * | 2017-07-13 | 2019-03-19 | 北京嘀嘀无限科技发展有限公司 | System and method for determining track |
| CN110288032A (en) * | 2019-06-27 | 2019-09-27 | 武汉中海庭数据技术有限公司 | A kind of vehicle driving trace type detection method and device |
| CN110304075A (en) * | 2019-07-04 | 2019-10-08 | 清华大学 | Track of vehicle prediction technique based on Mix-state DBN and Gaussian process |
| CN110836675A (en) * | 2019-10-25 | 2020-02-25 | 北京航空航天大学 | Decision tree-based automatic driving search decision method |
| CN111386559A (en) * | 2018-10-25 | 2020-07-07 | 北京嘀嘀无限科技发展有限公司 | Method and system for judging whether target road facilities exist at intersection |
| GB202011050D0 (en) * | 2020-07-17 | 2020-09-02 | Continental Automotive Systems Srl | Method of prediction of the trajectory, computing unit, computer program, computer readable medium, and data stream |
| US20210139026A1 (en) * | 2019-11-07 | 2021-05-13 | Hyundai-Aptiv Ad Llc | Trajectory prediction from precomputed or dynamically generated bank of trajectories |
| US20210174668A1 (en) * | 2019-12-10 | 2021-06-10 | Samsung Electronics Co., Ltd. | Systems and methods for trajectory prediction |
| CN113792782A (en) * | 2021-09-13 | 2021-12-14 | 一汽出行科技有限公司 | Track monitoring method and device for operating vehicle, storage medium and computer equipment |
| CN113903173A (en) * | 2021-10-18 | 2022-01-07 | 苏州工业园区测绘地理信息有限公司 | Vehicle track feature extraction method based on directed graph structure and LSTM |
| CN114091581A (en) * | 2021-11-03 | 2022-02-25 | 武汉长江通信智联技术有限公司 | Vehicle operation behavior type identification method based on sparse track |
| CN114169426A (en) * | 2021-12-02 | 2022-03-11 | 安徽庐峰交通科技有限公司 | Beidou position data-based highway traffic potential safety hazard investigation method |
| CN114639243A (en) * | 2022-03-31 | 2022-06-17 | 四川九洲视讯科技有限责任公司 | Intelligent traffic prediction and decision method, system and readable storage medium |
| CN115170611A (en) * | 2022-07-11 | 2022-10-11 | 湖北工业大学 | Complex intersection vehicle driving track analysis method, system and application |
| CN115588310A (en) * | 2022-06-28 | 2023-01-10 | 中国公路工程咨询集团有限公司 | Vehicle collision risk prediction method based on trajectory data |
| CN115631651A (en) * | 2022-09-23 | 2023-01-20 | 北京理工大学 | Automatic driving decision planning system and method for non-control intersection environment |
| CN115743179A (en) * | 2022-11-28 | 2023-03-07 | 长沙汽车创新研究院 | Vehicle probability multi-mode expected trajectory prediction method |
| CN116153108A (en) * | 2022-12-31 | 2023-05-23 | 浙江工业大学 | Method for evaluating safety influence of illumination on intersection by using random forest model |
| CN116434532A (en) * | 2022-12-14 | 2023-07-14 | 杭州电子科技大学 | Intersection track prediction method and device based on strategy intention |
-
2023
- 2023-10-17 CN CN202311337787.6A patent/CN117077042B/en active Active
Patent Citations (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160101779A1 (en) * | 2013-05-31 | 2016-04-14 | Toyota Jidosha Kabushiki Kaisha | Movement trajectory predicting device and movement trajectory predicting method |
| US20150332097A1 (en) * | 2014-05-15 | 2015-11-19 | Xerox Corporation | Short-time stopping detection from red light camera videos |
| CN109496288A (en) * | 2017-07-13 | 2019-03-19 | 北京嘀嘀无限科技发展有限公司 | System and method for determining track |
| CN109017799A (en) * | 2018-04-03 | 2018-12-18 | 张锐明 | A kind of new-energy automobile driving behavior prediction technique |
| CN108961758A (en) * | 2018-08-03 | 2018-12-07 | 东华理工大学 | A kind of crossing broadening lane detection method promoting decision tree based on gradient |
| CN111386559A (en) * | 2018-10-25 | 2020-07-07 | 北京嘀嘀无限科技发展有限公司 | Method and system for judging whether target road facilities exist at intersection |
| CN110288032A (en) * | 2019-06-27 | 2019-09-27 | 武汉中海庭数据技术有限公司 | A kind of vehicle driving trace type detection method and device |
| CN110304075A (en) * | 2019-07-04 | 2019-10-08 | 清华大学 | Track of vehicle prediction technique based on Mix-state DBN and Gaussian process |
| CN110836675A (en) * | 2019-10-25 | 2020-02-25 | 北京航空航天大学 | Decision tree-based automatic driving search decision method |
| US20210139026A1 (en) * | 2019-11-07 | 2021-05-13 | Hyundai-Aptiv Ad Llc | Trajectory prediction from precomputed or dynamically generated bank of trajectories |
| US20210174668A1 (en) * | 2019-12-10 | 2021-06-10 | Samsung Electronics Co., Ltd. | Systems and methods for trajectory prediction |
| GB202011050D0 (en) * | 2020-07-17 | 2020-09-02 | Continental Automotive Systems Srl | Method of prediction of the trajectory, computing unit, computer program, computer readable medium, and data stream |
| CN113792782A (en) * | 2021-09-13 | 2021-12-14 | 一汽出行科技有限公司 | Track monitoring method and device for operating vehicle, storage medium and computer equipment |
| CN113903173A (en) * | 2021-10-18 | 2022-01-07 | 苏州工业园区测绘地理信息有限公司 | Vehicle track feature extraction method based on directed graph structure and LSTM |
| CN114091581A (en) * | 2021-11-03 | 2022-02-25 | 武汉长江通信智联技术有限公司 | Vehicle operation behavior type identification method based on sparse track |
| CN114169426A (en) * | 2021-12-02 | 2022-03-11 | 安徽庐峰交通科技有限公司 | Beidou position data-based highway traffic potential safety hazard investigation method |
| CN114639243A (en) * | 2022-03-31 | 2022-06-17 | 四川九洲视讯科技有限责任公司 | Intelligent traffic prediction and decision method, system and readable storage medium |
| CN115588310A (en) * | 2022-06-28 | 2023-01-10 | 中国公路工程咨询集团有限公司 | Vehicle collision risk prediction method based on trajectory data |
| CN115170611A (en) * | 2022-07-11 | 2022-10-11 | 湖北工业大学 | Complex intersection vehicle driving track analysis method, system and application |
| CN115631651A (en) * | 2022-09-23 | 2023-01-20 | 北京理工大学 | Automatic driving decision planning system and method for non-control intersection environment |
| CN115743179A (en) * | 2022-11-28 | 2023-03-07 | 长沙汽车创新研究院 | Vehicle probability multi-mode expected trajectory prediction method |
| CN116434532A (en) * | 2022-12-14 | 2023-07-14 | 杭州电子科技大学 | Intersection track prediction method and device based on strategy intention |
| CN116153108A (en) * | 2022-12-31 | 2023-05-23 | 浙江工业大学 | Method for evaluating safety influence of illumination on intersection by using random forest model |
Non-Patent Citations (1)
| Title |
|---|
| 胡艳等: "城市道路交叉口防碰撞预警的研究", 无线互联科技 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117077042B (en) | 2024-01-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101002239B (en) | System and method for monitoring driving | |
| CN113486822B (en) | Surrounding vehicle track prediction method and system based on driving intention | |
| CN109840612A (en) | User's driving behavior analysis method and system | |
| CN108091132B (en) | Traffic flow prediction method and device | |
| US20170103121A1 (en) | Drive support apparatus | |
| CN111785023A (en) | Vehicle collision risk early warning method and system | |
| CN111563555A (en) | Driver driving behavior analysis method and system | |
| Zhang et al. | A framework for turning behavior classification at intersections using 3D LIDAR | |
| CN108062699A (en) | Personalized vehicle management method | |
| CN116383685A (en) | Vehicle lane change detection method based on space-time interaction diagram attention network | |
| Xue et al. | A context-aware framework for risky driving behavior evaluation based on trajectory data | |
| CN113581188A (en) | Commercial vehicle driver driving style identification method based on Internet of vehicles data | |
| CN115965235A (en) | Public transport city bus accident risk factor analysis method | |
| CN117077042B (en) | Rural level crossing safety early warning method and system | |
| CN112990563B (en) | Real-time prediction method for rear-end collision accident risk of expressway | |
| CN112435466B (en) | Method and system for predicting take-over time of CACC vehicle changing into traditional vehicle under mixed traffic flow environment | |
| CN108509999A (en) | It is a kind of indignation drive detection and safe early warning method | |
| CN114822044B (en) | Driving safety early warning method and device based on tunnel | |
| CN116946183A (en) | Commercial vehicle driving behavior prediction method considering driving capability and vehicle equipment | |
| Yasaswini et al. | Identifying road accidents severity using convolutional neural networks | |
| CN113192340B (en) | Method, device, equipment and storage medium for identifying highway construction vehicles | |
| CN115860461A (en) | Risk factor evaluation method for traffic conflict of non-motor vehicles at plane intersection | |
| Tanaka et al. | Large-scale driver identification using automobile driving data | |
| Hegazy et al. | Classifying aggressive drivers for better traffic signal control | |
| CN114802264A (en) | Vehicle control method and device and electronic equipment |
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 |