WO2021203972A1 - 基于碰撞模型的风险预防方法、装置、设备及存储介质 - Google Patents

基于碰撞模型的风险预防方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2021203972A1
WO2021203972A1 PCT/CN2021/082677 CN2021082677W WO2021203972A1 WO 2021203972 A1 WO2021203972 A1 WO 2021203972A1 CN 2021082677 W CN2021082677 W CN 2021082677W WO 2021203972 A1 WO2021203972 A1 WO 2021203972A1
Authority
WO
WIPO (PCT)
Prior art keywords
model
traffic
collision
parameters
simulation
Prior art date
Application number
PCT/CN2021/082677
Other languages
English (en)
French (fr)
Inventor
王健宗
程宁
Original Assignee
平安科技(深圳)有限公司
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 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2021203972A1 publication Critical patent/WO2021203972A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00Computer-aided design [CAD]
    • G06F30/20Design optimisation, verification or simulation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063Operations research, analysis or management
    • G06Q10/0635Risk analysis of enterprise or organisation activities
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40Business processes related to the transportation industry
    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G1/00Traffic control systems for road vehicles
    • G08G1/01Detecting movement of traffic to be counted or controlled
    • G08G1/0104Measuring and analyzing of parameters relative to traffic conditions
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02TCLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00Road transport of goods or passengers
    • Y02T10/10Internal combustion engine [ICE] based vehicles
    • Y02T10/40Engine management systems

Definitions

  • This application relates to the field of artificial intelligence, and in particular to a risk prevention method, device, equipment and storage medium based on a collision model.
  • the main purpose of this application is to solve the problem that there is no corresponding accident risk model for the mixed flow in the case of intelligent network connection to prevent the risk of traffic accidents and reduce the incidence of traffic accidents.
  • the first aspect of this application provides a risk prevention method based on a collision model, including:
  • a logistic regression algorithm is used to construct a collision risk model
  • the percentage of successful predicted traffic accidents of the collision risk model is calculated, and active safety-related intervention measures are executed based on the percentage of successful predicted traffic accidents.
  • the second aspect of the present application provides a risk prevention device based on a collision model, including:
  • the traffic simulation module is used to perform traffic simulation on the preset mixed flow and output the detector data and trajectory data collected during the traffic simulation process;
  • the data processing module is used to perform data preprocessing on the detector data to obtain a set of related parameters of the traffic scene, and perform data preprocessing on the trajectory data to obtain the number of accidents;
  • the model creation module is used to sequentially perform regression analysis on the traffic scene related parameters in the traffic scene related parameter set to obtain a significant correlation parameter set, and determine the most critical parameters of the model to be established according to the significant correlation parameter set; State the most critical parameters and the number of accidents, and use the logistic regression algorithm to build a collision risk model;
  • the risk prevention module is used to calculate the percentage of successful predicted traffic accidents of the collision risk model, and execute active safety-related intervention measures based on the percentage of successful predicted traffic accidents.
  • a third aspect of the present application provides a risk prevention device based on a collision model, including: a memory and at least one processor, the memory stores instructions; the at least one processor calls the instructions in the memory, So that the collision model-based risk prevention equipment executes the steps of the collision model-based risk prevention method as described below:
  • a logistic regression algorithm is used to construct a collision risk model
  • the percentage of successful predicted traffic accidents of the collision risk model is calculated, and active safety-related intervention measures are executed based on the percentage of successful predicted traffic accidents.
  • the fourth aspect of the present application provides a computer-readable storage medium having instructions stored in the computer-readable storage medium, which when run on a computer, cause the computer to execute the following collision model-based risk prevention method A step of:
  • a logistic regression algorithm is used to construct a collision risk model
  • the percentage of successful predicted traffic accidents of the collision risk model is calculated, and active safety-related intervention measures are executed based on the percentage of successful predicted traffic accidents.
  • the simulation output data is preprocessed by performing a traffic simulation on a mixed flow, and a collision risk model is constructed based on the preprocessed simulation output data, and finally the successful prediction of a traffic accident of the collision risk model is calculated Percentage, implement active safety-related intervention measures based on the percentage of successful predicted traffic accidents, thereby effectively preventing potential traffic accident risks and reducing the incidence of traffic accidents.
  • FIG. 1 is a schematic diagram of an embodiment of a risk prevention method based on a collision model in an embodiment of the application
  • FIG. 2 is a schematic diagram of an embodiment of a risk prevention device based on a collision model in an embodiment of the application
  • Fig. 3 is a schematic diagram of an embodiment of a risk prevention device based on a collision model in an embodiment of the application.
  • the embodiments of the present application provide a risk prevention method, device, equipment, and storage medium based on a collision model, which can effectively prevent potential traffic accident risks and reduce the incidence of traffic accidents.
  • An embodiment of the risk prevention method based on the collision model in the embodiment of the present application includes:
  • VISSIM simulation software is used for traffic simulation.
  • VISSIM is a microscopic simulation modeling tool based on time interval and driving behavior, which is used for traffic modeling of urban traffic and public transportation operations. It can analyze the operating conditions of urban traffic and public transportation under various traffic conditions, such as lane settings, traffic composition, traffic signals, bus stops, etc., and comprehensively consider various influences.
  • the factors of road network operation such as lane types, traffic composition , Signal control configuration, parking to give way, and at the same time.
  • Pedestrian models can be added to form the interaction between vehicles and pedestrians. Under the condition that various influencing factors are set, various data can be exported through complete cycle operation.
  • the VISSIM simulation software is internally composed of two parts: a traffic simulator and a signal state generator, which exchange detector data and signal state information through an interface.
  • the traffic simulator is a microscopic traffic simulation model, which includes a car following model and a lane change model.
  • the signal state generator is a signal control software that can realize the control logic of traffic flow through programs. The logic extracts detector data from the traffic simulator in each discrete time interval (which can be 1-0.1 seconds) to determine the signal state in the next simulation second. At the same time, the signal status information is returned to the traffic simulator.
  • VISSIM can not only generate visualized traffic conditions online, but also output various statistical data offline, such as travel time, queue length, etc.
  • the core model of the VISSIM simulation software which is the physiological-psychological driving model, allows VISSIM to simulate the real situation more realistically.
  • the basic idea of this model is: once the driver of the following vehicle thinks that the distance between him and the vehicle in front is less than At its psychological (safe) distance, the driver of the following vehicle starts to slow down. Since the driver of the following vehicle cannot accurately determine the speed of the preceding vehicle, the speed of the following vehicle will be lower than the speed of the preceding vehicle for a period of time, until the distance between the front and rear workshops reaches another psychological (safe) distance, the following driver begins to accelerate slowly. This is repeated, forming an iterative process of acceleration and deceleration.
  • the mixed flow described in this embodiment refers to mixed traffic flow, which refers to the phenomenon of mixed traffic of non-motor vehicles and motor vehicles, and motor vehicles with very different performances.
  • the mixed traffic flow of motor vehicles and non-motor vehicles and pure motor vehicle traffic There is no essential difference in flow. They are all behaviors on public roads that stem from the travel needs of individual traffic, and they are all dynamic systems that are discrete in time and space.
  • the interaction between individuals in a mixed traffic flow is very complex, and its complexity far exceeds that of a single type of traffic flow. This is mainly because different types of vehicles differ greatly in geometric dimensions, driving speeds, power characteristics, and safety requirements. , Which will lead to different driving rules for different types of vehicles. Therefore, the complexity of the composition and behavior of mixed traffic flow determines that its operating characteristics are quite different from pure motor vehicle flow.
  • the traffic simulation parameters directly affect the simulation results. For example, for a scene that simulates mixed traffic flow, draw the simulation in the VISSIM simulation software. Road section, setting the type of vehicle and its expected speed, assigning the driving route, setting and configuring the detector, etc.
  • the VISSIM simulation software will collect two types of data, one is the data collected by the detector, and the other is the trajectory data file.
  • the VISSIM simulation software completes the traffic simulation of the mixed flow, the data and trajectory data collected by the corresponding detectors are exported to the local storage in the form of files.
  • the above 101 further includes the following steps:
  • Simulating traffic simulation is performed on the mixed flow simulation initial model based on the simulation parameters.
  • the basic composition of a transportation system is the road network.
  • the drawing of simulated road sections in VISSIM is realized by the road network editor, while the use of VISSIM’s road network editor is relatively simple and convenient. It does not have a fixed template. You can draw a variety of road networks according to your needs. Sometimes due to the needs of the actual situation, the lane width or turning radius is not an ideal standard. Then you can adjust it according to the actual map of the road network. At this time, we can change the mixed flow
  • the actual road network map is imported into VISSIM as a base map.
  • the base map can be imported into multiple formats, and then the road network can be edited on the map, so that the simulation made is consistent with the actual road network.
  • the input of the intelligent networked vehicle is completed through the external driver model interface in VISSIM.
  • a new vehicle model is established in the vehicle type attribute under the vehicle definition module and the compiled ddl file is imported in the external driver model option.
  • Various types of cars can be defined in the ddl file according to the actual situation. For example, common taxis, cars, and small SUVs belong to the category of cars, so you need to define the car category as a car in the corresponding ddl file, and Define various branch types under the sub-branch, such as its length, width, height, and so on.
  • the desired speed for various types of vehicles in different road sections where the speed is expressed in the form of an interval, which can be adjusted.
  • the expected speed of the car type on the urban road segment is defined as [40, 60]
  • the expected speed of the school area road segment is defined as [10, 30]
  • the unit is km/h.
  • the traffic volume definition module set the traffic volume and the traffic proportion of different vehicle types, and allocate traffic volume for road sections.
  • the traffic flow is expressed in numerical form. For example, if the traffic flow is set to 1000, it means that there are a total of 1000 cars in the entire traffic simulation system; the traffic proportion is expressed in the form of a percentage, such as setting the traffic proportions of different models such as cars It accounts for 95%, buses account for 2%, freight trucks account for 2%, and other types account for 1%.
  • Assign traffic volume to the road segment divide the simulated road segment into multiple paths and add signs to them, such as path a, path b, and path c, and then select the specified path in the road segment selection options under the traffic volume definition module, and finally in the traffic Configure the traffic volume of the road section in the volume attribute, for example, select path a, configure the traffic volume of this path to be 10%, and the traffic volume in the entire simulation is 1000, then the traffic volume in path a is 100.
  • the traffic priority means that the driver finds that the speed of the vehicle in front will hinder his driving while driving. At the same time, when the road conditions ahead are good, the driver will choose to overtake , Then you need to change lanes. When changing lanes, you need to pay attention to the situation of your own lane and adjacent lanes. You can set it in this module. The user can customize the change method and change position. You can also set a certain lot that does not allow lanes Transform.
  • the method further includes:
  • the road section detector is set every 1Km on the road section to be detected, the collection time interval is set to 30s, and the data collection parameters of the road section detector are set, including the average speed of the vehicle, the average occupancy rate, and the average The number of passing vehicles.
  • the detector data is the data collected by the above-mentioned road section detector set on the road section to be detected, which mainly includes the average speed, average occupancy rate and average number of passing vehicles, and is stored in the coil file path of VISSIM in the form of a coil file. After sorting the data in the coil file, a collection of relevant parameters of the traffic scene is obtained.
  • Trajectory data uses SSAM software to set different TTC (Timetocollison, time to collision) thresholds to get the number of rear-end collisions.
  • SSAM software is a simulation conflict analysis software. The software can perform simulation conflict analysis on the vehicle trajectory files output by the four simulation models of VISSIM, PARAMICS, TEXAS and AIMSUN, and can output different types of simulation conflict data, including the number and location of simulation conflicts. And the severity, the number of rear-end collisions in this embodiment is the number of simulated conflicts.
  • the above 102 further includes the following steps:
  • two consecutive upstream and downstream data are integrated into traffic scene related parameters with a time interval of 5 minutes.
  • the two consecutive upstream and downstream data refer to the two consecutive data executed by the detector.
  • Collection, the collection time interval is the collection time interval set by the above-mentioned road section detector.
  • the data five minutes before the accident will be used as the precursor data of the accident, so that the precursor data collected can be used for the prediction of traffic accidents.
  • the first collected data and the second collected data of a certain road section detector are integrated into the relevant parameters of the traffic scene with a time interval of 5 minutes, and the first interval and the second interval are obtained. If it occurs within 5 minutes of the second interval If there is a traffic accident, then the traffic scene-related data corresponding to the first interval, that is, the first collected data will be used as the accident precursor data.
  • a small number of effective parameters are selected from the traffic scene related parameter set to establish a risk model.
  • the above 103 further includes the following steps:
  • the parameters are the most critical parameters of the model to be established.
  • the receiver operation curve referred to as ROC curve
  • ROC curve is often used in artificial intelligence scenarios to evaluate the performance of the classifier in the neural network model.
  • Its main analysis tool is a curve drawn on a two-dimensional plane-ROCcurve.
  • the abscissa of the plane is falsepositiverate (FPR), and the ordinate is truepositiverate (TPR).
  • FPR falsepositiverate
  • TPR truepositiverate
  • TPR is the ratio of all samples that are actually positive that are correctly judged as positive
  • FPR is the ratio of all samples that are actually negative.
  • the rate of false positives For a certain classifier, we can get a TPR and FPR point pair based on its performance on the test sample. In this way, this classifier can be mapped to a point on the ROC plane, and these points can be mapped.
  • ROC can more intuitively reflect the performance of the classifier, and AUC can more intuitively represent the performance of the classifier, because AUC is a numerical value, and ROC is a two-dimensional coordinate curve, the full name of AUC is AreaUnderrocCurve, which refers to under ROCcurve The size of that part of the area. The larger the AUC value, the better the classification effect of the classifier.
  • a Bayesian logistic model is established for the parameters in the significant correlation parameter set.
  • the corresponding significant correlation parameter set is ⁇ average vehicle speed, average occupancy rate ⁇ , parameter "
  • the model of “vehicle average speed” is denoted as M1
  • the model of parameter “average occupancy rate” is denoted as M2.
  • Both M1 and M2 are used to predict whether a collision will occur.
  • the predicted value of 0 means no collision, and 1 means collision.
  • the ROC curve of model M2 is drawn in the coordinates, then there will be two ROC curves on the coordinates, but the shape of the ROC curve is not easy to quantify and compare, so it is necessary to calculate the corresponding AUC value, which is ROC Obtain the ROC curve with the largest AUC value from the area under the curve, look up layer by layer according to the ROC curve, find the corresponding Bayesian logistic model according to the ROC curve, and find its model establishment according to the Bayesian logistic model At last, find the corresponding traffic scene related parameters according to the significant correlation parameters used in the time, and use them as the most critical parameters of the collision risk model.
  • regression analysis is performed on the most critical parameters and the number of accidents based on a logistic regression algorithm, and a regression equation with good correlation and its derivation are established.
  • the specific formula of the regression model of the Logistic model is:
  • p i represents the accident probability of the accident in the i-th observation sample
  • x ki represents the value of the variable k in the i-th observation sample
  • ⁇ k is the correlation coefficient of the variable k.
  • the Bayesian inference method based on Markov Chain Monte Carlo (MCMC) method is applied to test the posterior probability distribution.
  • the estimated value of the mean, standard deviation and quartile of each explanatory variable can be determined by the posterior distribution provided by the Bayesian method.
  • the posterior distribution of estimated parameters can be estimated using the following formula:
  • Y) is expressed as the joint posterior distribution of the parameter ⁇ on the condition of the data set Y.
  • f(Y, ⁇ ) represents the joint probability distribution of the data set Y and the parameter ⁇ .
  • ⁇ ) is the likelihood condition function of the parameter ⁇ .
  • the function ⁇ ( ⁇ ) is the prior distribution of the parameter ⁇ . The following is a non-informative prior distribution formula:
  • Y) can be the following formula:
  • the predictability of an accident risk model can be expressed as the conditional probability of an accident in the presence of accident precursors when it comes to the overall prediction accuracy. Based on Bayesian theory, the following equation can be introduced to calculate the conditional probability:
  • A)/P(A') (standardized predictable value) is based on the accident risk model. When the ratio is greater than 1, compared with the conventional accident frequency model, the predictability of the model is considered effective.
  • T exceeds the preset threshold t of the accident precursor, that is, T>t, the event is identified as a traffic conflict event, otherwise it is determined as a non-conflict event. Therefore, P(A'
  • y i 1) represents the sensitivity; P (T>t
  • y i 0) represents the false alarm rate.
  • the probability of standardization can be estimated as the ratio between the sensitivity and the false alarm rate. Therefore, the predictability of the accident risk model can be approximately expressed as:
  • A') is used to evaluate the accuracy and effectiveness of the collision risk model. Its reciprocal can be interpreted as the number of predictions that need to be made before an accident can be accurately predicted. When applied to real projects, it can be converted into a specific number of active safety-related interventions required to prevent accidents. Therefore, compared with traditional sensitivity and specificity, this indicator can better reflect the costs and benefits of real-time accident risk models in practical applications.
  • the degree of influence of different parameters such as the penetration rate of connected vehicles and the design factors of the expressway on the accident risk can be obtained.
  • the vehicle collision model can determine whether an accident occurs based on the precursor data, which provides a basis for judging traffic safety early warning. At the same time, it can also be combined with the highway traffic system under the intelligent network to provide real-time risk warning for the networked vehicle.
  • the most critical parameters of the model can be considered to have greater relevance to traffic safety, and can provide early traffic warnings to vehicles on road sections where traffic accidents may occur based on finding outliers.
  • the predictability of accident success can also be judged through the adjustment of different test times to find the number of active interventions with higher predictability and higher probability of successful prediction.
  • the risk prevention method based on the collision model in the embodiment of the application is described above, and the risk prevention device based on the collision model in the embodiment of the application is described below. Please refer to FIG. 2.
  • the risk prevention device based on the collision model in the embodiment of the application is described below.
  • One embodiment includes:
  • the traffic simulation module 201 is used to perform traffic simulation on a preset mixed flow and output detector data and trajectory data collected during the traffic simulation process;
  • the data processing module 202 is configured to perform data preprocessing on the detector data to obtain a set of related parameters of the traffic scene, and perform data preprocessing on the trajectory data to obtain the number of accidents;
  • the model creation module 203 is used to sequentially perform regression analysis on the traffic scene related parameters in the traffic scene related parameter set to obtain a significant correlation parameter set and determine the most critical parameters of the model to be established according to the significant correlation parameter set; State the most critical parameters and the number of accidents and use logistic regression algorithm to build a collision risk model;
  • the risk prevention module 204 is configured to calculate the percentage of successful predicted traffic accidents of the collision risk model, and execute active safety-related intervention measures based on the percentage of successful predicted traffic accidents.
  • the traffic simulation module 201 can also be specifically used for:
  • Simulating traffic simulation is performed on the mixed flow simulation initial model based on the simulation parameters.
  • the traffic simulation module 201 can also be specifically used for:
  • a road section detector is set based on the simulation parameters, and a data collection interval and data collection parameters are set for the road section detector.
  • model creation module 202 may also be specifically used for:
  • the parameters are the most critical parameters of the model to be established.
  • model creation module 203 can also be specifically used to:
  • the active safety-related intervention measures are executed.
  • the risk prevention module 204 may also be specifically used to:
  • the percentage of successful predicted traffic accidents of the collision risk model is calculated.
  • the risk prevention module 204 may also be specifically used to:
  • a standardized predictable value of the collision risk model is calculated based on the sensitivity and the false alarm rate.
  • the modular design allows the hardware of each part of the risk prevention device based on the collision model to focus on the realization of a certain function, which maximizes the performance of the hardware. At the same time, the modular design also reduces the number of modules of the device. The coupling between the two makes it easier to maintain.
  • FIGS 1 and 2 above describe in detail the collision model-based risk prevention device in this embodiment of the application from the perspective of modular functional entities.
  • the following describes the collision model-based risk prevention device in this embodiment of the application in detail from the perspective of hardware processing. describe.
  • FIG. 3 is a schematic structural diagram of a risk prevention device based on a collision model provided by an embodiment of the present application.
  • the risk prevention device 300 based on a collision model may have relatively large differences due to different configurations or performances, and may include one or more A processor (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, one or more storage media 330 for storing application programs 333 or data 332 (for example, one or one storage device with a large amount of storage).
  • the memory 320 and the storage medium 330 may be short-term storage or persistent storage.
  • the program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the risk prevention device 300 based on the collision model. Furthermore, the processor 310 may be configured to communicate with the storage medium 330 and execute a series of instruction operations in the storage medium 330 on the risk prevention device 300 based on the collision model.
  • the risk prevention device 300 based on the collision model may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and/or, one or more operating systems 331, For example, WindowsServe, MacOSX, Unix, Linux, FreeBSD and so on.
  • operating systems 331 For example, WindowsServe, MacOSX, Unix, Linux, FreeBSD and so on.
  • FIG. 3 does not constitute a limitation on the risk prevention equipment based on the collision model, and may include more or less components than shown in the figure, or a combination of certain components. Some components, or different component arrangements.
  • the present application also provides a risk prevention device based on a collision model.
  • the risk prevention device based on the collision model includes a memory and a processor.
  • the memory stores computer-readable instructions.
  • the computer-readable instructions are executed by the processor, the The device executes the steps of the collision model-based risk prevention method in the foregoing embodiments.
  • the computer-readable storage medium may be a non-volatile computer-readable storage medium, and the computer-readable storage medium may also be a volatile computer-readable storage medium.
  • the computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the computer executes the steps of the collision model-based risk prevention method.
  • the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium.
  • the technical solution of the present application essentially or the part that contributes to the existing technology or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , Including several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (read-only memory, ROM), random access memory (random access memory, RAM), magnetic disk or optical disk and other media that can store program codes.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Geometry (AREA)
  • Operations Research (AREA)
  • Educational Administration (AREA)
  • Development Economics (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Quality & Reliability (AREA)
  • General Engineering & Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Traffic Control Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种基于碰撞模型的风险预防方法、装置、设备及存储介质,所述方法包括:对混合流进行交通仿真,将仿真输出数据进行预处理,确定待建立模型的最关键参数,并基于预处理后的仿真输出数据和最关键参数构建碰撞风险模型,最后计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施(105)。该方法可以有效预防潜在的交通事故风险,减少交通事故的发生率。

Description

基于碰撞模型的风险预防方法、装置、设备及存储介质
本申请要求于2020年11月19日提交中国专利局、申请号为202011298663.8、发明名称为“基于碰撞模型的风险预防方法、装置、设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在申请中。
技术领域
本申请涉及人工智能领域,尤其涉及一种基于碰撞模型的风险预防方法、装置、设备及存储介质。
背景技术
随着智能交通系统的发展,网联车辆将逐步走进我们的生活,在智能网联交通系统中,前车可以通过V2V技术,将加速度等行驶信息实时传递给后车,最终实现以协调自适应巡航控制的方式行驶。实际的实验说明,网联车渗透率的增加可以提升交通安全,但是由于不同车辆驾驶方式在驾驶习惯以及驾驶技巧中都存在差异,也可能在其中存在新的安全风险。
目前智能网联车的渗透率较低并且道路实现智能网联的条件有限,大多关于智能网联车辆运动特性的研究都依靠交通仿真软件的帮助,然而其中与网联车相关的宏观和微观模型较多,对于交通安全方面的研究较少。
发明人发现,在现有技术中,传统的混合交通流方面的事故风险模型由于网联车辆和人工驾驶车辆在微观运动特性和运动学性能方面的差异,此类模型无法直接应用于智能网联情况下的混合交通流中,所以目前对于智能网联情况下的混合交通流来说,并没有相应的事故风险模型来预防交通事故风险,减少交通事故的发生率。
发明内容
本申请的主要目的在于解决智能网联情况下的混合流没有相应的事故风险模型来预防交通事故风险,减少交通事故的发生率的问题。
本申请第一方面提供了一种基于碰撞模型的风险预防方法,包括:
对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;
依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分 比执行主动安全相关干预措施。
本申请第二方面提供了一种基于碰撞模型的风险预防装置,包括:
交通仿真模块,用于对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
数据处理模块,用于对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
模型创建模块,用于依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
风险预防模块,用于计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
本申请第三方面提供了一种基于碰撞模型的风险预防设备,包括:存储器和至少一个处理器,所述存储器中存储有指令;所述至少一个处理器调用所述存储器中的所述指令,以使得所述基于碰撞模型的风险预防设备执行如下所述的基于碰撞模型的风险预防方法的步骤:
对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;
依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
本申请的第四方面提供了一种计算机可读存储介质,所述计算机可读存储介质中存储有指令,当其在计算机上运行时,使得计算机执行如下所述的基于碰撞模型的风险预防方法的步骤:
对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;
依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分 比执行主动安全相关干预措施。
本申请提供的技术方案中,通过对混合流进行交通仿真,将仿真输出数据进行预处理,并基于预处理后的仿真输出数据构建碰撞风险模型,最后计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施,从而有效预防潜在的交通事故风险,减少交通事故的发生率。
附图说明
图1为本申请实施例中基于碰撞模型的风险预防方法的一个实施例示意图;
图2为本申请实施例中基于碰撞模型的风险预防装置的一个实施例示意图;
图3为本申请实施例中基于碰撞模型的风险预防设备的一个实施例示意图。
具体实施方式
本申请实施例提供了一种基于碰撞模型的风险预防方法、装置、设备及存储介质,可以有效预防潜在的交通事故风险,减少交通事故的发生率。
本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”、“第三”、“第四”等(如果存在)是用于区别类似的对象,而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换,以便这里描述的实施例能够以除了在这里图示或描述的内容以外的顺序实施。此外,术语“包括”或“具有”及其任何变形,意图在于覆盖不排他的包含,例如,包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元,而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
为便于理解,下面对本申请实施例的具体流程进行描述,请参阅图1,本申请实施例中基于碰撞模型的风险预防方法的一个实施例包括:
101、对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
本实施例中采用了VISSIM仿真软件进行交通仿真,VISSIM是一种微观的、基于时间间隔和驾驶行为的仿真建模工具,用以城市交通和公共交通运行的交通建模。它可以分析各种交通条件下,如车道设置、交通构成、交通信号、公交站点等,城市交通和公共交通的运行状况,综合考虑各种影响.路网运行的因素,例如车道类型,交通组成,信号控制配置,停车让行,同时还.可以加入行人模型,形成车辆和行人的交互,在各种影响因素设置好的情况下,通过周期的完整运行,可以导出各种数据。
VISSIM仿真软件内部由交通仿真器和信号状态产生器两部分组成,它们之间通过接口交换检测器数据和信号状态信息,交通仿真器是一个微观交通仿真模型,它包括跟车模型和车道变换模型。信号状态产生器是一个信号控制软件,可以通过程序实现交通流的控制逻辑。逻辑在每一个离散的时间间隔(可以是1-0.1秒)内从交通仿真器中提取检测器数 据,用以确定下一仿真秒的信号状态。同时,将信号状态信息回传给交通仿真器。VISSIM既可以在线生成可视化的交通运行状况,也可以离线输出各种统计数据,例如行程时间、排队长度等。而VISSIM仿真软件的核心模型,也就是生理-心理驾驶模型,让VISSIM能够更真实地模拟出现实的情况,该模型的基本思路是:一旦后车驾驶员认为他与前车之间的距离小于其心理(安全)距离时,后车驾驶员开始减速。由于后车驾驶员无法准确判断前车车速,后车车速会在一段时间内低于前车车速,直到前后车间的距离达到另一个心理(安全)距离时,后车驾驶员开始缓慢地加速,由此周而复始,形成一个加速、减速的迭代过程。
而本实施例中所述的混合流是指混合交通流,是指非机动车与机动车、性能相差悬殊的机动车混行的现象,机动车与非机动车混合交通流与纯机动车交通流并没有本质上的差异,都是源于交通个体出行需求而在公共道路上的行进行为,也都是时间和空间都离散的动力系统。但是混合交通流中个体间的相互影响非常复杂,其复杂程度远超过单一类型车流的复杂程度,这主要是因为不同类型的车辆在几何尺寸、行驶速度、动力特性、安全需求等方面差异较大,这就会导致不同类型的车辆的行驶规则是不同的。因此,混合交通流的构成和行为上的复杂性,决定其在运行特性上与纯机动车流存在较大的差异。
在本实施例中,首先在VISSIM仿真软件中的参数配置入口设定好相关的交通仿真参数,其中交通仿真参数直接影响仿真结果,例如对于模拟混合交通流的场景,在VISSIM仿真软件中绘制仿真路段、设定车辆类型及其期望速度、分配行驶路径、设定并配置检测器等。在混合交通流的模拟过程中,VISSIM仿真软件会收集两类数据,一类为检测器收集得到的数据,另一类为轨迹数据文件。在VISSIM仿真软件完成对混合流的交通仿真时,将对应的检测器收集到的数据和轨迹数据以文件的形式导出到本地存储中。
在本实施例中,上述101还包括以下步骤:
在预置交通仿真软件中新建混合流仿真初始模型;
为所述混合流仿真初始模型设定仿真参数;
基于所述仿真参数对所述混合流仿真初始模型进行模拟交通仿真。
本实施例中,通过在仿真软件中选取待仿真路段,绘制所述仿真路段,设置不同智能网联车辆的交通占比、交通流量、期望速度,并为所述交通流量分配路径和交通优先级。
一个交通系统的基本组成就是路网,在VISSIM中对于仿真路段的绘制是通过路网编辑器来实现的,而VISSIM的路网编辑器的使用是比较简单简便的,它没有固定的模板,用户可以根据自己的需求画出各种路网,有时候由于现实情况的需要,车道宽度或者转弯半径不是理想的标准,那么可以根据路网的实际图进行调整,此时,我们可以将混合流的实际路网图导入VISSIM中,作为底图,底图可以导入多种格式,然后在图上进行路网编辑,这样就可以保证做出来的仿真与实际路网一致。
智能网联车辆的输入通过VISSIM中的外部驾驶员模型接口完成,在车辆定义模块下的车辆类型属性中建立新的车辆模型并在外部驾驶员模型选项中导入编译好的ddl文件。可 以根据实际情况在ddl文件中定义各种类型的车,例如常见的出租车,轿车,小型的SUV都是属于小汽车的范畴,那么需要在对应的ddl文件中定义汽车类别为小汽车,并在子分支下定义各种分支类型,例如它的长度,宽度,高度等等。
在车速定义模块下,对各种类型的车辆在不同的路段下定义期望速度,其中速度是以区间的形式表示,可以对此区间进行调整。例如在车速定义模块下将小汽车类型在城市路段的期望速度定义为[40,60],学校区域路段的期望速度定义为[10,30],单位为km/h。
在交通量定义模块下,设定交通流量、不同车型的交通占比,为路段分配交通量。交通流量用数值形式进行表示,例如设定交通流量为1000,代表整个交通仿真系统中一共有1000辆汽车;交通占比用百分比的形式进行表示,例如设定不同车型的交通占比例如小汽车占比95%、公交车占比2%、货运卡车占比2%,其他类型占比1%。路段分配交通量,将仿真路段被划分为多条路径并为其添加标识,例如路径a、路径b、路径c,然后在交通量定义模块下的路段选择选项中选择指定的路径,最后在交通量属性中配置该条路段的交通量,例如选择路径a,配置该条路径的交通量为10%,而整个仿真中的交通流量为1000,则路径a中的交通量即为100。
在交通优先级定义模块下,设定交通优先级,交通优先级是指驾驶员在驾驶的过程中发现前车的速度会妨碍自己行驶,同时前方路况较好的情况下,驾驶员会选择超车,那么就要进行车道变换,在变换车道的时候需要注意本车道和邻车道的情况,在此模块中可以进行设置,用户可以自定义变换方式,变换位置,也可以设置某个地段不允许车道变换。
在本实施例中,在为所述混合流仿真初始模型设定仿真参数之后还包括:
基于所述仿真参数设置路段检测器,为所述路段检测器设定数据收集间隔时间和数据收集参数;
在本实施例中,在需要检测的路段上每隔1Km设置路段检测器,设定收集的时间间隔为30s,设定路段检测器的数据收集参数,包括车辆的平均速度、平均占有率以及平均通过车辆数。
102、对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
检测器数据是上述在待检测路段上设置的路段检测器收集的数据,主要包括车辆的平均速度、平均占有率和平均通过车辆数,以线圈文件的形式存储于VISSIM的线圈文件路径中,对线圈文件中的数据进行整理后得到交通场景相关参数集合。
轨迹数据通过使用SSAM软件设定不同的TTC(Timetocollison,距离碰撞时间)阈值来得到产生的追尾事故数量。SSAM软件是一款仿真冲突分析软件,该软件能够对VISSIM、PARAMICS,TEXAS和AIMSUN四种仿真模型输出的车辆轨迹文件进行仿真冲突分析,并能输出不同类型仿真冲突数据,包括仿真冲突数量、位置以及严重程度,本实施例中的追尾事故数量即是仿真冲突数量。
在本实施例中,上述102还包括以下步骤:
将两个连续上下游的数据整合为预置时间间隔T的交通场景相关参数。
在本实施例中,对于检测器收集得到的数据,将两个连续上下游的数据整合为时间间隔为5分钟的交通场景相关参数,两个连续上下游是指检测器执行的连续两次数据采集,采集时间间隔为上述路段检测器设定的收集时间间隔。事故发生前5分钟的数据将作为事故前兆数据,这样收集的前兆数据可用于交通事故的预测。例如某一路段检测器第1次采集数据和第2次采集数据整合为时间间隔为5分钟的交通场景相关参数,得到第1间隔和第2间隔,如果在第2间隔的5分钟时间内发生了交通事故,那么第1间隔对应的交通场景相关数据,即第1次采集数据将作为事故前兆数据。
103、依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合并根据所述显著相关性参数集合确定待建立模型的最关键参数;
考虑到交通场景相关参数集合中参数的工程实际运用的有效性,以及参数之间可能存在相关性,从交通场景相关参数集合中选取少量且有效的参数进行建立风险模型。首先对交通场景相关参数集合中的所有参数进行逻辑回归分析,筛选出与事故发生有显著相关性的参数,得到显著相关性参数集合,然后再对每个单独的显著变量建立贝叶斯logistic模型,绘制相应的接受者操作特性曲线(receiveroperatingcharacteristiccurve,简称ROC曲线),最后计算每一条ROC曲线的AUC值(AreaUnderCurve),最终将最大AUC值对应的交通场景相关参数作为所在交通场景的最关键参数。
在本实施例中,上述103还包括以下步骤:
依次为所述显著相关性参数集合中的参数建立逻辑回归模型,得到逻辑回归模型集合;
依次对所述逻辑回归模型集合中的模型绘制接受者操作特性曲线,得到接受者操作特性曲线集合;
计算所述接受者操作特性曲线集合中的AUC值,得到AUC值集合;
获取所述AUC值集合中数值最大的AUC值,根据所述数值最大的AUC值查找对应的显著相关性参数,根据所述显著相关性参数查找对应的交通场景相关参数,将所述交通场景相关参数作为待建立模型的最关键参数。
其中,接受者操作曲线,简称ROC曲线,常用于人工智能场景下对神经网络模型中的分类器性能进行评估,其主要分析工具是一个画在二维平面上的曲线——ROCcurve。平面的横坐标是falsepositiverate(FPR),纵坐标是truepositiverate(TPR),TPR为在所有实际为阳性的样本中,被正确地判断为阳性之比,而FPR为在所有实际为阴性的样本中,被错误地判断为阳性之比率。对某个分类器而言,我们可以根据其在测试样本上的表现得到一个TPR和FPR点对。这样,此分类器就可以映射成ROC平面上的一个点,将这些点。通过ROC能够比较直观地反应分类器的性能,而AUC能够更加直观地表示分类器的性能,因为AUC是一个数值,而ROC是一张二维坐标曲线,AUC的全称为AreaUnderrocCurve,指的是处于ROCcurve下方的那部分面积的大小。而AUC值越大,则代表分类器的分类效果越好。
在本实施例中,为显著相关性参数集合中的参数建立贝叶斯logistic模型,例如发生交通碰撞场景下,相应的显著相关性参数集合为{车辆的平均速度,平均占有率},参数“车辆平均速度”的模型记为M1,参数“平均占有率”的模型记为M2,M1和M2均用于预测是否发生碰撞,用一组数据让M1进行预测,得到一组样本案例,请参考表一,再用一组数据让M2进行预测,也得到一组样本案例,请参考表二。
表一
序号 1 2 3 4 5 6
真实值 碰撞 没碰撞 没碰撞 碰撞 没碰撞 没碰撞
预测值 0.72 0.44 0.35 0.48 0.54 0.49
表二
序号 1 2 3 4 5 6
真实值 碰撞 碰撞 没碰撞 没碰撞 碰撞 没碰撞
预测值 0.68 0.44 0.45 0.39 0.38 0.69
其中预测值0代表没碰撞,1代表碰撞。数字越大越接近碰撞特征,数字越小越接近没碰撞特征。如果我们设定区分是否碰撞的阈值是0.5,那么预测值大于0.5的都是P正向碰撞,小于0.5都是N负向没碰撞。
那么对于模型M1来说,真实2个碰撞样本中只有1号查出来了,模型M1的查出率TPR=1/2=0.5;真实4个没碰撞样本中5号被查错,所以误检率FPR=1/4=0.25;精度是ACC=(1+3)/6≈0.666。但是注意,如果我们修改阈值等于0.4,那么就会变为2个碰撞样本全被检出,查出率TPR=1;而没碰撞样本则被误检3个,误检率FPR=0.75;精度为0.5。
上面我们都只是把从一组预测样本得到的[FPR,TPR]作为一个点描述,并且我们知道阈值的改变会严重影响FPR和TPR,那么,如果我们把所有可能的阈值都尝试一遍,再把样本集预测结果计算得到的所有[FPR,TPR]点都画在坐标上,就会得到一个曲线,也就是接受者操作特征曲线(ROC曲线)。
按照同样的原理将模型M2的ROC曲线在坐标中划出,那么在坐标上就会有2条ROC曲线,但是ROC曲线的形状不太好量化比较,因此需要计算对应的AUC值,也就是ROC曲线下面的面积,得到AUC值最大的ROC曲线,根据所述ROC曲线往上一层层地查找,根据ROC曲线找到对应的贝叶斯logistic模型,根据所述贝叶斯logistic模型找到其模型建立时所用的显著相关性参数,最后根据所述显著相关性参数找到对应的交通场景相关参数,将其作为碰撞风险模型的最关键参数。
104、依据所述最关键参数和所述事故产生数量并利用逻辑回归算法构建碰撞风险模型;
在本实施例中,将所述最关键参数和事故产生数量基于逻辑回归算法进行回归分析,建立了相关性较好的回归方程及其推导,其中Logistic模型的回归模型具体公式为:
y i~Bernoulli(p i)
logit(p i)=β 01x 1i2x 2i+···+β kix ki
y i表示观察的第i个样本事故是否发生,当y i=1时表示事故发生,当y i=0是表示事故并未发生。
p i表示事故在第i个观察样本中的事故可能性,x ki表示了在第i个观察样本下变量k的值,β k是变量k的相关系数。其似然函数参考公式为:
Figure PCTCN2021082677-appb-000001
采用基于马尔可夫链蒙特卡洛(MCMC)方法的贝叶斯推理方法被应用于检验的后验概率分布。每个解释变量的均值、标准差和四分位数的估计值都可以通过贝叶斯方法所提供的后验分布来确定。根据贝叶斯定理,估计参数的后验分布可以使用以下公式来估计:
Figure PCTCN2021082677-appb-000002
其中,f(β|Y)表示为以数据集Y为条件的关于参数β的联合后验分布。f(Y,β)表示数据集Y和参数β的联合概率分布。f(Y|β)是参数β的似然条件函数。函数π(β)是参数β的先验分布。下列的为非信息性先验分布公式:
β~N(0 k,10 6I k)
其中0 k是k×1的零向量,I k为k×k的单位矩阵,根据参数β的先验分布的规范,联合先验分布f(β|Y)可以为以下公式:
f(β|Y)∝f(Y|β)π(β)
Figure PCTCN2021082677-appb-000003
105、计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
事故风险模型的可预测性可以表示为当涉及总体的预测准确性时,在存在事故前兆的条件下发生事故的条件概率。基于贝叶斯理论,可以引入以下方程式来计算条件概率:
Figure PCTCN2021082677-appb-000004
其中A是发生事故的情况;A'是事故前兆;P(A)是实际发生事故的概率,或可以通过报告的事故数据获得的先验事故概率;P(A')是注意到事故前兆的概率;P(A|A')表示在有事故前兆的情况下发生事故的概率;P(A'|A)表示发生事故之前注意到前兆的可能性。
值得注意的是,P(A'|A)/P(A')(标准化可预测值),是基于事故风险模型的。当比值大于1时,与常规事故频率模型相比,该模型的可预测性被认为是有效的。当估计的事故概率T超过事故前兆的预设阈值t,即T>t时,该事件被识别为交通冲突事件,否则被确定为非冲突事件。因此,P(A'|A)/P(A')可以表示为以下公式:
Figure PCTCN2021082677-appb-000005
其中P(y i=1)和P(y i=0)分别表示在5分钟的时间间隔内集计的事故和非事故的比例; P(T>t|y i=1)表示灵敏度;P(T>t|y i=0)表示误报率。
实际上,事故的比例明显小于非事故的比例,即P(y i=1)≈0和P(y i=0)≈1。因此,P(T>t|y i=1)×P(y i=1)与P(T>t|y i=0)×P(y i=0)的值相比可忽略不计。同样,P(T>t|y i=0)×P(y i=0)可以认为等于P(T>t|y i=0)。因此式前式可以简化为:
Figure PCTCN2021082677-appb-000006
如上所述P(T>t|y i=1)是灵敏度,P(T>t|y i=0)是误警报率,也可以认为是(1-特异性)。可以解释为:
Figure PCTCN2021082677-appb-000007
通过将公式P(A'|A)/P(A')和公式P(A|A')组合在一起,可以将标准化的可能性估计为灵敏度和误报率之间的比率。因此,事故风险模型的可预测性可以近似表示为:
Figure PCTCN2021082677-appb-000008
指标P(A|A')用于评估碰撞风险模型的准确性和有效性。它的倒数可以解释为在准确预测事故之前需要进行的预测次数。当应用于实际项目时,可以将其转换为预防事故所需的特定数量的主动安全相关干预措施。因此,与传统的敏感性和特异性相比,该指标可以更好地反映实际应用中实时事故风险模型的成本和收益。
完成模型的建立之后,计算P(y i=1|T>t) -1的值达到1000、2000、3000、5000、6000、7000、8000、9000和10000次时,模型可成功预测交通事故的百分比柱状图。根据得到的成功预测交通事故百分比,选择预测事故需要进行的预测次数,作为工程上可运用的指标。
本申请实施例中,通过碰撞风险模型的建立,可以得出不同参数例如网联车渗透率、高速公路设计因素等对事故风险的影响程度。车辆碰撞模型可根据前兆数据来判断是否发生事故,为交通安全预警提供了判断依据。同时也可以与智能网联下的高速公路交通系统相结合,为网联车提供实时的风险预警。该模型的最关键参数,可认为与交通安全有较大的相关性,可根据查找其异常值来较快地对可以发生交通事故的路段上的车辆进行交通预警。事故成功的可预测性判断也可通过不同测试次数的调试,找到可预测性较高并且成功预测概率较大的主动干预次数。
上面对本申请实施例中基于碰撞模型的风险预防方法进行了描述,下面对本申请实施例中基于碰撞模型的风险预防装置进行描述,请参阅图2,本申请实施例中基于碰撞模型的风险预防装置一个实施例包括:
交通仿真模块201,用于用于对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
数据处理模块202,用于对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
模型创建模块203,用于依次对交通场景相关参数集合中的交通场景相关参数进行回 归分析,得到显著相关性参数集合并根据所述显著相关性参数集合确定待建立模型的最关键参数;依据所述最关键参数和所述事故产生数量并利用逻辑回归算法构建碰撞风险模型;
风险预防模块204,用于用于计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
可选的,交通仿真模块201还可以具体用于:
在预置交通仿真软件中新建混合流仿真初始模型;
为所述混合流仿真初始模型设定仿真参数;
基于所述仿真参数对所述混合流仿真初始模型进行模拟交通仿真。
可选的,交通仿真模块201还可以具体用于:
基于所述仿真参数设置路段检测器,为所述路段检测器设定数据收集间隔时间和数据收集参数。
可选的,模型创建模块202还可以具体用于:
依次为所述显著相关性参数集合中的参数建立逻辑回归模型,得到逻辑回归模型集合;
依次对所述逻辑回归模型集合中的模型绘制接受者操作特性曲线,得到接受者操作特性曲线集合;
计算所述接受者操作特性曲线集合中的AUC值,得到AUC值集合;
获取所述AUC值集合中数值最大的AUC值,根据所述数值最大的AUC值查找对应的显著相关性参数,根据所述显著相关性参数查找对应的交通场景相关参数,将所述交通场景相关参数作为待建立模型的最关键参数。
可选的,模型创建模块203还可以具体用于:
基于所述成功预测交通事故百分比,确定待预测次数;
根据所述待预测次数,确定待执行的主动安全相关干预措施的数量;
依据所述待执行的主动安全相关干预措施的数量,执行主动安全相关干预措施。
可选的,风险预防模块204还可以具体用于:
计算所述碰撞风险模型的标准化可预测值和实际发生事故的概率;
基于所述标准化可预测值和实际发生事故的概率,计算所述碰撞风险模型的成功预测交通事故百分比。
可选的,风险预防模块204还可以具体用于:
计算所述预置时间间隔T内的交通冲突事故的比例,得到灵敏度;
计算所述预置时间间隔T内的非交通冲突事故的比例,得到误报率;
基于所述灵敏度和所述误报率计算所述碰撞风险模型的标准化可预测值。
本申请实施例中,模块化的设计让基于碰撞模型的风险预防装置各部位的硬件专注于某一功能的实现,最大化实现了硬件的性能,同时模块化的设计也降低了装置的模块之间 的耦合性,更加方便维护。
上面图1和图2从模块化功能实体的角度对本申请实施例中的基于碰撞模型的风险预防装置进行详细描述,下面从硬件处理的角度对本申请实施例中基于碰撞模型的风险预防设备进行详细描述。
图3是本申请实施例提供的一种基于碰撞模型的风险预防设备的结构示意图,该基于碰撞模型的风险预防设备300可因配置或性能不同而产生比较大的差异,可以包括一个或一个以上处理器(centralprocessingunits,CPU)310(例如,一个或一个以上处理器)和存储器320,一个或一个以上存储应用程序333或数据332的存储介质330(例如一个或一个以上海量存储设备)。其中,存储器320和存储介质330可以是短暂存储或持久存储。存储在存储介质330的程序可以包括一个或一个以上模块(图示没标出),每个模块可以包括对基于碰撞模型的风险预防设备300中的一系列指令操作。更进一步地,处理器310可以设置为与存储介质330通信,在基于碰撞模型的风险预防设备300上执行存储介质330中的一系列指令操作。
基于碰撞模型的风险预防设备300还可以包括一个或一个以上电源340,一个或一个以上有线或无线网络接口350,一个或一个以上输入输出接口360,和/或,一个或一个以上操作系统331,例如WindowsServe,MacOSX,Unix,Linux,FreeBSD等等。本领域技术人员可以理解,图3示出的基于碰撞模型的风险预防设备结构并不构成对基于碰撞模型的风险预防设备的限定,可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件布置。
本申请还提供一种基于碰撞模型的风险预防设备,所述基于碰撞模型的风险预防设备包括存储器和处理器,存储器中存储有计算机可读指令,计算机可读指令被处理器执行时,使得处理器执行上述各实施例中的所述基于碰撞模型的风险预防方法的步骤。
本申请还提供一种计算机可读存储介质,该计算机可读存储介质可以为非易失性计算机可读存储介质,该计算机可读存储介质也可以为易失性计算机可读存储介质,所述计算机可读存储介质中存储有指令,当所述指令在计算机上运行时,使得计算机执行所述基于碰撞模型的风险预防方法的步骤。
所属领域的技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的系统,装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
所述集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(read-onlymemory, ROM)、随机存取存储器(randomaccessmemory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述,以上实施例仅用以说明本申请的技术方案,而非对其限制;尽管参照前述实施例对本申请进行了详细的说明,本领域的普通技术人员应当理解:其依然可以对前述各实施例所记载的技术方案进行修改,或者对其中部分技术特征进行等同替换;而这些修改或者替换,并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。

Claims (20)

  1. 一种基于碰撞模型的风险预防方法,其中,所述风险预防方法包括:
    对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
    对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
    依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;
    依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
    计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
  2. 根据权利要求1所述的基于碰撞模型的风险预防方法,其中,所述对预置混合流进行交通仿真包括:
    在预置交通仿真软件中新建混合流仿真初始模型;
    为所述混合流仿真初始模型设定仿真参数;
    基于所述仿真参数对所述混合流仿真初始模型进行模拟交通仿真。
  3. 根据权利要求2所述的基于碰撞模型的风险预防方法,其中,在所述为所述混合流仿真初始模型设定仿真参数之后,还包括:
    基于所述仿真参数设置路段检测器,为所述路段检测器设定数据收集间隔时间和数据收集参数。
  4. 根据权利要求1所述的基于碰撞模型的风险预防方法,其中,所述根据所述显著相关性参数集合确定待建立模型的最关键参数包括:
    依次为所述显著相关性参数集合中的参数建立逻辑回归模型,得到逻辑回归模型集合;
    依次对所述逻辑回归模型集合中的模型绘制接受者操作特性曲线,得到接受者操作特性曲线集合;
    计算所述接受者操作特性曲线集合中的AUC值,得到AUC值集合;
    获取所述AUC值集合中数值最大的AUC值,根据所述数值最大的AUC值查找对应的显著相关性参数,根据所述显著相关性参数查找对应的交通场景相关参数,将所述交通场景相关参数作为待建立模型的最关键参数。
  5. 根据权利要求1所述的基于碰撞模型的风险预防方法,其中,所述基于所述成功预测交通事故百分比执行主动安全相关措施包括:
    基于所述成功预测交通事故百分比,确定待预测次数;
    根据所述待预测次数,确定待执行的主动安全相关干预措施的数量;
    依据所述待执行的主动安全相关干预措施的数量,执行主动安全相关干预措施。
  6. 根据权利要求1所述的基于碰撞模型的风险预防方法,其中,所述计算所述碰撞风险模型的成功预测交通事故百分比包括:
    计算所述碰撞风险模型的标准化可预测值和实际发生事故的概率;
    基于所述标准化可预测值和实际发生事故的概率,计算所述碰撞风险模型的成功预测交通事故百分比。
  7. 根据权利要求6所述的基于碰撞模型的风险预防方法,其中,所述计算所述碰撞风险模型的标准化可预测值包括:
    计算所述预置时间间隔T内的交通冲突事故的比例,得到灵敏度;
    计算所述预置时间间隔T内的非交通冲突事故的比例,得到误报率;
    基于所述灵敏度和所述误报率计算所述碰撞风险模型的标准化可预测值。
  8. 一种基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备包括:存储器和至少一个处理器,所述存储器中存储有指令;
    所述至少一个处理器调用所述存储器中的所述指令,以使得所述基于碰撞模型的风险预防设备执行如下所述的基于碰撞模型的风险预防方法的步骤:
    对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
    对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
    依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;
    依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
    计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
  9. 根据权利要求8所述的基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备执行所述对预置混合流进行交通仿真的步骤时,包括:
    在预置交通仿真软件中新建混合流仿真初始模型;
    为所述混合流仿真初始模型设定仿真参数;
    基于所述仿真参数对所述混合流仿真初始模型进行模拟交通仿真。
  10. 根据权利要求9所述的基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备执行所述为所述混合流仿真初始模型设定仿真参数的步骤之后,还包括如下步骤:
    基于所述仿真参数设置路段检测器,为所述路段检测器设定数据收集间隔时间和数据收集参数。
  11. 根据权利要求8所述的基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备执行所述根据所述显著相关性参数集合确定待建立模型的最关键参数的 步骤时,包括:
    依次为所述显著相关性参数集合中的参数建立逻辑回归模型,得到逻辑回归模型集合;
    依次对所述逻辑回归模型集合中的模型绘制接受者操作特性曲线,得到接受者操作特性曲线集合;
    计算所述接受者操作特性曲线集合中的AUC值,得到AUC值集合;
    获取所述AUC值集合中数值最大的AUC值,根据所述数值最大的AUC值查找对应的显著相关性参数,根据所述显著相关性参数查找对应的交通场景相关参数,将所述交通场景相关参数作为待建立模型的最关键参数。
  12. 根据权利要求8所述的基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备执行所述基于所述成功预测交通事故百分比执行主动安全相关措施的步骤时,包括:
    基于所述成功预测交通事故百分比,确定待预测次数;
    根据所述待预测次数,确定待执行的主动安全相关干预措施的数量;
    依据所述待执行的主动安全相关干预措施的数量,执行主动安全相关干预措施。
  13. 根据权利要求8所述的基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备执行所述计算所述碰撞风险模型的成功预测交通事故百分比的步骤时,包括:
    计算所述碰撞风险模型的标准化可预测值和实际发生事故的概率;
    基于所述标准化可预测值和实际发生事故的概率,计算所述碰撞风险模型的成功预测交通事故百分比。
  14. 根据权利要求13所述的基于碰撞模型的风险预防设备,其中,所述基于碰撞模型的风险预防设备执行所述计算所述碰撞风险模型的标准化可预测值的步骤时,包括:
    计算所述预置时间间隔T内的交通冲突事故的比例,得到灵敏度;
    计算所述预置时间间隔T内的非交通冲突事故的比例,得到误报率;
    基于所述灵敏度和所述误报率计算所述碰撞风险模型的标准化可预测值。
  15. 一种计算机可读存储介质,所述计算机可读存储介质上存储有指令,其中,所述指令被处理器执行时实现如下所述的基于碰撞模型的风险预防方法的步骤:
    对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
    对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
    依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合,并根据所述显著相关性参数集合确定待建立模型的最关键参数;
    依据所述最关键参数和所述事故产生数量,并利用逻辑回归算法构建碰撞风险模型;
    计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
  16. 根据权利要求15所述的计算机可读存储介质,其中,所述指令被处理器执行时实现所述对预置混合流进行交通仿真的步骤时,包括:
    在预置交通仿真软件中新建混合流仿真初始模型;
    为所述混合流仿真初始模型设定仿真参数;
    基于所述仿真参数对所述混合流仿真初始模型进行模拟交通仿真。
  17. 根据权利要求15所述的计算机可读存储介质,其中,所述指令被处理器执行时实现所述根据所述显著相关性参数集合确定待建立模型的最关键参数的步骤时,包括:
    依次为所述显著相关性参数集合中的参数建立逻辑回归模型,得到逻辑回归模型集合;
    依次对所述逻辑回归模型集合中的模型绘制接受者操作特性曲线,得到接受者操作特性曲线集合;
    计算所述接受者操作特性曲线集合中的AUC值,得到AUC值集合;
    获取所述AUC值集合中数值最大的AUC值,根据所述数值最大的AUC值查找对应的显著相关性参数,根据所述显著相关性参数查找对应的交通场景相关参数,将所述交通场景相关参数作为待建立模型的最关键参数。
  18. 根据权利要求15所述的计算机可读存储介质,其中,所述指令被处理器执行时实现所述基于所述成功预测交通事故百分比执行主动安全相关措施的步骤时,包括:
    基于所述成功预测交通事故百分比,确定待预测次数;
    根据所述待预测次数,确定待执行的主动安全相关干预措施的数量;
    依据所述待执行的主动安全相关干预措施的数量,执行主动安全相关干预措施。
  19. 根据权利要求15所述的计算机可读存储介质,其中,所述指令被处理器执行时实现所述计算所述碰撞风险模型的成功预测交通事故百分比的步骤时,包括:
    计算所述碰撞风险模型的标准化可预测值和实际发生事故的概率;
    基于所述标准化可预测值和实际发生事故的概率,计算所述碰撞风险模型的成功预测交通事故百分比。
  20. 一种基于碰撞模型的风险预防装置,其中,所述基于碰撞模型的风险预防装置包括:
    交通仿真模块,用于对预置混合流进行交通仿真,并输出所述交通仿真过程中采集的检测器数据和轨迹数据;
    数据处理模块,用于对所述检测器数据进行数据预处理,得到交通场景相关参数集合,对所述轨迹数据进行数据预处理,得到事故产生数量;
    模型创建模块,用于依次对交通场景相关参数集合中的交通场景相关参数进行回归分析,得到显著相关性参数集合并根据所述显著相关性参数集合确定待建立模型的最关键参 数;依据所述最关键参数和所述事故产生数量并利用逻辑回归算法构建碰撞风险模型;
    风险预防模块,用于计算所述碰撞风险模型的成功预测交通事故百分比,基于所述成功预测交通事故百分比执行主动安全相关干预措施。
PCT/CN2021/082677 2020-11-19 2021-03-24 基于碰撞模型的风险预防方法、装置、设备及存储介质 WO2021203972A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011298663.8A CN112487617B (zh) 2020-11-19 2020-11-19 基于碰撞模型的风险预防方法、装置、设备及存储介质
CN202011298663.8 2020-11-19

Publications (1)

Publication Number Publication Date
WO2021203972A1 true WO2021203972A1 (zh) 2021-10-14

Family

ID=74931772

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/082677 WO2021203972A1 (zh) 2020-11-19 2021-03-24 基于碰撞模型的风险预防方法、装置、设备及存储介质

Country Status (2)

Country Link
CN (1) CN112487617B (zh)
WO (1) WO2021203972A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114241754A (zh) * 2021-12-03 2022-03-25 东南大学 基于高速公路合流影响区事故前兆特征的实时控制方法
CN114944055A (zh) * 2022-03-29 2022-08-26 浙江省交通投资集团有限公司智慧交通研究分公司 基于电子收费门架的高速公路碰撞风险动态预测方法
CN115100867A (zh) * 2022-07-27 2022-09-23 武汉微晶石科技股份有限公司 一种基于数字孪生的城市智能交通仿真方法
CN115206088A (zh) * 2022-06-06 2022-10-18 合肥工业大学 考虑动态交通流的城市快速路交通事故风险量化评估方法
CN115762140A (zh) * 2022-11-01 2023-03-07 山东高速集团有限公司 一种考虑变量异质性的高速公路交通事故风险预测方法
CN115857336A (zh) * 2022-11-22 2023-03-28 中科航宇(北京)自动化工程技术有限公司 悬臂式斗轮机防碰撞的方法、装置和设备
CN115906539A (zh) * 2023-01-10 2023-04-04 联友智连科技有限公司 一种基于Django实现自动驾驶仿真的系统
CN116631221A (zh) * 2023-05-05 2023-08-22 同济大学 一种基于蒙特卡洛模拟的在途车辆运行风险量化计算方法
CN115762140B (zh) * 2022-11-01 2024-06-07 山东高速集团有限公司 一种考虑变量异质性的高速公路交通事故风险预测方法

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112487617B (zh) * 2020-11-19 2024-03-08 平安科技(深圳)有限公司 基于碰撞模型的风险预防方法、装置、设备及存储介质
CN113345226A (zh) * 2021-05-31 2021-09-03 重庆交通大学 一种突发事件影响在高速公路网中的传播预测方法
CN113536628A (zh) * 2021-07-09 2021-10-22 深圳佰维存储科技股份有限公司 焊点回损预测方法、装置、可读存储介质及电子设备
CN114566051A (zh) * 2022-03-18 2022-05-31 武汉理工大学 基于logistic模型的高速公路转换区严重冲突识别方法
CN115293255B (zh) * 2022-08-02 2023-05-05 河北雄安荣乌高速公路有限公司 高速公路交通事故风险模型构建及风险判别方法
CN115830800A (zh) * 2022-11-28 2023-03-21 广州城建职业学院 车祸预警方法、系统、装置及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521438A (zh) * 2011-12-02 2012-06-27 东南大学 交通冲突仿真两阶段参数标定方法
CN106529076A (zh) * 2016-11-28 2017-03-22 东南大学 高速公路交通安全仿真分析的两阶段参数标定方法
US20170316127A1 (en) * 2016-04-29 2017-11-02 Baidu Online Network Technology (Beijing) Co., Ltd. Method and apparatus for constructing testing scenario for driverless vehicle
CN109118773A (zh) * 2018-09-30 2019-01-01 中交第公路勘察设计研究院有限公司 一种高速公路交通事故风险评估方法
CN112487617A (zh) * 2020-11-19 2021-03-12 平安科技(深圳)有限公司 基于碰撞模型的风险预防方法、装置、设备及存储介质

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103150930B (zh) * 2013-02-01 2015-09-16 东南大学 针对快速道路常发性拥堵路段的追尾事故实时预测方法
CN104820763B (zh) * 2015-05-25 2017-12-15 西华大学 一种基于vissim的交通事故三维仿真方法
CN111626461A (zh) * 2019-09-18 2020-09-04 东莞灵虎智能科技有限公司 一种安全风险预测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521438A (zh) * 2011-12-02 2012-06-27 东南大学 交通冲突仿真两阶段参数标定方法
US20170316127A1 (en) * 2016-04-29 2017-11-02 Baidu Online Network Technology (Beijing) Co., Ltd. Method and apparatus for constructing testing scenario for driverless vehicle
CN106529076A (zh) * 2016-11-28 2017-03-22 东南大学 高速公路交通安全仿真分析的两阶段参数标定方法
CN109118773A (zh) * 2018-09-30 2019-01-01 中交第公路勘察设计研究院有限公司 一种高速公路交通事故风险评估方法
CN112487617A (zh) * 2020-11-19 2021-03-12 平安科技(深圳)有限公司 基于碰撞模型的风险预防方法、装置、设备及存储介质

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114241754B (zh) * 2021-12-03 2023-01-03 东南大学 基于高速公路合流影响区事故前兆特征的实时控制方法
CN114241754A (zh) * 2021-12-03 2022-03-25 东南大学 基于高速公路合流影响区事故前兆特征的实时控制方法
CN114944055A (zh) * 2022-03-29 2022-08-26 浙江省交通投资集团有限公司智慧交通研究分公司 基于电子收费门架的高速公路碰撞风险动态预测方法
CN114944055B (zh) * 2022-03-29 2023-04-18 浙江省交通投资集团有限公司智慧交通研究分公司 基于电子收费门架的高速公路碰撞风险动态预测方法
CN115206088A (zh) * 2022-06-06 2022-10-18 合肥工业大学 考虑动态交通流的城市快速路交通事故风险量化评估方法
CN115206088B (zh) * 2022-06-06 2023-05-02 合肥工业大学 考虑动态交通流的城市快速路交通事故风险量化评估方法
CN115100867B (zh) * 2022-07-27 2022-11-29 武汉微晶石科技股份有限公司 一种基于数字孪生的城市智能交通仿真方法
CN115100867A (zh) * 2022-07-27 2022-09-23 武汉微晶石科技股份有限公司 一种基于数字孪生的城市智能交通仿真方法
CN115762140A (zh) * 2022-11-01 2023-03-07 山东高速集团有限公司 一种考虑变量异质性的高速公路交通事故风险预测方法
CN115762140B (zh) * 2022-11-01 2024-06-07 山东高速集团有限公司 一种考虑变量异质性的高速公路交通事故风险预测方法
CN115857336A (zh) * 2022-11-22 2023-03-28 中科航宇(北京)自动化工程技术有限公司 悬臂式斗轮机防碰撞的方法、装置和设备
CN115857336B (zh) * 2022-11-22 2024-02-02 江苏射阳港发电有限责任公司 悬臂式斗轮机防碰撞的方法、装置和设备
CN115906539A (zh) * 2023-01-10 2023-04-04 联友智连科技有限公司 一种基于Django实现自动驾驶仿真的系统
CN116631221A (zh) * 2023-05-05 2023-08-22 同济大学 一种基于蒙特卡洛模拟的在途车辆运行风险量化计算方法
CN116631221B (zh) * 2023-05-05 2024-01-23 同济大学 一种基于蒙特卡洛模拟的在途车辆运行风险量化计算方法

Also Published As

Publication number Publication date
CN112487617B (zh) 2024-03-08
CN112487617A (zh) 2021-03-12

Similar Documents

Publication Publication Date Title
WO2021203972A1 (zh) 基于碰撞模型的风险预防方法、装置、设备及存储介质
Jeong et al. Is vehicle automation enough to prevent crashes? Role of traffic operations in automated driving environments for traffic safety
Ali et al. Understanding the discretionary lane-changing behaviour in the connected environment
Sander et al. The potential of clustering methods to define intersection test scenarios: Assessing real-life performance of AEB
Jia et al. Long short‐term memory and convolutional neural network for abnormal driving behaviour recognition
Chen et al. Using vehicular trajectory data to explore risky factors and unobserved heterogeneity during lane-changing
Ali et al. Cooperate or not? Exploring drivers’ interactions and response times to a lane-changing request in a connected environment
Hu et al. Efficient mapping of crash risk at intersections with connected vehicle data and deep learning models
Goyani et al. Investigation of traffic conflicts at unsignalized intersection for reckoning crash probability under mixed traffic conditions
WO2022110611A1 (zh) 一种面向平面交叉口的行人过街行为预测方法
Zhu et al. Multi-vehicle interaction safety of connected automated vehicles in merging area: A real-time risk assessment approach
Zhang et al. Spatiotemporal interaction pattern recognition and risk evolution analysis during lane changes
Platho et al. Predicting velocity profiles of road users at intersections using configurations
Jia et al. Lane‐Changing Behavior Prediction Based on Game Theory and Deep Learning
Li et al. Development and evaluation of high-speed differential warning application using Vehicle-to-Vehicle communication
Yu et al. Exploring the associations between driving volatility and autonomous vehicle hazardous scenarios: Insights from field operational test data
Hua et al. Modeling and simulation of approaching behaviors to signalized intersections based on risk quantification
Wu et al. Lane-GNN: Integrating GNN for Predicting Drivers' Lane Change Intention
Shangguan et al. An empirical investigation of driver car-following risk evolution using naturistic driving data and random parameters multinomial logit model with heterogeneity in means and variances
Chen et al. Analyzing differences of highway lane-changing behavior using vehicle trajectory data
Katrakazas Developing an advanced collision risk model for autonomous vehicles
Wang et al. Driver’s Lane Selection Model Based on Phase‐Field Coupling and Multiplayer Dynamic Game with Incomplete Information
Chen et al. Platoon separation strategy optimization method based on deep cognition of a driver’s behavior at signalized intersections
Gao et al. Stage-specific impacts of hazy weather on car following
Lin et al. Optimization of Lane-changing Advisory in Mixed Traffic of Connected Vehicles and Human-driven Vehicles at Expressway Bottlenecks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21785126

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21785126

Country of ref document: EP

Kind code of ref document: A1