CN113096416B - Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway - Google Patents

Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway Download PDF

Info

Publication number
CN113096416B
CN113096416B CN202110333583.XA CN202110333583A CN113096416B CN 113096416 B CN113096416 B CN 113096416B CN 202110333583 A CN202110333583 A CN 202110333583A CN 113096416 B CN113096416 B CN 113096416B
Authority
CN
China
Prior art keywords
lane
traffic
vehicle
time
automatic driving
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110333583.XA
Other languages
Chinese (zh)
Other versions
CN113096416A (en
Inventor
郝威
吴其育
张兆磊
戎栋磊
易可夫
王正武
刘理
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changsha University of Science and Technology
Original Assignee
Changsha University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Changsha University of Science and Technology filed Critical Changsha University of Science and Technology
Priority to CN202110333583.XA priority Critical patent/CN113096416B/en
Publication of CN113096416A publication Critical patent/CN113096416A/en
Application granted granted Critical
Publication of CN113096416B publication Critical patent/CN113096416B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/07—Controlling traffic signals
    • G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • 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
    • G08G1/0133—Traffic data processing for classifying traffic situation
    • 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
    • G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed

Abstract

The invention discloses a dynamic cooperative control method for variable speed limit of an automatic driving special lane and a general lane in a confluence area on an expressway, which comprises the following steps: generating a traffic running state classifier; establishing preset function curves of three target functions; determining the arrangement number of the automatic driving special lanes; starting a dynamic cooperative control scheme when the traffic running state is crowded; the roadside communication unit issues lane control information special for automatic driving and dynamic variable speed limit control information; the traffic information cloud dynamically adjusts the variable speed limit information, and when the traffic flow tends to be in a stable free flow state, dynamic cooperative management and control are finished. The dynamic cooperative control method for the variable speed limit of the automatic driving special lane and the general lane in the confluence area of the expressway provided by the invention can be used for promoting the adaptability of CAVs by providing the priority right of use for automatic driving vehicles and providing an active management mode for a traffic manager, and can be used for carrying out variable speed limit on the traffic flow of the general lane, thereby achieving the purposes of improving the traffic efficiency and reducing traffic conflicts.

Description

Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway
Technical Field
The invention belongs to the technical field of intelligent traffic, and relates to a dynamic cooperative control method for variable speed limit of an automatic driving special lane and a general lane in a confluence area on a highway.
Background
In recent years, China successively puts forward 'compendium of the traffic compendium' and 'new capital construction' national key strategic demands, and clearly indicates that the future intelligent traffic development needs to be developed around intelligent internet automatic driving automobiles and digital traffic infrastructures. Under the background, the key technology of automatic driving is continuously broken through, automatic driving vehicles start to carry out road tests continuously all over the country, the situation that the automatic driving vehicles and manual driving vehicles are mixed in the future appears for a long time, and the traditional single traffic flow gradually evolves into mixed heterogeneous traffic flow.
At present, the specific characteristics of heterogeneous traffic flow, such as traffic flow stability, operation safety, traffic efficiency, etc., are not clear, and the defects of unclear traffic safety factors and incomplete control measures exist. This drawback is not highlighted in the open flow state, but it is more highlighted with the complication of the composition of the traffic flow and the densification of the traffic density, and when the traffic is crowded, the mixing of the automatic driving vehicles and the manual driving vehicles lowers the utilization rate of road resources and increases the instability of the traffic flow, thereby having a negative influence on the traffic efficiency, traffic safety, running time and the like. Therefore, it is necessary to provide a dynamic exclusive lane for the autonomous vehicle to alleviate the contradiction between the autonomous vehicle and the manually driven vehicle in the complex traffic environment and the travel peak period.
However, currently, common lanes dedicated to motor vehicles, such as a bus lane and a High Occupancy Vehicle (HOV) lane, are fixedly set in different time intervals and different areas, and cannot be adjusted in real time according to the congestion state of the traffic flow, and the management method is to draw a mark on the road by a traffic control department or set a fixed sign at the roadside. The method can not realize flexible and mobile configuration of the special lane, and can also influence the use efficiency of the lane. In addition, more and more research shows that road speed limit is closely related to traffic safety. However, the road speed limit in China is mostly fixed at present, which has negative influence on road traffic efficiency and traffic safety to a certain extent.
The invention provides a dynamic cooperative control method for variable speed limit of an automatic driving special Lane and a general Lane in a confluence area on an expressway, which aims to dynamically lay the automatic driving special Lane (CAV _ Lane) according to a real-time traffic flow state and the permeability of an automatic driving vehicle to separate heterogeneous traffic flows, and simultaneously carry out variable speed limitation on the traffic flow of the general Lane according to real-time traffic flow information, thereby achieving the purposes of improving traffic efficiency and reducing traffic conflicts and running time.
Disclosure of Invention
The invention aims to provide a dynamic cooperative control method for variable speed limit of an automatic driving special lane and a general lane in a confluence area on a highway, which can promote the adaptability of CAVs by providing priority use right for automatic driving vehicles and provide an active management mode for a traffic manager, and can carry out variable speed limit on the traffic flow of the general lane according to real-time traffic flow information so as to achieve the purposes of improving traffic efficiency and reducing traffic conflict and running time; secondly, the CAV vehicle can keep smaller headway, faster running speed and higher stability on a special lane by means of communication technologies such as V2V and V2X, so that traffic safety is improved, road traffic capacity is improved, and the problems that heterogeneous traffic flow running characteristics are not clear, a control method needs to be improved and the like in the future transition to a full-automatic driving environment are solved, and further the problems of low traffic capacity, large potential safety hazard and high running delay are caused.
The technical scheme adopted by the specific embodiment of the invention is that the dynamic cooperative control method for the variable speed limit of the automatic driving special lane and the general lane in the confluence area on the expressway comprises the following steps:
generating a traffic running state classifier; establishing a preset function curve of three objective functions of total traffic capacity, traffic safety agency indexes and total vehicle stopping time; determining the arrangement number of the automatic driving special lanes;
when the traffic running state is a crowded state, a dynamic automatic driving special lane and general lane variable speed-limiting dynamic cooperative control scheme is started, the traffic information cloud inputs the permeability of the automatic driving vehicle at the next moment obtained by prediction into preset function curves of three target functions to obtain function values of the three target functions, and the optimal arrangement number of the automatic driving special lanes at the next moment under the permeability of the automatic driving vehicle is determined;
the roadside communication unit receives the traffic information cloud instruction, issues the automatic driving special lane control information on the dynamic special lane information prompt board, and issues the dynamic variable speed limit control information on the general lane variable speed limit information prompt board; the traffic information cloud calculates traffic flow data in real time, dynamically adjusts variable speed limit information, and transmits closing information of an automatic driving special lane to a roadside communication unit when the traffic flow tends to be in a stable free flow state, automatic driving vehicles and manual driving vehicles on a road start mixed driving, and dynamic cooperative control of the automatic driving special lane and the variable speed limit of a general lane in a confluence area on a highway is finished;
the process for establishing the preset function curve of the traffic safety agent index target function specifically comprises the following steps:
simulation experiments are carried out under each scene model by utilizing SUMO simulation software, road network attribute data and vehicle attribute data corresponding to each scene model are written into the SUMO simulation software, the SUMO simulation software is operated, and the road network attribute data and the vehicle attribute data are output to comprise vehicle information, traffic safety agency indexes and total vehicle stopping time
Figure GDA0003394618510000021
The traffic safety agent index comprises the rear-end collision time; preset autonomous vehicle market penetration P with respective scene modelsACalculating a function value of a traffic safety agent index TIT by taking the rear-end collision time extracted from the corresponding data table as an intermediate quantity as an independent variable, and fitting to obtain a preset function curve of each scene model by taking the traffic safety agent index TIT as a target function;
the road network attribute data comprises road section length, road line type, lane width, lane number and road speed limit;
the vehicle attribute data comprises a vehicle following model, vehicle length, vehicle headway and a vehicle lane change model;
wherein, the automatic driving special lane does not limit the speed; the road speed limit of the general lane adopts a variable speed limit model; fixing and limiting speed on a ramp;
the vehicle information comprises vehicle speed, vehicle travel time, vehicle stop waiting time and lane space occupancy;
the expression of the objective function of the traffic safety agent index TIT is shown as follows:
Figure GDA0003394618510000031
Figure GDA0003394618510000032
in the formula, TTCn(t) time of collision, TTC, of nth vehicle at time t*Is a collision time threshold, delta t is a simulation time step length, N is the total number of vehicles in the simulation, M is a simulation total step length, xn-1(t) is the front vehicle position of the conflict point at the time t, xn(t) the position of the vehicle behind the conflict point at the time t, l the length of the vehicle, vn-1(t) the speed of the vehicle ahead of the conflict point at time t, vnAnd (t) is the rear vehicle speed of the conflict point at the time t.
Further, a traffic running state classifier is generated, specifically:
s1, acquiring an original vehicle data set, performing data preprocessing on the original vehicle data set to obtain a training vehicle data set for state recognition, performing traffic congestion state recognition on the training vehicle data set based on cluster analysis, and acquiring a feature data set with a congestion state label:
s11, acquiring an original vehicle data set: acquiring images of vehicles on the highway at intervals of the same frame number per second, extracting vehicle track data of each image as original vehicle data through image identification, and taking a set of the original vehicle data as an original vehicle data set;
s12, carrying out data preprocessing on the original vehicle data set: the method comprises the steps of screening and integrating original vehicle data in an original vehicle data set, namely screening 3 characteristic data and 1 conventional data from each original vehicle data, wherein the 3 characteristic data are vehicle speed, lane space occupancy and vehicle travel time, the 1 conventional data are standard time, and performing data integration on the screened characteristic data of each original vehicle data to form a vehicle data set after data integration;
s13, carrying out traffic congestion state recognition on the training vehicle data set based on cluster analysis, and acquiring a characteristic data set with a traffic congestion state label;
and S2, generating a traffic running state classifier by using the characteristic data with the traffic congestion state label as the prior knowledge of classification analysis.
Further, in S2, the traffic operation state classifier adopts a traffic operation state classifier based on a BP neural network, and uses a feature data set with a traffic congestion state label as a data set of the BP neural network, and then trains by performing an offline classification evaluation model, learns a traffic state represented by a traffic flow parameter, and finally obtains the traffic operation state classifier, and the specific implementation process includes the following steps:
step S21, establishing traffic behavior based on BP neural networkThe state classifier model: the network topology structure of the established traffic running state classifier model has 3 layers including an input layer, an output layer and a hidden layer; wherein the input layer variable is the average vehicle speed VelAverage occupancy rate OccAnd an average travel time Ttravel(ii) a The output layer variables are clear, crowded and blocked; the number of hidden layer nodes is determined according to the number of input layer nodes and the number of output layer nodes;
s22, building and training a traffic operation state classifier model based on a BP (Back propagation) neural network based on TensorFlow, designing a transfer process, a training function and an activation function of the traffic operation state classifier model based on the BP neural network, dividing a characteristic data set with a traffic congestion state label obtained in S1 into a training set and a prediction set, executing a training process on the traffic operation state classifier model based on the BP neural network by using the training set, calculating a cross entropy loss function according to a Levensberg-Marquardt training algorithm in each training, and then adjusting network parameters according to the cross entropy loss function to obtain the traffic operation state classifier model based on the BP neural network;
and S23, continuously training a traffic running state classifier model based on the BP neural network according to the steps S21-S22, comparing the obtained errors, and selecting a training function with smaller error so as to finally determine the optimal node number, thereby generating the final traffic running state classifier.
Further, establishing a preset function curve of three objective functions of the total traffic capacity, the traffic safety agency index and the total vehicle stopping time, specifically:
firstly, establishing an expressway scene comprising L main line lanes and single ramp lanes, as a modeling scene of a preset function curve of three target functions of total traffic capacity, traffic safety agent indexes and total vehicle stopping time, and setting the traffic capacity C of a downstream lanedownAnd upstream traffic demand DupOf the downstream lane traffic capacity CdownLess than upstream traffic demand DupSimulating that the confluence area is in a crowded state;
second, the autopilot vehicle market penetration P is presetALane l special for automatic drivingATraffic q of traffic lanerampTo generate a plurality of scene models to each preset autonomous vehicle market penetration rate PAAnd respectively fitting to obtain preset function curves of three objective functions of each scene model by taking the total traffic capacity, the traffic safety agency index and the total vehicle stopping time as dependent variables.
Further, the objective function expression of the total traffic capacity Q is shown as follows:
Figure GDA0003394618510000051
wherein Q is the total traffic capacity of the confluence area of the expressway, L is the total number of lanes, and L isAFor the number of lanes dedicated to autonomous driving, L-L accordinglyANumber of general lanes, qmixTraffic volume in general lanes, qAFor autonomous driving lane traffic, qrampIs the traffic volume of the ramp;
wherein the traffic volume q of the autodrive laneASee the following formula:
Figure GDA0003394618510000052
in the formula, PAFor autonomous vehicle market penetration, D is traffic demand per lane, lAFor driveways exclusively for automatic driving, CAThe lane passing capacity is the traffic capacity of the automatic driving special lane;
the general lane traffic is calculated as follows:
Figure GDA0003394618510000053
Figure GDA0003394618510000054
Figure GDA0003394618510000055
Figure GDA0003394618510000056
in the formula (I), the compound is shown in the specification,
Figure GDA0003394618510000057
assigning a ratio in a common lane to the overflowing autonomous vehicle,
Figure GDA0003394618510000058
the traffic capacity of the general-purpose lane when the ratio of the general-purpose lane is p-is allocated to the overflowing autonomous vehicle,
Figure GDA0003394618510000059
the ratio of the allocation of the spilled autonomous vehicles to the common lane is
Figure GDA00033946185100000510
Average headway of a time-universal lane; h isccAn average headway representing the autonomous vehicle following the autonomous vehicle; h isaccRepresenting the average headway of the automatically driven vehicle following the manually driven vehicle; h ismIndicating that the artificial vehicle follows the autonomous vehicle and the average headway of the artificial vehicle.
Further, the functional relationship of the variable rate-limiting model is expressed as follows:
Figure GDA0003394618510000061
Figure GDA0003394618510000062
in the formula, VSL_k(xiT + Δ t) indicates that the k-th general lane is at x at the time of t + Δ tiSpeed limit of (V)k(xi+1And t) represents that the k-th general lane is at x at the moment of ti+1At the average vehicle speed detected by the detector, a represents the desired deceleration of the vehicle, taIndicating the driver perceived reaction time, S the line of sight,
Figure GDA0003394618510000065
denotes the average vehicle length, Occ_k(xi+1And t) represents that the k-th general lane is at x at the moment of ti+1At the average occupancy detected by the detector, k ∈ [1, L-L ∈ [ ]A]When k is 1, it indicates the innermost general lane, and k is L-LATime represents the outermost general lane, Vk(xi-1And t) represents that the k-th general lane is at x at the moment of ti-1Average vehicle speed, V, detected by the detectorsafe(xi-1And t) represents that the k-th general lane is at x at the moment of ti-1At a safe speed value, i.e. vehicle is decelerated to Vsafe(xi-1T) the maximum speed at which a collision with the preceding vehicle can be avoided, Δ t representing the time interval;
and (3) correcting the variable speed limit model: introducing a speed threshold Vmin、VmaxAnd a maximum speed change rate Δ V1、ΔV2Correcting the speed limit values of the same lane at the adjacent time and the same time of the adjacent lane;
and correcting the speed limit value of the same lane at adjacent time as follows:
Figure GDA0003394618510000063
Figure GDA0003394618510000064
in the formula,. DELTA.V1Representing the speed change rate of the same lane at the adjacent time;
and correcting the speed limit value of the adjacent lanes at the same time as shown in the following formula:
Figure GDA0003394618510000071
in the formula, VSL_k(xi，t+Δt)、VSL_k+1(xiT + Δ t) indicates that the k-th and k + 1-th general lanes t + Δ t are at time xiThe limiting value, abs (. circle.) represents the absolute value sign, Δ V2Indicating the rate of change of the speed of the adjacent lanes at the same time.
Further, the target function expression of the total stop time of the vehicle is as follows:
Figure GDA0003394618510000072
in the formula (I), the compound is shown in the specification,
Figure GDA0003394618510000073
for the total time that vehicle n stopped and waited on the ramp area, Q is the total number of vehicles stopped and waited on the ramp area within the simulation time, vnIs the speed of the vehicle n in the total simulation time.
Further, determining the number of the lanes dedicated for automatic driving, specifically:
constructing weight vector A ═ a (a) of preset function curves of three target functions under different scene models1，a2，a3)；a1，a2，a3The importance degrees of the three target functions to lane layout values are respectively set;
constructing the optimal layout quantity vector U-U (U-U) of the automatic driving special lane of the preset function curves of the three target functions under different scene models under different market penetration rates of the automatic driving vehicles1，u2，u3)T；u1，u2，u3The optimal layout number of the automatic driving special lanes of the three target functions under different scenes is respectively set;
and combining the weight vector A and the vector U to obtain an optimal arrangement numerical value R of the automatic driving special lane, namely A.U, and determining the arrangement number of the automatic driving special lane.
Further, the calculation of the predicted driveability of the autonomous vehicle at the next time is as follows:
PA_predict(t+Δt)＝(1+α)PA_real(t)
Figure GDA0003394618510000074
in the formula, PA_real(t) permeability of the autonomous vehicle at the present moment, α attenuation coefficient, PA_real(t- Δ t) is the autonomous vehicle permeability at the previous time.
The specific embodiment of the invention has the beneficial effects that:
(1) the method is characterized in that an automatic driving special lane is dynamically laid through the real-time traffic flow state of a confluence area on the highway and the permeability of automatic driving vehicles at the next moment to dynamically separate heterogeneous traffic flows in real time, so that the automatic driving vehicles run on the special lane, the automatic driving vehicles on the general lane can be reduced or even eliminated, and the negative influence on the traffic flow caused by the mixed running of the automatic driving vehicles and the manual driving vehicles in a crowded traffic state is effectively solved; because the automatic driving vehicle on the automatic driving special lane has the communication technology of V2V and V2X, the automatic driving technology can be fully exerted, the interference of manual driving vehicles is avoided, the vehicle traffic capacity and the running speed are greatly improved, and the traffic time is effectively reduced; due to the fact that automatic driving vehicles are greatly reduced or even not arranged on the general lane, the safety and the stability of traffic flow on the general lane are greatly improved.
(2) The specific embodiment of the invention achieves the purposes of relieving traffic jam, reducing traffic conflict and improving traffic capacity by a dynamic collaborative control method of variable speed limit of the automatic driving special lane and the general lane, and simultaneously performs variable speed limit on the traffic flow of the general lane according to real-time traffic flow information, thereby achieving the purposes of improving traffic efficiency and reducing traffic conflict and running time.
(3) The specific embodiment of the invention formulates a multi-objective function, aims to find the numerical balance point of the layout of the automatic driving lane and balances the passing performance, the safety and the high efficiency; by adopting a dynamic variable speed limit control scheme for the general lane, the collision risk of the traffic flow of the bottleneck road section in the highway confluence area can be effectively reduced, and the proportion of the total stop time increased by vehicles on the average lane is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a topology structure diagram of a BP neural network according to an embodiment of the present invention.
Fig. 2 is a diagram of headways of different vehicle-following modes under heterogeneous traffic flow according to an embodiment of the present invention.
Fig. 3 is a diagram of a dynamic exclusive lane scenario in accordance with an embodiment of the present invention.
Fig. 4 is a diagram of a speed limiting plate and detector deployment configuration in accordance with an embodiment of the present invention.
Fig. 5 is a graph of three preset functions of the objective function in a scenario of a traffic volume on a ramp of 400 vehicles/hour according to an embodiment of the present invention.
Fig. 6 is a graph of three preset function curves of the objective function in a scenario of 600 vehicles/hour on the ramp traffic volume according to the embodiment of the present invention.
Fig. 7 is a preset function graph of three objective functions in a scenario of 800 vehicles/hour on a ramp traffic volume according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The specific embodiment of the invention provides a dynamic cooperative control method for variable speed limit of an automatic driving special lane and a general lane in a confluence area on an expressway, which comprises the following steps:
s1, acquiring an original vehicle data set, performing data preprocessing on the original vehicle data set to obtain a training vehicle data set for state recognition, performing traffic congestion state recognition on the training vehicle data set based on cluster analysis, and acquiring a feature data set with a congestion state label, wherein the method specifically comprises the following steps:
s11, acquiring an original vehicle data set: the method comprises the steps of collecting images of vehicles on the expressway at intervals of the same number of frames per second, extracting vehicle track data of each image through image recognition to serve as original vehicle data, and using a set of the original vehicle data as an original vehicle data set.
The vehicle track data comprises vehicle position data, vehicle speed data, vehicle type data and vehicle head distance data.
In the specific embodiment of the invention, a U.S. 101 highway (u.s.freeway 101) is taken as a specific example for specific analysis, and the steps are as follows:
acquiring an original vehicle data set for training, wherein the data acquisition time is 7:50-13:50 and is 360 minutes in total, the data set comprises vehicle image acquisition of a 101 expressway at intervals of 10 frames per second, vehicle track data is extracted through an image recognition technology and serves as original vehicle data, and a set of the original vehicle data serves as an original vehicle data set, wherein the vehicle track data comprises vehicle position data, vehicle speed data, vehicle type data and vehicle headway data.
S12, carrying out data preprocessing on the original vehicle data set: the method comprises the steps of screening and integrating original vehicle data in an original vehicle data set, namely screening 3 characteristic data and 1 conventional data from each original vehicle data, wherein the 3 characteristic data are vehicle speed, lane space occupancy and vehicle travel time, the 1 conventional data are standard time, carrying out data integration on the characteristic data screened from each original vehicle data to form a vehicle data set after data integration, aggregating the vehicle data set after data integration according to a certain time granularity, taking the aggregated vehicle data set after data integration as a sample, and forming a training vehicle data set for state recognition by using a sample set.
The lane space occupancy is a traffic flow parameter index and describes the ratio of the amount of roads which are already occupied by the vehicle projected on the ground in a specific area at a specific time to the total amount of roads in the area.
The specific embodiment of the invention carries out data preprocessing on an original vehicle data set: the method comprises the steps of screening and integrating original vehicle data in an original vehicle data set, aggregating the vehicle data set after data integration with the time granularity of 30 seconds, taking the aggregated vehicle data set after data integration as a sample, forming 720 training vehicle data sets for state recognition by the sample set, wherein each data set comprises 3 types of characteristic data of vehicle speed, lane space occupancy and vehicle travel time and 1 type of standard time conventional data.
The method specifically comprises the following steps of performing data integration on feature data screened from each original vehicle data:
step S121, performing data standardization processing on the feature data screened from each original vehicle data to obtain a vehicle data set subjected to data standardization processing;
the specific embodiment of the present invention provides a specific data normalization processing method, specifically, a minimum-maximum normalization method is adopted to perform feature transformation, as shown in formula (1), a ratio of a difference between a value of feature data screened from each piece of original vehicle data and a minimum value of a feature data category in which the data is located and a difference between maximum values of feature data categories in which the data is located is calculated as a value after feature transformation, and a value range of the ratio is [0, 1 ]:
Figure GDA0003394618510000101
wherein x isnewRepresenting the feature-transformed values, x representing the value of the feature data screened out for each of the raw vehicle data, xminMinimum value, x, representing the type of the feature data in which each of the screened feature data of the raw vehicle data is locatedmaxAnd representing the maximum value of the feature data category of the feature data screened out by each piece of original vehicle data.
Step S122, removing outlier data in the vehicle data set after data standardization processing;
the specific embodiment of the present invention provides a preferred method for removing Outlier data, which specifically uses a Local Outlier Factor (LOF) algorithm to remove Outlier data, and determines an Outlier according to a Local Outlier Factor LOF degree by calculating a ratio of an average density of positions of sample points around a sample point to a density of the position of the sample point as a Local Outlier Factor LOF, and specifically includes the following steps:
(1) and calculating the Euclidean distance between each feature data in the vehicle data set after data standardization processing and feature data of other same types, and sequencing the Euclidean distances to find the first k neighbors of each feature data.
(2) Calculating the k-th adjacent distance, p, of all the feature dataiIs the ith characteristic data point of the vehicle data set X after the data normalization process,
Figure GDA0003394618510000102
representing characteristic data points piIs the k-th neighbor, the k-th neighbor distance dk(pi) Is defined as piEuclidean distance from its k-th adjacent characteristic data point, i.e.
Figure GDA0003394618510000103
(3) Calculating the reachable distance of each characteristic data point to k neighbors:
Figure GDA0003394618510000111
wherein the content of the first and second substances,
Figure GDA0003394618510000112
is piK neighbors, 1 is more than or equal to a is less than or equal to k,
Figure GDA0003394618510000113
representing characteristic data points
Figure GDA0003394618510000114
Is close to the k-th distance of (c),
Figure GDA0003394618510000115
represents piAnd
Figure GDA0003394618510000116
the euclidean distance of (c).
(4) Calculating the local reachable density of each characteristic data point:
Figure GDA0003394618510000117
in the formula, lrdk(pi) Representing the local reachable density of the ith sample point in the vehicle data set X after data normalization, the higher the density, the higher piThe more likely it is to belong to the same cluster as the surrounding points, conversely, the lower the density the more likely it is an outlier, i.e. the local reachable density is inversely proportional to the probability of becoming an outlier; n is a radical ofk(pi) Representing a sample point piI.e. the number of all points within its k-th neighbourhood, including points at the k-th distance, excluding point pi。
(5) Calculating local outlier factors for all feature data:
Figure GDA0003394618510000118
in the formula, LOFk(pi) Representing the ith characteristic data point p in the data-normalized vehicle data set XiK-th distance neighborhood point Nk(pi) And piThe higher the average ratio of the local achievable densities of points, the more significant the difference in density characteristics between neighboring points and their neighbors, and the greater the probability that they will be outliers.
And step S13, carrying out traffic congestion state recognition on the training vehicle data set based on cluster analysis, and acquiring a feature data set with a traffic congestion state label.
The specific embodiment of the invention provides a specific clustering analysis method, namely, the traffic congestion state identification is carried out based on K-Means algorithm clustering analysis.
After the clustering number u is determined, iterative calculation is carried out by setting u random points as initial centroids. Each training vehicle data sample is then assigned to the nearest centroid, measured using euclidean distance, and a new cluster is recalculated. And (5) repeatedly iterating the calculation process to minimize the objective function, and stopping the algorithm when the iteration termination condition is met. The clustering error sum of squares function of the K-Means clustering method is as follows:
Figure GDA0003394618510000119
wherein J (X, C) represents a clustering error sum of squares function, g represents a total number of training vehicle data samples, u is a number of clusters, and XiFor the ith training vehicle data sample entry, cjFor the jth cluster center, the specific embodiment of the present invention sets u to 3, that is, the three traffic states of clear, congested and blocked are divided.
Figure GDA0003394618510000121
In the formula, cjJ is more than or equal to 1 and less than or equal to u, represents the mean value of all training vehicle data in each category, SjRepresenting the number of training vehicle data samples, x, in class jiRepresenting the ith training vehicle data sample item in the jth category, i is more than or equal to 1 and less than or equal to Sj。
And S2, generating a traffic running state classifier by using the characteristic data with the traffic congestion state label of S1 as the prior knowledge of classification analysis. And classifying the real-time traffic flow by using the generated traffic running state classifier to obtain the actual road traffic running state.
In the specific embodiment of the invention, a traffic running state classifier based on a BP (back propagation) neural network is selected as the traffic running state classifier, a characteristic data set with a traffic congestion state label is used as a data set of the BP neural network, then an offline classification evaluation model is trained to learn the traffic state represented by traffic flow parameters, and finally the traffic running state classifier is obtained to carry out real-time or future migration classification prediction on the traffic flow state under the same road scene, and the specific implementation process comprises the following steps:
step S21, establishing a traffic operation state classifier model based on the BP neural network: the network topology structure of the established traffic operation state classifier model is shown in fig. 1, and has 3 layers including an input layer, an output layer and a hidden layer; wherein the input layer variable is the average vehicle speed VelAverage occupancy rate OccAnd an average travel time Ttravel(ii) a The output layer variables are clear, crowded and blocked; the number of hidden layer nodes is determined according to the number of input layer nodes and the number of output layer nodes.
Through the analysis of the output of the traffic flow parameters in different states, the input layer variable of the traffic operation state classifier model is determined as the average vehicle speed VelAverage occupancy rate OccAnd an average travel time TtravelAs shown in table 1.
TABLE 1 input variables for traffic behavior classifier model
Input variable Inputting variable names
Vel Average vehicle speed (km/h)
Occ Average occupancy (%)
Ttravel Mean time of flight(s)
The traffic congestion state classification forms are various, the traffic running state classifier model adopted by the specific embodiment of the invention can distinguish three congestion states, namely clear, congested and blocked, and the output layer design of the traffic running state classifier model is shown in table 2. When the traffic state is congestion, the control is started until the traffic is recovered, and the ideal result is that no congestion occurs, and if the congestion occurs, the processing method is adopted.
y1 y2 y3 Meaning of output variables
1 0 0 Clear
0 1 0 Congestion of the earth
0 0 1 Blocking of a vessel
The number of hidden layer nodes in the BP neural network is determined according to the number of input layer nodes and the number of output layer nodes, and the determination formula is shown as the formula (7):
Figure GDA0003394618510000131
wherein r is the number of hidden layer nodes, g is the number of input layer nodes, h is the number of output layer nodes, delta is a constant, and a is more than or equal to 1 and less than or equal to 10.
Step S22, building and training a traffic operation state classifier model based on the BP neural network based on TensorFlow, designing a transfer process, a training function and an activation function of the traffic operation state classifier model based on the BP neural network, and the characteristic data set with the traffic congestion state label obtained in the S1 is divided into a training set and a prediction set, the training set is utilized to execute a training process to a traffic operation state classifier model based on the BP neural network, and calculates a cross entropy loss function (matlab toolbox call) according to the Levenberg-Marquardt (LM) algorithm (matlab toolbox call) in each round of training, and then, adjusting network parameters (including the number of nodes of each network layer, a node connection weight parameter w and a node offset value b) according to the cross entropy loss function to obtain a traffic operation state classifier model based on the BP neural network.
The BP neural network is a multilayer network which forwards transmits working signals and backwards transmits error signals based on the Widrop-Hoff learning rule, and the forward transmission subprocesses are shown as formulas (8) and (9):
Figure GDA0003394618510000132
Figure GDA0003394618510000133
in the formula (I), the compound is shown in the specification,
Figure GDA0003394618510000134
is the input of the jth node of the l layer, mlThe number of the nodes in the layer I is,
Figure GDA0003394618510000135
representing the connection weight value between the ith node from level l-1 and the jth node from level l,
Figure GDA0003394618510000136
is the output of the ith node of the l-1 layer,
Figure GDA0003394618510000137
for the bias value of the jth node of level l,
Figure GDA0003394618510000138
for the output of the jth node at level l, f (-) is the activation function.
In the specific embodiment of the invention, a sigmoid function is selected as an activation function, as shown in a formula (10); the Levenberg-Marquardt (LM) algorithm was chosen as the training function train.
Figure GDA0003394618510000141
f (epsilon) represents the activation function,ε represents S in formula 9j (l)。
And S23, continuously training a traffic running state classifier model based on the BP neural network according to the steps S21-S22, comparing the obtained errors, and selecting a training function with smaller error so as to finally determine the optimal node number, thereby generating the final traffic running state classifier. The specific embodiment of the invention finally determines that the hidden layer is 2 layers, and the number of nodes of the input layer, the hidden layer and the output layer is 3-5-3 respectively.
Step S3, respectively establishing preset function curves of three objective functions of total traffic capacity, traffic safety agency indexes and total vehicle stopping time, specifically:
firstly, establishing an expressway scene comprising L main line lanes and single ramp lanes, as a modeling scene of a preset function curve of three target functions of total traffic capacity, traffic safety agent indexes and total vehicle stopping time, and setting the traffic capacity C of a downstream lanedownAnd upstream traffic demand DupOf the downstream lane traffic capacity CdownLess than upstream traffic demand DupSimulating that the confluence area is in a crowded state;
second, the autopilot vehicle market penetration P is presetALane l special for automatic drivingATraffic q of traffic lanerampTo generate a plurality of scene models to each preset autonomous vehicle market penetration rate PAAnd respectively fitting to obtain preset function curves of three objective functions of each scene model by taking the total traffic capacity, the traffic safety agency index and the total vehicle stopping time as dependent variables.
The method for presetting three target function curves in advance enables the market penetration rate P of the real-time automatic driving vehicle fed back by the detector in the step S5ACan be input into preset function curves of three target functions to quickly obtain the market penetration rate P of the automatic driving vehicle in real timeAThe values of the corresponding three target functions are obtained without repeating the operation process of the three target functions every time, so that the operation time and the operation amount are greatly reducedThe speed and efficiency of operation are improved.
In the specific embodiment of the invention, an expressway scene comprising four main lanes and single lane ramps is designed as a modeling scene of preset function curves of three target functions, as shown in fig. 3. When the downstream traffic capacity of the ramp is smaller than the upstream traffic demand time, the ramp is in a crowded state. Setting the traffic capacity C of a downstream lane for simulating that a ramp is in a crowded statedownSet up an upstream traffic demand D for 1800 cars/hour/lane up2000; market penetration P for autonomous vehiclesAEleven sets of 0%/10%/…/100%, driveway l for automatic drivingA0/1/2/3 four groups; meanwhile, the influence of the ramp traffic volume on the running stability of the road section traffic flow is large, and the ramp traffic volumes q are respectively setrampThree sets of 400/600/800 cars/hour/lane, and finally, 132 scene models are generated.
(1) One of the indexes, namely, the overall traffic capacity objective function is an integer nonlinear programming model, and the dynamic special lane control problem is converted into a function optimization problem for solving the optimal number of automatic driving lanes to realize the maximum overall traffic capacity Q of the highway confluence area, which is specifically shown in a formula (11):
Figure GDA0003394618510000151
wherein Q is the total traffic capacity of the confluence area of the expressway, L is the total number of lanes (L is more than or equal to 2), and LAFor the number of lanes dedicated to autonomous driving, L-L accordinglyAThe number of general lanes (0 is less than or equal to l)A≤L)，
Figure GDA0003394618510000156
Traffic volume in general lanes, qAFor autonomous driving lane traffic, qrampIs the traffic volume of the ramp.
Respectively according to the traffic volume q of the general lanemixTraffic volume q of automatic driving laneATraffic q of traffic lanerampThe total traffic capacity Q of the confluence area of the target function highway is subjected to numerical analysis and solution to obtainAnd (3) the optimal special lane arrangement number when the total traffic capacity Q of the highway confluence area under different scene models is maximum, namely scheme one.
In the total traffic capacity Q of the expressway convergence area, the traffic volume Q of the automatic driving laneASee formula (12):
Figure GDA0003394618510000152
in the formula, PAFor the market penetration (%) of the autonomous vehicle, D is the traffic demand of each lane, preset to 2000 vehicles/hour/lane, lane l dedicated for autonomous drivingA0/1/2/3 four groups, CAFor driveway-only capacity (vehicle/hour/lane), CAPreset at 3200 vehicles/hour/lane.
In the total traffic capacity Q of the merging area of the expressway, the calculation of the traffic volume of the general lane is shown in formulas (13) to (16):
Figure GDA0003394618510000153
Figure GDA0003394618510000154
Figure GDA0003394618510000155
Figure GDA0003394618510000161
in the formula (I), the compound is shown in the specification,
Figure GDA0003394618510000162
assigning a ratio in a common lane to the overflowing autonomous vehicle,
Figure GDA0003394618510000163
the ratio of the allocation of the spilled autonomous vehicles to the common lane is
Figure GDA0003394618510000164
The traffic capacity of the general-purpose lane at the time,
Figure GDA0003394618510000165
the ratio of the allocation of the spilled autonomous vehicles to the common lane is
Figure GDA0003394618510000166
Average headway of the time-universal lane.
Ramp traffic qrampPreset to 400, 600, 800 vehicles/hour/lane.
The ratio of the allocation of the overflowed autonomous vehicle to the general lane in equation (16) is
Figure GDA0003394618510000167
Average headway of time-universal lane
Figure GDA0003394618510000168
According to autonomous vehicle market penetration PAAverage headway of heterogeneous traffic flow
Figure GDA0003394618510000169
The calculation is as follows:
firstly, a vehicle following model of heterogeneous traffic flow is established, and 4 vehicle following modes of the heterogeneous traffic flow are established:
for the heterogeneous traffic flow of the specific embodiment of the invention, the arrival of all the automatic driving vehicles and the manual driving vehicles is random, the car following mode of the heterogeneous traffic flow appears in 4 car following modes as shown in fig. 2, and the average head time distances under the 4 car following modes are respectively
Figure GDA00033946185100001610
Wherein the content of the first and second substances,
Figure GDA00033946185100001611
representing the average headway of a human vehicle following an autonomous vehicle,
Figure GDA00033946185100001612
represents the average headway of the artificial vehicle following the artificial vehicle,
Figure GDA00033946185100001613
indicating that the autonomous vehicle follows the average headway of the autonomous vehicle,
Figure GDA00033946185100001614
representing the average headway of an autonomous vehicle following a manned vehicle. In the embodiments of the present invention, the following descriptions are provided
Figure GDA00033946185100001615
hmIndicating that the artificial vehicle follows the autonomous vehicle and the average headway of the artificial vehicle,
Figure GDA00033946185100001616
hcaccindicating that the autonomous vehicle follows the average headway of the autonomous vehicle,
Figure GDA00033946185100001617
haccindicating the average headway of an autonomous vehicle following a manually driven vehicle. Market penetration P for autonomous vehiclesAThe total number of vehicles N ═ N in the descending flowC+NMWherein total number of autonomous vehicles NC＝NCC+NMCTotal number of man-made vehicles NM＝NMM+NCM(ii) a Then N isMM+NCM＝N(1-PA)；NCC+NMC＝NPA；
Figure GDA00033946185100001618
Then, the average head time under the heterogeneous traffic flow is solved according to the 4 vehicle following modes of the heterogeneous traffic flowDistance between
Figure GDA00033946185100001619
As shown in equation (17):
Figure GDA00033946185100001620
in the formula, NCMRepresenting a total number of human vehicles following the autonomous vehicle; n is a radical ofMMRepresenting the total number of human vehicles following the other human-driven vehicles; n is a radical ofCCRepresenting a total number of autonomous vehicles following the other autonomous vehicles; n is a radical ofMCRepresenting the total number of autonomous vehicles following the artificial vehicle,
Figure GDA0003394618510000171
the headway of the artificial vehicle is represented,
Figure GDA0003394618510000172
indicating that the autonomous vehicle follows the headway under the artificial vehicle.
Then, the autonomous vehicle market penetration PAThe value of (A) is set to be 0.1, 0.2, …, 1, and the market penetration rate P of different automatic driving vehicles is obtainedAAverage headway of heterogeneous traffic flow
Figure GDA0003394618510000173
The expression of (c) is represented by the following formula (18):
Figure GDA0003394618510000174
finally, when the ratio of the overflowed autonomous vehicles allocated to the general lane is as follows
Figure GDA0003394618510000175
Will be provided with
Figure GDA0003394618510000176
Substitution in formula (18) to obtain overflowThe ratio of the movable driving vehicle to the general lane is
Figure GDA0003394618510000177
Average headway of time-universal lane
Figure GDA0003394618510000178
The traffic flow in the merging area of the upper ramp of the expressway mostly considers the maximum traffic capacity and fails to take traffic safety and traffic delay into consideration. Based on the above, the embodiment of the invention utilizes the SUMO simulation software to respectively carry out simulation experiments under the different scene models aiming at traffic safety and traffic delay, and measures the traffic safety degree by taking the minimum traffic safety agent index as a target function, namely scheme two; and measuring the traffic efficiency by taking the minimum total stop time of the vehicle as an objective function, namely scheme III. And respectively obtaining different schemes of 3 target functions through numerical analysis and simulation, establishing weight matrixes of the different schemes, and determining a final lane layout scheme.
(2) The second index: simulation experiments are carried out under each scene model by utilizing SUMO simulation software, road network attribute data and vehicle attribute data corresponding to each scene model are written into the SUMO simulation software, the SUMO simulation software is operated, and the road network attribute data and the vehicle attribute data are output to comprise vehicle information, traffic safety agency indexes and total vehicle stopping time
Figure GDA0003394618510000179
The traffic safety agent index comprises the rear-end collision time; preset autonomous vehicle market penetration P with respective scene modelsAAnd calculating a function value of the traffic safety agent index TIT by taking the rear-end collision time TTC extracted from the corresponding data table as an intermediate quantity as an independent variable, and fitting to obtain a preset function curve of each scene model by taking the traffic safety agent index TIT as a target function.
The road network attribute data comprises road section length, road line type, lane width, lane number and road speed limit;
the vehicle attribute data comprises a vehicle following model, vehicle length, vehicle headway and a vehicle lane change model;
wherein, the automatic driving special lane does not limit the speed; the road speed limit of the general lane adopts a variable speed limit model; the fixed speed limit of the ramp is set to be 60 km/h; and forming a dynamic variable speed limit control scheme.
The vehicle information includes a vehicle speed, a vehicle travel time, a vehicle stop waiting time, and a lane space occupancy.
With respective preset autonomous vehicle market penetration PAAs an argument, a traffic safety agent index TIT (Time Integrated Time-to-precision) is used as an objective function, see equations (19) to (20), and the superiority and inferiority of the scheme are measured from the aspect of traffic safety:
Figure GDA0003394618510000181
Figure GDA0003394618510000182
in the formula, TTCn(t) is the rear-end collision time of the nth vehicle at the t moment, TTC is the rear-end collision time threshold, 3s is taken, delta t is the simulation time step length and is set to be 0.1s, N is the total number of vehicles in the simulation, M is the total simulation step length and is set to be 3600s, x is the total simulation step lengthn-1(t) is the front vehicle position of the conflict point at the time t, xn(t) the position of the vehicle behind the conflict point at the time t, l the length of the vehicle, and 5m, vn-1(t) the speed of the vehicle ahead of the conflict point at time t, vnAnd (t) is the rear vehicle speed of the conflict point at the time t.
The traditional speed limit does not consider the downstream traffic state and has a fixed mode, when the traffic volume on the ramp is large, the confluence area is in a crowded state, the upstream is in a smooth state, and if the upstream traffic speed is not subjected to step-by-step speed limit control, the vehicle can not slow down to enter the crowded area, so that the crowded state is further aggravated, and certain risks are generated.
In the setting process of the preset function curves of the second index and the third index, the road speed limit of the general lane adopts a variable speed limit control model, real-time adjustment is carried out according to the monitoring data of the downstream traffic flow detector, and speed limit control is carried out step by step, the speed limit is not fixed speed limit of the existing road, namely, when the downstream occupancy is higher, the calculation result of the speed limit value is lower, otherwise, the calculation result is higher.
The functional relation expression of the variable speed limit control model is shown as formula (21) and formula (22):
Figure GDA0003394618510000183
Figure GDA0003394618510000184
in the formula, VSL_k(xiT + Δ t) indicates that the k-th general lane is at x at the time of t + Δ tiSpeed limit of (V)k(xi+1And t) represents that the k-th general lane is at x at the moment of ti+1Where the average vehicle speed detected by the detector, a represents the desired deceleration of the vehicle, embodiments of the present invention preferably take on a value of 1.57m/s2，taIndicating the perceived reaction time of the driver, the preferred value of the embodiment of the invention is 0.88S, S indicates the viewing distance,
Figure GDA0003394618510000191
representing the average vehicle length, embodiments of the present invention preferably take values of 5m, Occ_k(xi+1And t) represents that the k-th general lane is at x at the moment of ti+1At the average occupancy detected by the detector, k ∈ [1, L-L ∈ [ ]A]When k is 1, it indicates the innermost general lane, and k is L-LATime represents the outermost general lane, Vk(xi-1And t) represents that the k-th general lane is at x at the moment of ti-1Average vehicle speed, V, detected by the detectorsafe(xi-1And t) represents that the k-th general lane is at x at the moment of ti-1At a safe speed value, i.e. vehicle is decelerated to Vsafe(xi-1T) maximum speed at which a collision with the preceding vehicle can be avoided, a specific embodiment of the present invention is to maximize safetyAll conditions, Vsafe(xi-1T) is 0, Δ t represents a time interval, 30s, where details are shown in fig. 4.
And (3) correcting the variable speed limit model: the variable speed-limiting model calculates the speed-limiting value of the current sub-road section at the t + delta t moment by controlling the speed and occupancy rate information of the next sub-road section at the t moment of the area, so that the speed of the current sub-road section is gradually reduced in sections, and the purposes of reducing accident risks and improving traffic safety are achieved. In order to avoid accidents caused by sudden speed drop, the calculated speed limit value of the model needs to be corrected. The specific embodiment of the invention introduces a speed threshold Vmin、VmaxAnd a maximum speed change rate Δ V1、ΔV2The speed limit value of the same lane at the adjacent time and the speed limit value of the same lane at the adjacent time are corrected.
And correcting the speed limit value of the same lane at adjacent time:
Figure GDA0003394618510000192
Figure GDA0003394618510000193
in the formula, Vmm、VmaxRespectively 15km/h, 80km/h, delta V1The speed change rate of the same lane at the adjacent time is expressed, and the value is 15 km/h.
Correcting the speed limit value of the adjacent lanes at the same time:
Figure GDA0003394618510000201
in the formula, VSL_k(xi，t+Δt)、VSL_k+1(xiT + Δ t) denotes the time at which the k-th and k + 1-th general lanes t + Δ t are in xiThe limiting value, abs (. circle.) represents the absolute value sign, Δ V2And the speed change rate of the adjacent lanes at the same time is represented, and the value is 20 km/h.
(3) Third index: carrying out simulation experiment under each scene model by utilizing SUMO simulation software so as toPreset autonomous vehicle market penetration P for each scene modelAAs independent variable, total stop time of vehicle extracted from corresponding data table obtained from simulation result
Figure GDA0003394618510000204
And calculating a function value of the total stopping time of the vehicle as an intermediate quantity, and fitting to obtain a preset function curve of each scene model by taking the total stopping time of the vehicle as a target function. The rest of the data are all the same as the setting process of the preset function curve of the second index.
The target function relational expression (26) of the total stopping time of the vehicle is used for measuring the advantages and disadvantages of the schemes from the aspect of traffic efficiency, the three schemes are jointly restricted to obtain the final scheme, and the selection of different indexes is facilitated according to different weight value settings.
Figure GDA0003394618510000202
In the formula (I), the compound is shown in the specification,
Figure GDA0003394618510000203
for the total time that vehicle n stopped and waited on the ramp area, Q is the total number of vehicles stopped and waited on the ramp area within the simulation time, vnIs the speed of the vehicle n in the total simulation time.
Step S4, determining the number of lanes dedicated for automatic driving:
the distribution principle of the lane layout scheme special for automatic driving is as follows: that is, when the autonomous driving lane routing scheme is enabled, the conventional vehicle can only travel on the general lane, the autonomous driving vehicle will preferentially travel on the autonomous driving lane, and when the autonomous driving vehicle is greater than the exclusive lane capacity, the overflowing autonomous driving vehicle will be assigned to the general lane.
Firstly, constructing weight vector A ═ a (a) of preset function curves of three target functions under different scene models1，a2，a3) (ii) a Then, preset function curves of three target functions under different scene models are constructed for different automatic driving vehiclesThe optimal layout quantity vector U of the driveway special for automatic driving under market permeability is equal to (U)1，u2，u3)T(ii) a And finally, combining the weight vector A and the vector U to obtain the optimal arrangement numerical value R of the automatic driving special lanes, namely A.U, and determining the arrangement number of the automatic driving special lanes.
According to the specific embodiment of the invention, the weight matrix A is determined to be (0.38, 0.38 and 0.24) according to the SUMO simulation analysis result and the expert experience knowledge in the field. a is1，a2，a3The importance degree u of the three objective functions to the lane layout value1，u2，u3The optimal layout quantity of the automatic driving special lanes of the three target functions under different scenes is respectively, A is a row vector of 1X3, U is a column vector of 3X1, R obtained by A and U is a specific number, and the final automatic driving special lane layout scheme is obtained by rounding.
Step S5, sending the detected real-time traffic flow data to a traffic information cloud for preprocessing and storing according to step S1, classifying the traffic flow data stored by the traffic information cloud by using the traffic running state classifier generated in step S2, starting an automatic driving special lane and general lane variable speed-limiting dynamic cooperative control scheme by the traffic information cloud when the traffic running state output by the classifier is a crowded state, and utilizing formulas (27) to (28) by the traffic information cloud to realize the penetration rate P of the automatic driving vehicles at the next moment in the traffic flowA_predict(t + Deltat) and predicting the permeability P of the autonomous vehicle at the next timeA_predict(t + Δ t) is input to the preset function curves of the three objective functions obtained in step S3 to obtain function values of the three objective functions, and the optimal number of lanes for autonomous driving under the permeability of the autonomous vehicle at the next moment is determined by the method for determining the lane layout scheme for autonomous driving in step S4.
PA_predict(t+Δt)＝(1+α)PA_real(t) (27)
Figure GDA0003394618510000211
In the formula, PA_real(t) permeability of the autonomous vehicle at the present moment, α attenuation coefficient, PA_real(t- Δ t) is the autonomous vehicle permeability at the previous time.
Step S6, the road side communication unit receives the traffic information cloud end instruction, the automatic driving special lane control information is issued on the dynamic special lane information prompt board, and the vehicle which is allowed to run on the road is prompted, the artificial vehicle in the automatic driving special lane should be driven away as soon as possible, and the artificial vehicle which does not enter the control road section is prohibited from entering the automatic driving special lane; meanwhile, dynamic variable speed limit control information is issued on the variable speed limit information prompt board of the general lane, all vehicles in the general lane run according to the speed limit value issued by the variable speed limit information prompt board, and the variable speed limit information prompt board of the general lane updates the variable speed limit information in real time every 30 seconds; the road detector monitors and controls road traffic flow in real time, real-time monitored traffic flow information is transmitted to the traffic information cloud end, the traffic information cloud end carries out real-time calculation on traffic flow data, variable speed limit information is dynamically adjusted, when the traffic flow tends to be in a stable free flow state, the traffic information cloud end transmits closing information of the automatic driving special lane to the roadside communication unit, the roadside communication unit transmits information to the dynamic special lane information prompt board, the automatic driving special lane lamp is turned off, automatic driving vehicles and manual driving vehicles on a road start mixed driving, and dynamic cooperative management and control of the automatic driving special lane and the variable speed limit of the general lane in a confluence area on a highway are finished.
And the traffic flow entering the control road section can be monitored in real time through a high-definition camera at the entrance of the confluence area, the artificial vehicles entering the automatic driving special lane and the overspeed driving vehicles on the general lane in violation are captured, the information of the violation vehicles is uploaded to a traffic information cloud end in real time, the traffic information cloud end is interconnected with a traffic police management platform, and the violation vehicles are processed according to laws.
In order to ensure that the vehicles in the general lane can travel at the variable speed limit updated in real time, the layout of the variable speed limit information board and the detector is performed every 0.5km before the downstream of the control section, as shown in detail in fig. 4.
It is noted that, in the present application, relational terms such as first, second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. The term "comprising", without further limitation, means that the element so defined is not excluded from the group consisting of additional identical elements in the process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims (9)

1. The dynamic cooperative control method for the variable speed limit of the automatic driving special lane and the general lane in the confluence area on the expressway is characterized by comprising the following steps of:
generating a traffic running state classifier; establishing a preset function curve of three objective functions of total traffic capacity, traffic safety agency indexes and total vehicle stopping time; determining the arrangement number of the automatic driving special lanes;
when the traffic running state is a crowded state, starting a variable speed-limiting dynamic cooperative control scheme of the automatic driving special lane and the general lane, inputting the permeability of the automatic driving vehicle at the next moment obtained by prediction into preset function curves of three target functions by a traffic information cloud end to obtain function values of the three target functions, and determining the arrangement number of the optimal automatic driving special lane at the permeability of the automatic driving vehicle at the next moment;
the roadside communication unit receives the traffic information cloud instruction, issues the automatic driving special lane control information on the dynamic special lane information prompt board, and issues the dynamic variable speed limit control information on the general lane variable speed limit information prompt board; the traffic information cloud calculates traffic flow data in real time, dynamically adjusts variable speed limit information, and transmits closing information of an automatic driving special lane to a roadside communication unit when the traffic flow tends to be in a stable free flow state, automatic driving vehicles and manual driving vehicles on a road start mixed driving, and dynamic cooperative control of the automatic driving special lane and the variable speed limit of a general lane in a confluence area on a highway is finished;
the process for establishing the preset function curve of the traffic safety agent index target function specifically comprises the following steps:
simulation experiments are carried out under each scene model by utilizing SUMO simulation software, road network attribute data and vehicle attribute data corresponding to each scene model are written into the SUMO simulation software, the SUMO simulation software is operated, and the road network attribute data and the vehicle attribute data are output to comprise vehicle information, traffic safety agency indexes and total vehicle stopping time
Figure FDA0003394618500000011
The traffic safety agent index comprises the rear-end collision time; preset autonomous vehicle market penetration P with respective scene modelsACalculating a function value of a traffic safety agent index TIT by taking the rear-end collision time extracted from the corresponding data table as an intermediate quantity as an independent variable, and fitting to obtain a preset function curve of each scene model by taking the traffic safety agent index TIT as a target function;
the road network attribute data comprises road section length, road line type, lane width, lane number and road speed limit;
the vehicle attribute data comprises a vehicle following model, vehicle length, vehicle headway and a vehicle lane change model;
wherein, the automatic driving special lane does not limit the speed; the road speed limit of the general lane adopts a variable speed limit model; fixing and limiting speed on a ramp;
the vehicle information comprises vehicle speed, vehicle travel time, vehicle stop waiting time and lane space occupancy;
the expression of the objective function of the traffic safety agent index TIT is shown as follows:
Figure FDA0003394618500000021
Figure FDA0003394618500000022
in the formula, TTCn(t) time of collision, TTC, of nth vehicle at time t*Is a collision time threshold, delta t is a simulation time step length, N is the total number of vehicles in the simulation, M is a simulation total step length, xn-1(t) is the front vehicle position of the conflict point at the time t, xn(t) the position of the vehicle behind the conflict point at the time t, l the length of the vehicle, vn-1(t) the speed of the vehicle ahead of the conflict point at time t, vnAnd (t) is the rear vehicle speed of the conflict point at the time t.
2. The dynamic cooperative management and control method for the variable speed limit of the automatic driving special lane and the general lane in the confluence area on the expressway as claimed in claim 1, wherein the traffic state classifier is generated by:
s1, acquiring an original vehicle data set, performing data preprocessing on the original vehicle data set to obtain a training vehicle data set for state recognition, performing traffic congestion state recognition on the training vehicle data set based on cluster analysis, and acquiring a feature data set with a congestion state label:
s11, acquiring an original vehicle data set: acquiring images of vehicles on the highway at intervals of the same frame number per second, extracting vehicle track data of each image as original vehicle data through image identification, and taking a set of the original vehicle data as an original vehicle data set;
s12, carrying out data preprocessing on the original vehicle data set: the method comprises the steps of screening and integrating original vehicle data in an original vehicle data set, namely screening 3 characteristic data and 1 conventional data from each original vehicle data, wherein the 3 characteristic data are vehicle speed, lane space occupancy and vehicle travel time, the 1 conventional data are standard time, and performing data integration on the screened characteristic data of each original vehicle data to form a vehicle data set after data integration;
s13, carrying out traffic congestion state recognition on the training vehicle data set based on cluster analysis, and acquiring a characteristic data set with a traffic congestion state label;
and S2, generating a traffic running state classifier by using the characteristic data with the traffic congestion state label as the prior knowledge of classification analysis.
3. The dynamic cooperative control method for the variable speed limit of the automatic driving dedicated lane and the general lane in the confluence area on the expressway as recited in claim 2, wherein in S2, the traffic status classifier adopts a traffic status classifier based on a BP neural network, and uses a feature data set with a traffic congestion status label as a data set of the BP neural network, and then trains by performing an offline classification evaluation model, learns the traffic status represented by the traffic flow parameters, and finally obtains the traffic status classifier, and the specific implementation process comprises the following steps:
step S21, establishing a traffic operation state classifier model based on the BP neural network: the network topology structure of the established traffic running state classifier model has 3 layers, includingAn input layer, an output layer and a hidden layer; wherein the input layer variable is the average vehicle speed VelAverage occupancy rate OccAnd an average travel time Ttravel(ii) a The output layer variables are clear, crowded and blocked; the number of hidden layer nodes is determined according to the number of input layer nodes and the number of output layer nodes;
s22, building and training a traffic operation state classifier model based on a BP (Back propagation) neural network based on TensorFlow, designing a transfer process, a training function and an activation function of the traffic operation state classifier model based on the BP neural network, dividing a characteristic data set with a traffic congestion state label obtained in S1 into a training set and a prediction set, executing a training process on the traffic operation state classifier model based on the BP neural network by using the training set, calculating a cross entropy loss function according to a Levensberg-Marquardt training algorithm in each training, and then adjusting network parameters according to the cross entropy loss function to obtain the traffic operation state classifier model based on the BP neural network;
and S23, continuously training a traffic running state classifier model based on the BP neural network according to the steps S21-S22, comparing the obtained errors, and selecting a training function with smaller error so as to finally determine the optimal node number, thereby generating the final traffic running state classifier.
4. The dynamic cooperative management and control method for the variable speed limit of the automatic driving special lane and the general lane in the confluence area on the expressway as claimed in claim 1, wherein the preset function curves of three objective functions of total traffic capacity, traffic safety agency index and total vehicle stopping time are established, and specifically:
firstly, establishing an expressway scene comprising L main line lanes and single ramp lanes, as a modeling scene of a preset function curve of three target functions of total traffic capacity, traffic safety agent indexes and total vehicle stopping time, and setting the traffic capacity C of a downstream lanedownAnd upstream traffic demand DupOf the downstream lane traffic capacity CdownLess than upstreamTraffic demand DupSimulating that the confluence area is in a crowded state;
second, the autopilot vehicle market penetration P is presetALane l special for automatic drivingATraffic q of traffic lanerampTo generate a plurality of scene models to each preset autonomous vehicle market penetration rate PAAnd respectively fitting to obtain preset function curves of three objective functions of each scene model by taking the total traffic capacity, the traffic safety agency index and the total vehicle stopping time as dependent variables.
5. The method as claimed in claim 4, wherein the objective function expression of the total traffic capacity Q is as follows:
Figure FDA0003394618500000041
wherein Q is the total traffic capacity of the confluence area of the expressway, L is the total number of lanes, and L isAFor the number of lanes dedicated to autonomous driving, L-L accordinglyANumber of general lanes, qmixTraffic volume in general lanes, qAFor autonomous driving lane traffic, qrampIs the traffic volume of the ramp;
wherein the traffic volume q of the autodrive laneASee the following formula:
Figure FDA0003394618500000042
in the formula, PAFor autonomous vehicle market penetration, D is traffic demand per lane, lAFor driveways exclusively for automatic driving, CAThe lane passing capacity is the traffic capacity of the automatic driving special lane;
the general lane traffic is calculated as follows:
Figure FDA0003394618500000043
Figure FDA0003394618500000044
Figure FDA0003394618500000045
Figure FDA0003394618500000046
in the formula (I), the compound is shown in the specification,
Figure FDA0003394618500000047
assigning a ratio in a common lane to the overflowing autonomous vehicle,
Figure FDA0003394618500000048
the ratio of the allocation of the spilled autonomous vehicles to the common lane is
Figure FDA0003394618500000049
The traffic capacity of the general-purpose lane at the time,
Figure FDA00033946185000000410
the ratio of the allocation of the spilled autonomous vehicles to the common lane is
Figure FDA0003394618500000051
Average headway of a time-universal lane; h isccAn average headway representing the autonomous vehicle following the autonomous vehicle; h isaccRepresenting the average headway of the automatically driven vehicle following the manually driven vehicle; h ismIndicating that an artificial vehicle follows an autonomous vehicle and of an artificial vehicleAnd averaging the headway.
6. The dynamic cooperative management and control method for the variable speed limit of the automatic driving special lane and the general lane in the confluence area on the expressway as claimed in claim 1, wherein the functional relationship of the variable speed limit model is expressed as follows:
Figure FDA0003394618500000052
Figure FDA0003394618500000053
in the formula, VSL_k(xiT + Δ t) indicates that the k-th general lane is at x at the time of t + Δ tiSpeed limit of (V)k(xi+1And t) represents that the k-th general lane is at x at the moment of ti+1At the average vehicle speed detected by the detector, a represents the desired deceleration of the vehicle, taIndicating the driver perceived reaction time, S the line of sight,
Figure DEST_PATH_GDA0003394618510000065
denotes the average vehicle length, Occ_k(xi+1And t) represents that the k-th general lane is at x at the moment of ti+1At the average occupancy detected by the detector, k ∈ [1, L-L ∈ [ ]A]When k is 1, it indicates the innermost general lane, and k is L-LATime represents the outermost general lane, Vk(xi-1And t) represents that the k-th general lane is at x at the moment of ti-1Average vehicle speed, V, detected by the detectorsafe(xi-1And t) represents that the k-th general lane is at x at the moment of ti-1At a safe speed value, i.e. vehicle is decelerated to Vsafe(xi-1T) the maximum speed at which a collision with the preceding vehicle can be avoided, Δ t representing the time interval;
and (3) correcting the variable speed limit model: introducing a speed threshold Vmin、VmaxAnd a maximum speed change rate Δ V1、ΔV2Correcting the speed limit values of the same lane at the adjacent time and the same time of the adjacent lane;
and correcting the speed limit value of the same lane at adjacent time as follows:
Figure FDA0003394618500000054
Figure FDA0003394618500000061
in the formula,. DELTA.V1Representing the speed change rate of the same lane at the adjacent time;
and correcting the speed limit value of the adjacent lanes at the same time as shown in the following formula:
Figure FDA0003394618500000062
in the formula, VSL_k(xi,t+Δt)、VSL_k+1(xiT + Δ t) indicates that the k-th and k + 1-th general lanes t + Δ t are at time xiThe limiting value, abs (. circle.) represents the absolute value sign, Δ V2Indicating the rate of change of the speed of the adjacent lanes at the same time.
7. The dynamic cooperative management and control method for the variable speed limit of the automatic driving special lane and the general lane used in the confluence area on the expressway of claim 4, wherein the target function expression of the total stopping time of the vehicle is as follows:
Figure FDA0003394618500000063
in the formula (I), the compound is shown in the specification,
Figure FDA0003394618500000064
for vehicles n in the ramp zoneTotal time of stopping and waiting in the field, Q is the total number of vehicles staying and waiting in the ramp area in the simulation time, vnIs the speed of the vehicle n in the total simulation time.
8. The dynamic cooperative management and control method for the variable speed limit of the driveway and the general lane for the confluence area on the expressway as claimed in claim 1, wherein the determining the number of the driveway for automatic use is specifically as follows:
constructing weight vector A ═ a (a) of preset function curves of three target functions under different scene models1,a2,a3)；a1,a2,a3The importance degrees of the three target functions to lane layout values are respectively set;
constructing the optimal layout quantity vector U-U (U-U) of the automatic driving special lane of the preset function curves of the three target functions under different scene models under different market penetration rates of the automatic driving vehicles1,u2,u3)T；u1,u2,u3The optimal layout number of the automatic driving special lanes of the three target functions under different scenes is respectively set;
and combining the weight vector A and the vector U to obtain an optimal arrangement numerical value R of the automatic driving special lane, namely A.U, and determining the arrangement number of the automatic driving special lane.
9. The method as claimed in claim 1, wherein the predicted penetration of the autonomous vehicle at the next time is calculated as follows:
PA_predict(t+Δt)＝(1+α)PA_real(t)
Figure FDA0003394618500000071
in the formula, PA_real(t) is the permeability of the autonomous vehicle at the present moment, and α isAttenuation coefficient, PA_real(t- Δ t) is the autonomous vehicle permeability at the previous time.
CN202110333583.XA 2021-03-29 2021-03-29 Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway Active CN113096416B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110333583.XA CN113096416B (en) 2021-03-29 2021-03-29 Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110333583.XA CN113096416B (en) 2021-03-29 2021-03-29 Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway

Publications (2)

Publication Number Publication Date
CN113096416A CN113096416A (en) 2021-07-09
CN113096416B true CN113096416B (en) 2022-02-15

Family

ID=76670413

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110333583.XA Active CN113096416B (en) 2021-03-29 2021-03-29 Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway

Country Status (1)

Country Link
CN (1) CN113096416B (en)

Families Citing this family (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113327448B (en) * 2021-08-02 2021-10-29 长沙理工大学 Vehicle track optimization method under special phase for automatic driving
CN113345241B (en) * 2021-08-05 2021-11-09 华砺智行(武汉)科技有限公司 Distributed intersection lane occupancy fusion estimation method and system
CN113393690B (en) * 2021-08-17 2021-11-19 长沙理工大学 Hybrid traffic balance distribution method considering automatic driving special lane
CN113763708B (en) * 2021-08-20 2022-11-04 东南大学 Multi-type special lane control method based on traffic simulation
CN113674546B (en) * 2021-08-20 2022-06-21 东南大学 Sumo simulation-based highway management and control method in V2X environment
CN113689703B (en) * 2021-09-06 2022-06-28 季华实验室 Vehicle shunting control method and device, electronic equipment and storage medium
CN114067559B (en) * 2021-09-27 2022-11-15 北京交通大学 Confluence optimization control method for merging special lane for automatic vehicle into common lane
CN114255586A (en) * 2021-11-12 2022-03-29 浙江银江智慧交通工程技术研究院有限公司 Method and system for optimizing multi-mode network traffic distribution under open strategy
CN114627647A (en) * 2022-03-16 2022-06-14 重庆大学 Hybrid traffic flow optimization control method based on combination of variable speed limit and lane change
CN114758513B (en) * 2022-04-14 2023-05-05 清华大学 Cooperative control method and device for ramp converging, electronic equipment and storage medium
CN115035734B (en) * 2022-05-18 2023-05-23 北京航空航天大学 Urban expressway entrance ramp signal control method based on congestion state classification
CN115116225B (en) * 2022-06-23 2023-08-04 上海交通大学 Data-driven random model predictive control method for mixed traffic flow
CN115148016A (en) * 2022-06-27 2022-10-04 云控智行(上海)汽车科技有限公司 Highway CAV vehicle-specific lane control algorithm and system based on intelligent internet automobile permeability monitoring
CN114997451A (en) * 2022-08-05 2022-09-02 深圳市城市交通规划设计研究中心股份有限公司 Reserved passing method and computing device for automatically driving vehicle in highway scene
CN115410369B (en) * 2022-08-30 2024-02-20 合肥工业大学 Real-time road traffic speed interval construction method
CN115512537A (en) * 2022-09-13 2022-12-23 中远海运科技股份有限公司 Expressway monitoring method and system based on deep reinforcement learning algorithm
CN115953893B (en) * 2022-11-30 2024-01-30 东南大学 Expressway traffic flow stability analysis method in heterogeneous traffic environment
CN116229706B (en) * 2022-12-07 2024-03-26 南京理工大学 Lane-dividing variable speed limiting control method based on intelligent network-connected special lane environment
CN115601954B (en) * 2022-12-13 2023-03-31 西南交通大学 Lane change judgment method, device, equipment and medium for intelligent networked fleet
CN116246466B (en) * 2023-03-13 2024-01-23 长安大学 Hybrid traffic flow management method and system considering autopilot multimode characteristics
CN116238497B (en) * 2023-05-11 2024-01-26 安徽蔚来智驾科技有限公司 Vehicle speed limit control method, electronic device, storage medium and vehicle
CN116363905B (en) * 2023-05-19 2023-09-05 吉林大学 Heterogeneous traffic flow converging region lane change time judging and active safety control method

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108399762A (en) * 2018-05-08 2018-08-14 北京航空航天大学 A kind of automatic Pilot and pilot steering vehicle mix the intersection control method under the conditions of row
CN109949587A (en) * 2019-03-25 2019-06-28 武汉理工大学 A kind of Adjacent Intersections public transportation lane signal coordinated control optimization method
CN110503823A (en) * 2019-07-11 2019-11-26 江苏大学 A kind of automatic driving vehicle dedicated Lanes control system and method
CN110910669A (en) * 2019-11-28 2020-03-24 深圳成谷科技有限公司 Virtual isolation-based control method and device for automatic driving special lane
CN112116822A (en) * 2020-09-21 2020-12-22 长沙理工大学 Expressway traffic capacity cooperative regulation and control method based on CAVs mixed traffic flow lane dynamic allocation
CN112201070A (en) * 2020-09-29 2021-01-08 上海交通大学 Deep learning-based automatic driving expressway bottleneck section behavior decision method
CN112258849A (en) * 2020-10-26 2021-01-22 上海市城市建设设计研究总院（集团）有限公司 Dynamic automatic driving special lane based on expressway and use method thereof
CN112298195A (en) * 2019-07-30 2021-02-02 上海丰豹商务咨询有限公司 Automatic and manual mode switching system for automatic driving special lane of bidirectional 6-10 lane highway and using method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020014131A1 (en) * 2018-07-10 2020-01-16 Cavh Llc Connected automated vehicle highway systems and methods related to transit vehicles and systems

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108399762A (en) * 2018-05-08 2018-08-14 北京航空航天大学 A kind of automatic Pilot and pilot steering vehicle mix the intersection control method under the conditions of row
CN109949587A (en) * 2019-03-25 2019-06-28 武汉理工大学 A kind of Adjacent Intersections public transportation lane signal coordinated control optimization method
CN110503823A (en) * 2019-07-11 2019-11-26 江苏大学 A kind of automatic driving vehicle dedicated Lanes control system and method
WO2021004222A1 (en) * 2019-07-11 2021-01-14 江苏大学 Control system and method for dedicated lane for autonomous vehicle
CN112298195A (en) * 2019-07-30 2021-02-02 上海丰豹商务咨询有限公司 Automatic and manual mode switching system for automatic driving special lane of bidirectional 6-10 lane highway and using method
CN110910669A (en) * 2019-11-28 2020-03-24 深圳成谷科技有限公司 Virtual isolation-based control method and device for automatic driving special lane
CN112116822A (en) * 2020-09-21 2020-12-22 长沙理工大学 Expressway traffic capacity cooperative regulation and control method based on CAVs mixed traffic flow lane dynamic allocation
CN112201070A (en) * 2020-09-29 2021-01-08 上海交通大学 Deep learning-based automatic driving expressway bottleneck section behavior decision method
CN112258849A (en) * 2020-10-26 2021-01-22 上海市城市建设设计研究总院（集团）有限公司 Dynamic automatic driving special lane based on expressway and use method thereof

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
双车道自动-手动驾驶汽车混合交通流博弈模型及其仿真;魏修建等;《系统工程》;20181128(第11期);全文 *
面向新型混合交通流的快速路合流区通行能力建模;胡笳 等;《交通信息与安全》;20210131;第39卷(第1期);全文 *

Also Published As

Publication number Publication date
CN113096416A (en) 2021-07-09

Similar Documents

Publication Publication Date Title
CN113096416B (en) Dynamic cooperative control method for variable speed limit of automatic driving special lane and general lane in confluence area on expressway
CN112116822B (en) Expressway traffic capacity cooperative regulation and control method based on CAVs mixed traffic flow lane dynamic allocation
Liu et al. Trajectory planning for autonomous intersection management of connected vehicles
CN110823235B (en) Intelligent vehicle navigation system, method and control logic for deriving road segment speed limits
CN108877256B (en) Wireless communication-based method for controlling scattered cooperative self-adaptive cruise near intersection
WO2021218235A1 (en) Road traffic jam early warning method and system
CN113781788B (en) Automatic driving vehicle management method based on stability and safety
CN113345268A (en) CAV lane change decision-making method for expressway down-ramp diversion area based on automatic driving special lane deployment scene
CN111695418A (en) Method and system for safe driving based on road condition detection
CN110544380A (en) real-time lane-level safety situation assessment method for road confluence area
CN111243296B (en) Ramp confluence cooperative control method and system based on confluence time optimization
CN110182217A (en) A kind of traveling task complexity quantitative estimation method towards complicated scene of overtaking other vehicles
DE102021132722A1 (en) SELECTING TEST SCENARIOS TO EVALUATE AUTONOMOUS VEHICLE PERFORMANCE
CN108711285A (en) A kind of mixed traffic analogy method based on road cross
Jehad et al. Modelling and optimizing of electronic toll collection (ETC) at Malaysian toll plazas using microsimulation models
Pan et al. Research on the behavior decision of connected and autonomous vehicle at the unsignalized intersection
CN107622309B (en) Road congestion detection method based on VANETs and improved D-S evidence theory
Ito et al. Multi-vehicle coordination on merging roads based on pseudo-perturbation-based broadcast control
Si et al. Optimization model and algorithm for mixed traffic of urban road network with flow interference
CN116013101B (en) System and method for suggesting speed of signal-free intersection based on network environment
Ramaswamy et al. Lane assignment on automated highway systems
CN108241866A (en) A kind of method, apparatus guided to driving behavior and vehicle
CN114475607B (en) Method and device for changing lanes for automatic driving vehicle, vehicle and storage medium
DE102022103059A1 (en) AUTOMATIC ANNOTATION OF PASSABLE ROAD SEGMENTS
CN116033387B (en) Road environment collaborative perception decision-making method and device

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