CN112215487A - A method of vehicle driving risk prediction based on neural network model - Google Patents

A method of vehicle driving risk prediction based on neural network model Download PDF

Info

Publication number
CN112215487A
CN112215487A CN202011076552.2A CN202011076552A CN112215487A CN 112215487 A CN112215487 A CN 112215487A CN 202011076552 A CN202011076552 A CN 202011076552A CN 112215487 A CN112215487 A CN 112215487A
Authority
CN
China
Prior art keywords
vehicle
data
neural network
network model
label
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202011076552.2A
Other languages
Chinese (zh)
Other versions
CN112215487B (en
Inventor
胡宏宇
王�琦
杜来刚
鲁子洋
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.)
Jilin University
Original Assignee
Jilin University
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 Jilin University filed Critical Jilin University
Priority to CN202011076552.2A priority Critical patent/CN112215487B/en
Publication of CN112215487A publication Critical patent/CN112215487A/en
Application granted granted Critical
Publication of CN112215487B publication Critical patent/CN112215487B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • 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
    • G08G1/0125Traffic data processing
    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G1/00Traffic control systems for road vehicles
    • G08G1/16Anti-collision systems
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Software Systems (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • General Business, Economics & Management (AREA)
  • Artificial Intelligence (AREA)
  • Tourism & Hospitality (AREA)
  • Molecular Biology (AREA)
  • Marketing (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Educational Administration (AREA)
  • Development Economics (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Primary Health Care (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a vehicle running risk prediction method based on a neural network model, which comprises the following steps: collecting vehicle running data to form vehicle historical data; step two: extracting the characteristics of the historical data of the vehicle by adopting a context time window to form statistical characteristics; step three: extracting the statistical characteristics, and step four: dividing the extraction result data in the third step, and the fifth step: constructing a neural network; construction of LSTM encoder-1 DCNN-LSTAn M decoder network architecture; step six: unlabeled dataset is denoted as { XU}; the tagged data set is divided into a training set and a testing set, wherein the training set is marked as SL={XL,YL}; using tagged data sets SLPre-training a neural network; then entering a self-learning stage to carry out a self-learning process on the set YPLPredicting, and directly taking the predicted value as a real label; after the completion, all the labels of the label-free data and the trained network model are obtained.

Description

一种基于神经网络模型的车辆行驶风险预测方法A method of vehicle driving risk prediction based on neural network model

技术领域technical field

本发明涉及机器学习领域,更具体的是,本发明涉及一种基于神经网络模型的车辆行驶风险预测方法。The present invention relates to the field of machine learning, and more particularly, to a method for predicting vehicle driving risk based on a neural network model.

背景技术Background technique

根据世界卫生组织于2018年的统计,每年有135万人因交通事故丧生,并且道路交通碰撞带来的损失可达大部分国家国内生产总值的3%。此外,在交通相关的死亡事故中,约94%是由驾驶员引起的,驾驶员的不当驾驶行为成为了交通事故中的一个主要因素。这些驾驶行为往往是由于驾驶员对周围环境的不良感知、冒失或者激进的判断与决策、不适当的车辆驾驶操纵造成的。行驶风险评估是根据当前与过去时刻的各项行驶特征(包括驾驶员、车辆与周围环境)进行分析,给出当前车辆碰撞或发生其他交通事故的可能性。对车辆行驶安全性进行评估与预测,及时向驾驶员进行反馈,以提高车辆行驶安全性,进而减少交通事故。因此,对车辆行驶风险进行评估与预测是必不可少的。According to the World Health Organization's statistics in 2018, 1.35 million people are killed in traffic accidents every year, and the damage caused by road traffic collisions can reach 3% of the gross domestic product of most countries. In addition, about 94% of traffic-related fatalities are caused by drivers, and driver's improper driving behavior has become a major factor in traffic accidents. These driving behaviors are often caused by the driver's poor perception of the surrounding environment, rash or aggressive judgments and decisions, and inappropriate vehicle driving maneuvers. Driving risk assessment is based on the analysis of various driving characteristics (including driver, vehicle and surrounding environment) at the current and past moments, and gives the possibility of current vehicle collision or other traffic accidents. Evaluate and predict vehicle driving safety, and provide timely feedback to drivers to improve vehicle driving safety and reduce traffic accidents. Therefore, it is essential to evaluate and predict vehicle driving risks.

然而在行驶风险评估任务中,对带有风险的行驶数据进行标签是一个颇具挑战的任务。若运用无监督学习的方法对数据进行分类,所得到的结果可能并不是严格按照风险水平的高低进行类别划分,并且在准确率上也难以达到满意的效果。此外,行驶风险评估任务需要面对海量高维度、带时序、类别不均衡的车辆行驶数据。最后,行驶风险评估对准确率有较高的要求,并且难以接受高风险被判定为无风险这种情况。综上所述,对行驶风险进行全面、准确、高效的评估存在着巨大的挑战。However, in the driving risk assessment task, labeling the driving data with risk is a rather challenging task. If the unsupervised learning method is used to classify the data, the obtained results may not be classified strictly according to the level of risk, and it is difficult to achieve satisfactory results in terms of accuracy. In addition, the task of driving risk assessment needs to deal with massive high-dimensional, time-series and unbalanced vehicle driving data. Finally, driving risk assessment has high requirements on accuracy, and it is difficult to accept the situation that high risk is judged as no risk. To sum up, it is a huge challenge to conduct a comprehensive, accurate and efficient assessment of driving risk.

经过检索,中国发明专利CN201711234967.6公开了一种行驶风险警示方法和装置,用预先建立的BP网络将相应路段分类为高风险路段或低风险路段;向所述车辆发出警示信息,以控制所述车辆在行驶至高风险路段时发出警示;CN201910574565.3公开了一种基于北斗定位系统的车辆违规行驶风险分析方法,通过准确计算出车辆的违规行驶的风险分值,由用户根据该车辆的违规行驶的风险分值生成违规行驶的风险分析报告,提醒和督促司机改善驾驶行为,对司机本人的驾驶行为起到了预警和考核的作用。但是上述方法,对车辆行驶过程的海量高维度、带时序、类别不均衡的车辆行驶数据考虑不够周全,并且难以做到精细化的行驶风险预测,在准确性上较差。After retrieval, Chinese invention patent CN201711234967.6 discloses a driving risk warning method and device, which uses a pre-established BP network to classify corresponding road sections into high-risk sections or low-risk sections; The above-mentioned vehicle issues a warning when driving to a high-risk road section; CN201910574565.3 discloses a method for analyzing the risk of illegal driving of vehicles based on the Beidou positioning system. The driving risk score generates a risk analysis report of illegal driving, reminds and urges the driver to improve driving behavior, and plays an early warning and assessment role for the driver's own driving behavior. However, the above methods do not fully consider the massive, high-dimensional, time-series, and unbalanced vehicle driving data of the vehicle driving process, and it is difficult to achieve refined driving risk prediction, which is poor in accuracy.

发明内容SUMMARY OF THE INVENTION

本发明设计开发了一种基于神经网络模型的车辆行驶风险预测方法,该方法可以仅在一小部分数据上人工标签,并且自动学习潜在的特征,建立神经网络模型,并且可以预测将来一段时间内的风险值。The present invention designs and develops a method for predicting vehicle driving risk based on a neural network model, which can only manually label a small part of data, and automatically learn potential features, establish a neural network model, and predict a certain period of time in the future. value at risk.

本发明的另一个目的是一种用于车辆行驶风险预测的神经网络模型训练方法,该方法可以仅在一小部分数据上人工标签,并且自动学习潜在的特征,建立神经网络模型。Another object of the present invention is a neural network model training method for vehicle driving risk prediction, which can manually label only a small part of data, and automatically learn potential features to build a neural network model.

一种基于神经网络模型的车辆行驶风险预测方法,A vehicle driving risk prediction method based on a neural network model,

步骤一:采集车辆行驶数据,形成车辆历史数据;Step 1: Collect vehicle driving data to form vehicle historical data;

步骤二:采用上下文时间窗对所述车辆历史数据进行特征的提取,形成统计特征;Step 2: using the context time window to extract features from the vehicle historical data to form statistical features;

步骤三:对所述统计特征进行提取,包括:车辆的类型,车辆的长和宽;转向熵值;参数逆碰撞时间TTC-1、逆车头时距THW-1和逆车头间距DHW-1;车辆在无换道意图时压虚线时长,车辆压实线时长,车辆驶出到实线外时长;局部交通流密度、局部速度差异和局部加速度差异。Step 3: extracting the statistical features, including: the type of the vehicle, the length and width of the vehicle; the steering entropy value; the parameters reverse collision time TTC -1 , reverse headway THW -1 and reverse head distance DHW -1 ; The duration of the vehicle pressing the dotted line when there is no intention to change lanes, the duration of the vehicle pressing the line, and the duration of the vehicle driving out of the solid line; the local traffic flow density, the local speed difference and the local acceleration difference.

步骤四:对步骤三的提取结果数据进行划分,从中随机抽出不超过5%的数据进行标签,形成带标签数据集;剩余的数据为无标签数据集,用于半监督学习的无标签训练与测试;Step 4: Divide the extraction result data of Step 3, and randomly extract no more than 5% of the data for labeling to form a labeled data set; the remaining data is an unlabeled data set, which is used for unlabeled training and semi-supervised learning. test;

步骤五:构建神经网络;构建LSTM编码器-1DCNN-LSTM解码器网络架构;Step 5: Build a neural network; build an LSTM encoder-1DCNN-LSTM decoder network architecture;

步骤六:无标签数据集记做{XU};将带标签数据集划分为训练集与测试集,其中训练集记做SL={XL,YL};运用带标签数据集SL对神经网络进行预训练;Step 6: The unlabeled data set is recorded as {X U }; the labeled data set is divided into training set and test set, and the training set is recorded as SL ={X L , Y L }; the labeled data set SL is used Pre-training the neural network;

然后进入自学习阶段,将无标签数据集{XU}运用预训练的网络,生成伪标签{YP};对于每个生成的伪标签,都带有一定的置信度ε,将该置信度与阈值εth比较,大于该阈值的集合记做SP t={XUh t,YPh t},小于该阈值的集合记做{YPL t},t为迭代次数;对于集合{YPh t},根据流形假设,认为该伪标签即真实标签;将集合SL与SP t合并,形成新的集合SL t,再将其用于网络的训练中;对于{XU t},运用再次训练好的网络进行伪标签重新生成;对自学习最后阶段仍未标签的数据集合{XU mst}进行预测,以预测值直接作为其真实标签;完成后,得到了所有的无标签数据的标签与训练好的网络模型。Then enter the self-learning stage, use the unlabeled data set {X U } to use the pre-trained network to generate pseudo-labels {Y P }; for each generated pseudo-label, there is a certain degree of confidence ε, the confidence degree Compared with the threshold ε th , the set greater than the threshold is denoted as SP t = {X Uh t ,Y Ph t }, the set smaller than the threshold is denoted as {Y PL t }, and t is the number of iterations; for the set {Y Ph t t }, according to the manifold hypothesis, the pseudo-label is considered to be the real label; the set SL and SP t are combined to form a new set SL t , which is then used in the training of the network; for {X U t } , use the retrained network to regenerate pseudo-labels; predict the unlabeled data set {X U mst } in the final stage of self-learning, and use the predicted value directly as its true label; after completion, all unlabeled data sets are obtained. The labels of the data and the trained network model.

作为一种优选,所述神经网络的惩罚函数为:As a preference, the penalty function of the neural network is:

Figure BDA0002716989520000031
Figure BDA0002716989520000031

其中,分布P的概率质量函数f可以被定义为下式where the probability mass function f of the distribution P can be defined as

Figure BDA0002716989520000032
Figure BDA0002716989520000032

Figure BDA0002716989520000033
Figure BDA0002716989520000033

AOBC(0,k)=AOBC(k)AOBC(0,k)=AOBC(k)

Figure BDA0002716989520000034
Figure BDA0002716989520000034

其中,AOBC(k)中的N(t,k)=|SL,k t|,OBC(k)中的N(0,k)=|SL,k 0|;N为mini-batch中数据的数量,m为类别数,yik为真实值,

Figure BDA0002716989520000035
为预测值。Among them, N(t,k)=| SL,k t | in AOBC(k), N(0,k)=| SL,k 0 | in OBC(k); N is in the mini-batch The number of data, m is the number of categories, y ik is the real value,
Figure BDA0002716989520000035
is the predicted value.

作为一种优选,还包括极限值惩罚函数,如下式所示:As an option, the limit value penalty function is also included, as shown in the following formula:

Figure BDA0002716989520000041
Figure BDA0002716989520000041

其中,ev为极限值。Among them, ev is the limit value.

作为一种优选,所述转向熵值SRE:As a preference, the steering entropy value SRE:

Figure BDA0002716989520000042
Figure BDA0002716989520000042

作为一种优选,所述逆碰撞时间TTC-1 As a preference, the inverse collision time TTC -1

Figure BDA0002716989520000047
Figure BDA0002716989520000047

作为一种优选,所述局部交通流密度具体计算公式下:As a preference, the specific calculation formula of the local traffic flow density is as follows:

Figure BDA0002716989520000043
Figure BDA0002716989520000043

其中,Xj=(xj,yj)T为感兴趣车辆,μ=(x,y)T为目标车辆中心坐标,

Figure BDA0002716989520000044
其中σx与σy通过下式定义:Among them, X j =(x j ,y j ) T is the vehicle of interest, μ=(x, y) T is the center coordinate of the target vehicle,
Figure BDA0002716989520000044
where σ x and σ y are defined by:

σx=|vx|+k1Lσ x =|v x |+k 1 L

σy=|vy|+k2Wσ y =|v y |+k 2 W

其中,vx,vy为车辆的横向与纵向速度,k1与k2为补偿因子。Among them, v x , v y are the lateral and longitudinal speeds of the vehicle, and k 1 and k 2 are compensation factors.

作为一种优选,所述局部速度差异计算如下:As a preference, the local velocity difference is calculated as follows:

Figure BDA0002716989520000045
Figure BDA0002716989520000045

作为一种优选,所述局部加速度差异计算如下:As a preference, the local acceleration difference is calculated as follows:

Figure BDA0002716989520000046
Figure BDA0002716989520000046

本发明所述的有益效果:The beneficial effects of the present invention:

该方法首先提取了目标车辆行驶、车路交互、局部交通状况等特征。对海量的车辆行驶数据进行特征提取,取一小部分进行人工标签,从而获得了带标签的小数据集与无标签的大数据集。搭建了一维卷积神经网络(1D-CNN)与长短时记忆网络(LSTM)相结合的卷积神经网络,并且采用了自适应过均衡交叉熵作为神经网络的损失函数。将上述神经网络嵌入至半监督学习框架中,对两个数据集进行预训练、自学习与微调后得到了最终的标签结果与训练好的网络。最终,添加极限值惩罚模块以完善模型。The method first extracts the characteristics of target vehicle driving, vehicle-road interaction, and local traffic conditions. Feature extraction is performed on the massive vehicle driving data, and a small part is manually labeled, so as to obtain a small data set with labels and a large data set without labels. A convolutional neural network combining a one-dimensional convolutional neural network (1D-CNN) and a long short-term memory network (LSTM) is built, and an adaptive over-balanced cross-entropy is used as the loss function of the neural network. The above neural network is embedded into the semi-supervised learning framework, and the final label results and trained network are obtained after pre-training, self-learning and fine-tuning on the two datasets. Finally, a limit penalty block is added to refine the model.

采用了代价敏感的半监督深度学习方法,对车辆行驶数据进行分析,得到当前与将来的行驶风险分数,且该分数为介于0~3的连续值,对风险评估更加精细。该方法可以应用于ADAS中的行驶风险警示系统,从而给予驾驶员及时的反馈。A cost-sensitive semi-supervised deep learning method is used to analyze the vehicle driving data to obtain the current and future driving risk scores, and the score is a continuous value between 0 and 3, which makes the risk assessment more refined. This method can be applied to the driving risk warning system in ADAS to give timely feedback to the driver.

仅运用小部分标签数据就可以得到良好的结果,极大地解决了大量无标签数据的标签问题。采用了自适应过均衡交叉熵损失函数,对类别不均衡的半监督深度学习的训练性能有着较大的性能提升。该损失函数可以使得网络在整个训练过程处于过均衡状态,该状态可以有效的提高高风险数据的检测准确率。该方法可以应用于其他相关的类似场景之中。Good results can be obtained by using only a small part of labeled data, which greatly solves the labeling problem of a large amount of unlabeled data. The adaptive over-balanced cross-entropy loss function is adopted, which greatly improves the training performance of semi-supervised deep learning with unbalanced categories. The loss function can make the network in an over-balanced state during the entire training process, which can effectively improve the detection accuracy of high-risk data. This method can be applied to other related similar scenarios.

提出了局部交通状况描述子,对目标车辆周围的交通状况、速度与加速度差异进行描述,并给予了一个描述子,从而简化了周围场景的描述。该描述子也可以用于其他类似的领域之中。A local traffic condition descriptor is proposed to describe the traffic conditions, speed and acceleration differences around the target vehicle, and a descriptor is given to simplify the description of the surrounding scene. The descriptor can also be used in other similar fields.

附图说明Description of drawings

图1为本发明上下文窗口示意图。FIG. 1 is a schematic diagram of the context window of the present invention.

图2为本发明以目标车辆为基准的感兴趣车辆示意图。FIG. 2 is a schematic diagram of a vehicle of interest based on a target vehicle according to the present invention.

图3为本发明深度学习网络模型。FIG. 3 is a deep learning network model of the present invention.

图4为本发明半监督学习框架。Figure 4 is the semi-supervised learning framework of the present invention.

图5为本发明总体框架。Fig. 5 is the general framework of the present invention.

具体实施方式Detailed ways

下面结合附图对本发明做进一步的详细说明,以令本领域技术人员参照说明书文字能够据以实施。The present invention will be further described in detail below with reference to the accompanying drawings, so that those skilled in the art can implement it with reference to the description.

本发明提出了一种基于自适应代价敏感半监督深度学习的行驶风险评估与预测方法。该方法首先提取了目标车辆行驶、车路交互、局部交通状况等特征。对海量的车辆行驶数据进行特征提取,取一小部分进行人工标签,从而获得了带标签的小数据集与无标签的大数据集。搭建了一维卷积神经网络(1D-CNN)与长短时记忆网络(LSTM)相结合的卷积神经网络,并且采用了自适应过均衡交叉熵作为神经网络的损失函数。将上述神经网络嵌入至半监督学习框架中,对两个数据集进行预训练、自学习与微调后得到了最终的标签结果与训练好的网络。作为一种优选,最终,添加极限值惩罚模块以完善模型。该方法可以仅在一小部分数据上人工标签,并且自动学习潜在的特征,并且可以预测将来一段时间内的风险值,从而及时的向驾驶员及时反馈。The invention proposes a driving risk assessment and prediction method based on adaptive cost-sensitive semi-supervised deep learning. The method first extracts the characteristics of target vehicle driving, vehicle-road interaction, and local traffic conditions. The feature extraction is carried out on the massive vehicle driving data, and a small part is manually labeled, so as to obtain a small data set with labels and a large data set without labels. A convolutional neural network combining a one-dimensional convolutional neural network (1D-CNN) and a long short-term memory network (LSTM) is built, and an adaptive over-balanced cross-entropy is used as the loss function of the neural network. The above neural network is embedded into the semi-supervised learning framework, and the final label results and trained network are obtained after pre-training, self-learning and fine-tuning on the two datasets. As a preference, finally, a limit value penalty module is added to refine the model. This method can manually label only a small part of the data, and automatically learn the potential features, and can predict the risk value in a period of time in the future, so as to provide timely feedback to the driver.

步骤1:对车辆行驶历史数据进行转换与统计。由于直接将原始车辆行驶数据作为输入将会导致特征过于稀疏、特征阶数过低,从而增加训练难度。采用上下文时间窗(如图1所示)对一个固定的窗口进行特征的提取。选择窗宽为Ww=3秒,每秒的采样频率不小于10Hz。共选择5秒的历史数据,每个时间窗重叠率Ov=66.67%,即2秒。故共获得3个时间窗,并且对窗口内的所有数据进行统计。Step 1: Convert and count the vehicle driving history data. Directly using the original vehicle driving data as input will result in too sparse features and too low feature order, thus increasing the difficulty of training. The context time window (as shown in Figure 1) is used to extract features from a fixed window. The selected window width is W w =3 seconds, and the sampling frequency per second is not less than 10 Hz. A total of 5 seconds of historical data is selected, and the overlap rate of each time window is Ov=66.67%, that is, 2 seconds. Therefore, a total of 3 time windows are obtained, and all data in the window are counted.

步骤2:对统计特征进行提取。共提取47维特征,包含了车辆基本信息、车辆的行驶与交互信息、车辆周边环境信息等。每个统计数据都存在均值、最大值、最小值、方差统计量,故每一个数据存在4个统计量。Step 2: Extract statistical features. A total of 47-dimensional features are extracted, including basic vehicle information, vehicle driving and interaction information, and vehicle surrounding environment information. Each statistical data has mean, maximum, minimum and variance statistics, so each data has 4 statistics.

A)车辆的基本信息:A) Basic information of the vehicle:

车辆的类型、车辆的长L与宽W。Type of vehicle, length L and width W of the vehicle.

B)目标车辆行驶特征:B) The driving characteristics of the target vehicle:

定义车辆纵向行驶方向为x轴,横向行驶方向为y轴。选择车辆在横向、纵向的速度与加速度数据,对每个时间窗内的车辆在横向、纵向的速度与加速度数据提取均值、最大值、最小值、方差作为输入特征,总计16维特征。驾驶员在稳定、正常驾驶时,转向行为会比较平滑,然而驾驶员在疲劳、分心等状态下转向行为可能会有所混乱。运用转向熵来定量的指示驾驶员的方向操纵特性,以反映转向平稳程度与行驶安全性。Define the longitudinal travel direction of the vehicle as the x-axis and the lateral travel direction as the y-axis. Select the horizontal and vertical speed and acceleration data of the vehicle, and extract the mean, maximum, minimum, and variance of the vehicle's horizontal and vertical speed and acceleration data in each time window as input features, with a total of 16-dimensional features. When the driver is driving steadily and normally, the steering behavior will be relatively smooth, but the steering behavior may be chaotic when the driver is fatigued or distracted. Steering entropy is used to quantitatively indicate the driver's directional handling characteristics to reflect the degree of steering stability and driving safety.

记时间窗内航向角为θ=(θ123,L,θm),m为时间窗内数据的数目。运用给定时间内泰勒二阶展开式预测下一个转角,如下式所示:The heading angle in the time window is recorded as θ=(θ 1 , θ 2 , θ 3 , L, θ m ), and m is the number of data in the time window. Use Taylor's second-order expansion for a given time to predict the next corner as follows:

Figure BDA0002716989520000071
Figure BDA0002716989520000071

即:

Figure BDA0002716989520000072
which is:
Figure BDA0002716989520000072

误差函数定义:Error function definition:

Figure BDA0002716989520000073
Figure BDA0002716989520000073

式中,

Figure BDA0002716989520000077
为第n时刻θn的预测值。设置一个α值,作为一种优选α取值0.035;使90%数据的预测误差落在-α与α之间。将这个预测误差区间划分为9段,即共9个区间,然后求取各段分布频率,即pi,i=1,2,..9。运用下式计算转向熵值SRE:In the formula,
Figure BDA0002716989520000077
is the predicted value of θ n at the nth time. Set an α value, as a preferred α value of 0.035; make the prediction error of 90% of the data fall between -α and α. This prediction error interval is divided into 9 sections, that is, a total of 9 sections, and then the distribution frequency of each section is obtained, that is, p i , i=1,2,..9. Use the following formula to calculate the steering entropy value SRE:

Figure BDA0002716989520000074
Figure BDA0002716989520000074

C)车辆间的交互特征:C) Interaction characteristics between vehicles:

目标车辆速度记为ve,纵向坐标xe,前车速度记为vp,纵向坐标xp。THW车头时距,即指的是在同一车道上行驶的车辆队列中,两连续车辆车头端部通过某一断面的时间间隔;DHW车头间距,即指的是在同一车道上行驶的车辆队列中,两连续车辆车头端部通过某一断面的距离间隔;TTC碰撞时间,即后车与前车均按照当前速度行驶时,两车的碰撞时间。则THW、DHW、TTC的计算公式如下所示。The speed of the target vehicle is denoted as ve , the longitudinal coordinate x e , the speed of the preceding vehicle is denoted as v p , and the longitudinal coordinate x p . THW head-to-head distance refers to the time interval between the end of two consecutive vehicles passing through a certain section in the vehicle queue driving on the same lane; DHW head-to-head distance refers to the vehicle queue driving on the same lane. , the distance interval between the front ends of two consecutive vehicles passing through a certain section; TTC collision time, that is, the collision time of the two vehicles when both the rear vehicle and the front vehicle are traveling at the current speed. The calculation formulas of THW, DHW, and TTC are as follows.

Figure BDA0002716989520000075
Figure BDA0002716989520000075

DHW=(xp-xe)DHW=(x p -x e )

Figure BDA0002716989520000076
Figure BDA0002716989520000076

然而,以TTC为例,直接运用TTC可能会出现以下两种情况:当后车速度比前车速度低时,即以当前的相对速度后车永远无法追上前车,TTC为负值;有如果后车速度比前车速度仅快一点,即以当前的相对速度后车追上前车需要耗费很长时间,TTC为很大的正值。故TTC的值域在理论上为(-∞,+∞),而真正高风险TTC区间是很小的。故直接将TTC作为输入可能会导致模型准确率下降,此外通过后续特征标准化处理可能会再次压缩高风险TTC区间。所以,采用逆碰撞时间TTC-1,如下式所示。However, taking TTC as an example, the following two situations may occur when using TTC directly: when the speed of the following vehicle is lower than the speed of the preceding vehicle, that is, the following vehicle can never catch up with the preceding vehicle at the current relative speed, and the TTC is negative; If the speed of the car behind is only a little faster than the speed of the car in front, that is, it takes a long time for the car behind to catch up with the car in front at the current relative speed, and the TTC is a large positive value. Therefore, the value range of TTC is theoretically (-∞, +∞), and the real high-risk TTC interval is very small. Therefore, directly using TTC as an input may lead to a decrease in the accuracy of the model. In addition, the high-risk TTC interval may be compressed again through subsequent feature standardization. Therefore, the inverse collision time TTC -1 is used, as shown in the following equation.

Figure BDA0002716989520000081
Figure BDA0002716989520000081

此外,对于负值的TTC,为了简化特征,将所有负值TTC统一赋予一个足够大的正值TTC-1 max(一般取50秒),即经过很久目标车辆才会与前车碰撞。这种无关值剔除替换的方法可以减小对模型的负面混淆影响,提高训练准确度。用公式表述如下:In addition, for negative TTCs, in order to simplify the features, all negative TTCs are uniformly assigned a sufficiently large positive value TTC -1 max (usually 50 seconds), that is, the target vehicle will collide with the preceding vehicle after a long time. This method of eliminating and replacing irrelevant values can reduce the negative confounding effect on the model and improve the training accuracy. The formula is expressed as follows:

Figure BDA0002716989520000082
Figure BDA0002716989520000082

Figure BDA0002716989520000083
其中,THW-1 max=10;
Figure BDA0002716989520000083
Wherein, THW -1 max =10;

Figure BDA0002716989520000084
其中,DHW-1 max=200;
Figure BDA0002716989520000084
Wherein, DHW -1 max =200;

经过处理,相对于安全的TTC值被压缩到了0至一个很小的值之间,而高风险的TTC值被放大,值域也被放大,从而提高模型的准确度。同理,将THW与DHW变换为逆车头时距THW-1和逆车头间距DHW-1,以放大高风险THW、DHW的值域。最后选择最大值,均值以及方差作为输出特征,共9维。After processing, the safe TTC value is compressed to between 0 and a small value, while the high-risk TTC value is enlarged, and the value range is also enlarged, thereby improving the accuracy of the model. In the same way, THW and DHW are transformed into reverse headway time distance THW -1 and reverse headway distance DHW -1 to enlarge the value range of high-risk THW and DHW. Finally, the maximum value, mean and variance are selected as output features, with a total of 9 dimensions.

出现换道意图时,驾驶员对待换车道的感知是十分重要的。若没有对本车道与待换车道进行良好的感知,直接进行换道操作,可能会带来较高的行驶风险。故选择换道意图出现时待换车道上是否有并列行驶的车辆,与待换车道前车、后车最大TTC-1和THW-1这3维特征作为输入。When there is an intention to change lanes, the driver's perception of changing lanes is very important. If you do not have a good perception of the current lane and the lane to be changed, and directly change the lane, it may bring a higher driving risk. Therefore, it is selected whether there are vehicles running side by side in the lane to be changed when the lane change intention appears, and the 3-dimensional features of the maximum TTC -1 and THW -1 of the vehicle in front of and behind the lane to be changed are used as input.

D)车辆与道路之间的交互D) Interaction between vehicle and road

车辆在无换道意图时压虚线时长,车辆压实线时长,车辆驶出到实线外时长。The length of time that the vehicle presses the dotted line when there is no intention to change lanes, the length of time that the vehicle presses the line, and the time that the vehicle drives out of the solid line.

E)局部交通状况描述子E) Local Traffic Situation Descriptor

为了更好的描述车辆行驶过程中其他行驶的车辆、道路、障碍物等,需要相应的描述子进行描述。首先考虑目标车辆周围行驶的车辆、障碍物等因素。提出基于高斯权重的局部车流密度描述子(LTD)。以目标车辆为基准,考虑前车、左前车、右前车、左车、右车、左后车、右后车、后车这8辆车作为感兴趣车辆(如图2所示,若无此车,则记为0),计算感兴趣车辆对目标车辆的车流密度贡献度。In order to better describe other running vehicles, roads, obstacles, etc. during the driving process of the vehicle, corresponding descriptors are required for description. First, consider factors such as vehicles and obstacles driving around the target vehicle. A Gaussian weight-based local traffic density descriptor (LTD) is proposed. Taking the target vehicle as the benchmark, consider the front vehicle, the left front vehicle, the right front vehicle, the left vehicle, the right vehicle, the left rear vehicle, the right rear vehicle, and the rear vehicle as the vehicle of interest (as shown in Figure 2, if there is no such vehicle) If the vehicle is a vehicle, it is recorded as 0), and the contribution of the vehicle of interest to the traffic density of the target vehicle is calculated.

具体计算公式下:The specific calculation formula is as follows:

Figure BDA0002716989520000091
Figure BDA0002716989520000091

其中,Xj=(xj,yj)T为感兴趣车辆,xj,yj分别是感兴趣车辆横坐标与纵坐标;μ=(x,y)T为目标车辆中心坐标,

Figure BDA0002716989520000092
其中σx与σy通过下式定义:Among them, X j =(x j , y j ) T is the vehicle of interest, x j , y j are the abscissa and ordinate of the vehicle of interest respectively; μ = (x, y) T is the center coordinate of the target vehicle,
Figure BDA0002716989520000092
where σ x and σ y are defined by:

σx=|vx|+k1Lσ x =|v x |+k 1 L

σy=|vy|+k2Wσ y =|v y |+k 2 W

其中,vx,vy为车辆的横向与纵向速度,k1与k2为补偿因子,作为一种优选,k1=0.625,k2=1.25。Wherein, v x , v y are the lateral and longitudinal speeds of the vehicle, and k 1 and k 2 are compensation factors. As an example, k 1 =0.625 and k 2 =1.25.

至此,每个时间窗内的每条数据都形成一个对应的局部车流密度描述子。此外,将局部车流密度作为权重,求解局部速度差异,得到局部速度差异描述子,用来描述以目标车辆为基准周围车辆的行驶速度差异(LVD)。同理,求解局部加速度差异(LAD)。如下式所示:So far, each piece of data in each time window forms a corresponding local traffic density descriptor. In addition, using the local traffic density as the weight, the local speed difference is solved, and the local speed difference descriptor is obtained, which is used to describe the speed difference (LVD) of the surrounding vehicles with the target vehicle as the benchmark. Similarly, solve for the local acceleration difference (LAD). As shown in the following formula:

Figure BDA0002716989520000093
Figure BDA0002716989520000093

Figure BDA0002716989520000094
Figure BDA0002716989520000094

ve为目标车辆速度,前文提过,vj为感兴趣车辆的速度,ae为目标车辆加速度,aj为感兴趣车辆加速度,Ni为感兴趣车辆数目,为最大值为8,如图2所示,若没有对应车辆则减少之,最少为0,即一辆感兴趣车辆都没有,即周围无车辆。v e is the speed of the target vehicle, as mentioned above, v j is the speed of the vehicle of interest, a e is the acceleration of the target vehicle, a j is the acceleration of the vehicle of interest, Ni is the number of vehicles of interest, the maximum value is 8, such as As shown in Figure 2, if there is no corresponding vehicle, it will be reduced, at least 0, that is, there is no vehicle of interest, that is, there is no vehicle around.

至于障碍物,可以将其看做行驶速度为零的车辆。最后,对上述3个描述子分别求取其均值、最大值、最小值、标准差等统计指标,共12维特征。As for the obstacle, think of it as a vehicle traveling at zero speed. Finally, the statistical indicators such as the mean, maximum, minimum, and standard deviation of the above-mentioned three descriptors are obtained respectively, with a total of 12-dimensional features.

下表为特征统计:The following table is the characteristic statistics:

表1统计输入特征Table 1 Statistical input features

Figure BDA0002716989520000101
Figure BDA0002716989520000101

步骤3:将所有经过统计的数据进行划分,从中随机抽出不超过5%的数据进行标签,剩余的数据用于半监督学习的无标签训练与测试。按照2~5%的上分位值(即该数据超过了所有数据的值的95~98%)对当前时刻的风险分数进行评估,标注值为0(良好)、1(一般)、2(较差)、3(很差)。采用标签方法是对数据进行标签,对最终的结果取平均后四舍五入为整数。Step 3: Divide all the statistical data, randomly extract no more than 5% of the data for labeling, and use the remaining data for unlabeled training and testing of semi-supervised learning. The risk score at the current moment is evaluated according to the upper quantile value of 2 to 5% (that is, the data exceeds 95 to 98% of the value of all data), and the marked values are 0 (good), 1 (fair), 2 ( poor), 3 (very poor). The labeling method is to label the data, and the final result is averaged and rounded to an integer.

步骤4:构建神经网络。构建LSTM编码器-1DCNN-LSTM解码器网络架构如图所示。其中,Convolution为1D-CNN,MaxPooling为最大池化层,Dropout为丢弃层,FC为全连接层,Softmax为激活层。除最后一层外,其他的所有卷积层与全连接层激活函数为ReLU。Step 4: Build the neural network. Building the LSTM encoder-1DCNN-LSTM decoder network architecture is shown in the figure. Among them, Convolution is 1D-CNN, MaxPooling is the maximum pooling layer, Dropout is the discarding layer, FC is the fully connected layer, and Softmax is the activation layer. Except for the last layer, all other convolutional layers and fully connected layers have activation functions of ReLU.

具体如下:构建神经网络。构建LSTM编码器-1DCNN-LSTM解码器网络架构,如图所示。首先,将统计数据进行嵌入,通过一个嵌入层,将数据映射至LSTM编码器。嵌入层输入节点数为特征维度,输出为128。LSTM编码器共128个隐藏单元。最终取最后一个隐藏单元的张量进行激活与变形,得到一个一维张量。该张量经过三次一维卷积-激活-一维池化-随机丢弃,得到经过卷积后的张量。该卷积过程主要是提取不同时间序列下的隐藏特征间的深层模式,该模式可以更好地反映风险信息。三个卷积层卷积通道数目分别为64,128,256。最后一层的输出展开后,第一个分支接两个全连接层与Softmax层得到当前的风险分数,隐藏单元个数分别为128,64,4。另外的一个分支接入一个LSTM解码器,对当前的隐藏特征进行解码,得到未来风险值的预测,隐藏单元个数分别为解码器128个LSTM单元,全连接层128,64,4。计算机采用的环境是Win10,使用软件名称Python3.7,深度学习框架为Keras2.2.4,backend为Tensorflow1.14The details are as follows: Build a neural network. Build the LSTM encoder-1DCNN-LSTM decoder network architecture as shown. First, the statistics are embedded, and the data is mapped to the LSTM encoder through an embedding layer. The number of input nodes in the embedding layer is the feature dimension, and the output is 128. The LSTM encoder has a total of 128 hidden units. Finally, the tensor of the last hidden unit is taken for activation and deformation, and a one-dimensional tensor is obtained. The tensor undergoes three 1D convolutions - activation - 1D pooling - random discarding to obtain a convolved tensor. The convolution process is mainly to extract deep patterns between hidden features under different time series, which can better reflect risk information. The number of convolutional channels of the three convolutional layers is 64, 128, and 256, respectively. After the output of the last layer is expanded, the first branch connects two fully connected layers and Softmax layer to obtain the current risk score, and the number of hidden units is 128, 64, and 4, respectively. The other branch is connected to an LSTM decoder, decodes the current hidden features, and obtains the prediction of future risk value. The environment used by the computer is Win10, the software name is Python3.7, the deep learning framework is Keras2.2.4, and the backend is Tensorflow1.14

LSTM编码器将输入进行编码,采用Dropout方法,即在深度学习网络训练的过程中,对于神经网络单元,按照一定的概率将其从网络中随机丢弃,以减小过拟合,作为一种优选一定概率为0.2;接下来是1DCNN(一维卷积神经网络)以减少2DCNN对特征之间卷积所带来的误差。配合最大池化Maxpooling可以选择对风险值贡献大的特征。经过三次卷积-池化-Dropout后,第一个分支接两个全连接层与Softmax层得到当前的风险分数。另一个分支接入一个LSTM解码器,对当前的高级特征进行解码,得到未来风险值的预测。The LSTM encoder encodes the input and adopts the Dropout method, that is, in the process of deep learning network training, for the neural network unit, it is randomly discarded from the network according to a certain probability to reduce overfitting. A certain probability is 0.2; next is a 1DCNN (one-dimensional convolutional neural network) to reduce the error caused by the convolution between the features of the 2DCNN. With the maximum pooling Maxpooling, the features that contribute greatly to the risk value can be selected. After three convolution-pooling-Dropout, the first branch connects two fully connected layers and Softmax layers to get the current risk score. The other branch taps into an LSTM decoder to decode the current high-level features to obtain predictions of future risk values.

步骤5:将上述网络嵌入半监督学习架构,如图4所示。无标签数据集记做{XU}。将带标签数据集划分为训练集与测试集,其中训练集记做SL={XL,YL}。首先,运用带标签数据集SL对网络进行预训练。训练过程中采用Early-Stopping方法,即当某个监控值在经过patience次迭代后网络性能没有提升时,网络自动终止,并且返回停止之前性能最佳的网络权值。监控值选择loss,即loss经过patience次后不再下降。在此处patience选择相对较大,原因是,让网络对少量数据进行充分的学习。Step 5: Embed the above network into a semi-supervised learning architecture, as shown in Figure 4. The unlabeled dataset is denoted as {X U }. The labeled data set is divided into a training set and a test set, wherein the training set is denoted as S L ={ XL , Y L }. First, the network is pre-trained with the labeled dataset SL . The Early-Stopping method is used in the training process, that is, when a certain monitoring value does not improve the network performance after patience iterations, the network automatically terminates, and the network weight with the best performance before the stop is returned. The monitoring value selects loss, that is, the loss will not decrease after patience times. The choice of patience is relatively large here, the reason is to allow the network to learn enough from a small amount of data.

随后进入自学习阶段,将无标签数据集{XU}运用预训练的网络,生成伪标签{YP}。对于每个生成的伪标签,都带有一定的置信度ε(预训练的网络通过Softmax层后直接得到的)。将该置信度与阈值εth比较,大于该阈值的集合记做SP t={XUh t,YPh t},(t为迭代次数)小于该阈值的集合记做{YPL t},t为迭代次数。对于集合{YPh t},根据流形假设,认为该伪标签即真实标签。将集合SL与SP t合并,形成新的集合SL t,再将其用于网络的训练中。对于{XU t},(未被一个迭代过程所接纳的未标签数据集)运用再次训练好的网络进行伪标签重新生成。该过程如下公式所示:Then enter the self-learning stage, and use the unlabeled dataset {X U } to use the pre-trained network to generate pseudo-labels {Y P }. For each generated pseudo-label, there is a certain confidence ε (the pre-trained network is directly obtained after passing through the Softmax layer). Comparing the confidence with the threshold ε th , the set greater than the threshold is denoted as S P t ={X Uh t ,Y Ph t }, and the set (t is the number of iterations) less than the threshold is denoted as {Y PL t }, t is the number of iterations. For the set {Y Ph t }, according to the manifold hypothesis, the pseudo-label is considered to be the real label. The set SL and SP t are merged to form a new set SL t , which is then used in the training of the network. For {X U t }, (unlabeled dataset not accepted by an iterative process) pseudo-label regeneration using the retrained network. The process is shown in the following formula:

Figure BDA0002716989520000121
Figure BDA0002716989520000121

Figure BDA0002716989520000122
Figure BDA0002716989520000122

Figure BDA0002716989520000123
Figure BDA0002716989520000123

S表示训练数据集合,L表示带标签,t表示迭代次数,X表示输入特征,Y表示标签,U表示未标签,P表示伪标签,h表示大于阈值εth,l表示小于阈值εth,mst为总迭代次数。对应的,SL t为第t次迭代时带标签的训练数据集合。其余同理。S represents the training data set, L represents the label, t represents the number of iterations, X represents the input feature, Y represents the label, U represents the unlabeled, P represents the pseudo-label, h represents greater than the threshold ε th , l represents less than the threshold ε th , m st is the total number of iterations. Correspondingly, SL t is the labeled training data set at the t-th iteration. The rest are the same.

当i值越小,该阈值迭代次数逐次减少,原因是置信度越低,越容易引入噪声。同理,patience的次数选择也会随着置信度的下降而减少。When the value of i is smaller, the number of iterations of the threshold decreases successively, because the lower the confidence, the easier it is to introduce noise. Similarly, the number of patient selections will also decrease as the confidence decreases.

作为一种优选,还可以进行模型的微调。微调过程将所有的CNN层与LSTM层设置为不可训练的,仅对全连接层进行微调。原因是,特征提取层已经训练好,且已经学习到了无标签数据集的潜在特征,只需要调节全连接层权值即可。对自学习最后阶段仍未标签的数据集合{XU mst}进行预测,以预测值直接作为其真实标签。完成后,得到了所有的无标签数据的标签,与训练好的网络模型。As an option, fine-tuning of the model can also be performed. The fine-tuning process sets all CNN and LSTM layers as non-trainable, and only fine-tunes the fully connected layers. The reason is that the feature extraction layer has already been trained and the latent features of the unlabeled dataset have been learned, and it is only necessary to adjust the weights of the fully connected layer. Predict the unlabeled data set {X U mst } in the final stage of self-learning, and use the predicted value directly as its true label. After completion, all unlabeled data labels are obtained, and the trained network model is obtained.

步骤6:设置神经网络损失函数。由于类别不均衡,即高风险的数据总是远少于无风险的数据,会导致网络的性能变差。故设置损失函数补偿类别不均衡。选择多类别交叉熵损失函数(CE)作为基本的损失函数,以作为多类别的损失函数,如下式所示:Step 6: Set up the neural network loss function. The performance of the network is degraded due to class imbalance, i.e., high-risk data is always far less than risk-free data. Therefore, the loss function is set to compensate the category imbalance. The multi-class cross-entropy loss function (CE) is selected as the basic loss function as the multi-class loss function, as shown in the following formula:

Figure BDA0002716989520000131
Figure BDA0002716989520000131

其中,Eyi:P为具有分布P的随机变量yi,在yi,k

Figure BDA0002716989520000132
分布P下的数学期望;N为mini-batch中数据的数量,m为类别数,yik为真实值,
Figure BDA0002716989520000133
为预测值。对上述损失函数添加过权重(OBC),即:where E yi:P is a random variable yi with distribution P, where yi,k
Figure BDA0002716989520000132
Mathematical expectation under distribution P; N is the number of data in the mini-batch, m is the number of categories, y ik is the true value,
Figure BDA0002716989520000133
is the predicted value. Overweight (OBC) is added to the above loss function, namely:

Figure BDA0002716989520000134
Figure BDA0002716989520000134

其中,in,

Figure BDA0002716989520000135
Figure BDA0002716989520000135

Figure BDA0002716989520000136
Figure BDA0002716989520000136

然而,在半监督学习不断迭代的过程中,已标签数据数量会不断变化,但是每个类别变化不均。故需要对这一点进行自适应条件,如下式所示:However, during the iterative process of semi-supervised learning, the amount of labeled data is constantly changing, but each category varies unevenly. Therefore, it is necessary to make adaptive conditions for this point, as shown in the following formula:

N(t,k)=|SL t|N(t,k)=|S L t |

Figure BDA0002716989520000141
Figure BDA0002716989520000141

AOBC(0,k)=OBC(k)AOBC(0,k)=OBC(k)

其中,OBC(k)中的N=|SL 0|。Among them, N=|S L 0 | in OBC(k).

上式中||表示取集合中元素个数;AOBC(t,k)为随着迭代次数t的变化,第k类损失的权重。In the above formula || represents the number of elements in the set; AOBC(t,k) is the weight of the k-th type of loss with the change of the number of iterations t.

作为一种优选,步骤7:对上述网络添加极限值惩罚模块。为了弥补神经网络极特殊的严重分类错误,需要引入极限值惩罚模块。该模块采用模糊逻辑,当某些值越接近碰撞发生时刻的参数,基于的惩罚越高。如下式所示:As a preference, step 7: add a limit value penalty module to the above network. In order to make up for the very special serious classification errors of the neural network, a limit value penalty module needs to be introduced. The module uses fuzzy logic, and when certain values are closer to the parameters at the moment of collision, the higher the penalty is based on. As shown in the following formula:

Figure BDA0002716989520000142
Figure BDA0002716989520000142

其中,ev为极限值;一般取为在所有值中超过99.99%的数据作为极限值。Among them, ev is the limit value; generally, the data exceeding 99.99% of all values is taken as the limit value.

作为一种优选,步骤8:对网络进行预训练、自学习与微调。网络的优化器选择为Adam优化器,学习率为10-3,衰减为10-6,εth选择为0.999999、0.99999、0.9999、0.999、0.99、0.95、0.9。经过微调后,即得到所有的数据标签与训练好的网络。每个输出的标签会对应与一个置信度,将所有标签的值与置信度相乘则可得到当前的风险分数。训练好的网络则可以直接用来使用(即用来评估新的数据,而不采用半监督方式)。如图5所示总体框架。As an option, step 8: pre-training, self-learning and fine-tuning of the network. The optimizer of the network is selected as the Adam optimizer, the learning rate is 10 -3 , the decay is 10 -6 , and the ε th is selected as 0.999999, 0.99999, 0.9999, 0.999, 0.99, 0.95, 0.9. After fine-tuning, all the data labels and the trained network are obtained. Each output label corresponds to a confidence level, and the current risk score is obtained by multiplying the values of all labels by the confidence level. The trained network can then be used directly (ie, to evaluate new data without semi-supervised methods). The overall framework is shown in Figure 5.

尽管本发明的实施方案已公开如上,但其并不仅仅限于说明书和实施方式中所列运用,它完全可以被适用于各种适合本发明的领域,对于熟悉本领域的人员而言,可容易地实现另外的修改,因此在不背离权利要求及等同范围所限定的一般概念下,本发明并不限于特定的细节和这里示出与描述的图例。Although the embodiment of the present invention has been disclosed as above, it is not limited to the application listed in the description and the embodiment, and it can be applied to various fields suitable for the present invention. For those skilled in the art, it can be easily Therefore, the invention is not limited to the specific details and illustrations shown and described herein without departing from the general concept defined by the appended claims and the scope of equivalents.

Claims (8)

1. A vehicle running risk prediction method based on a neural network model is characterized in that,
the method comprises the following steps: collecting vehicle running data to form vehicle historical data;
step two: extracting the characteristics of the historical data of the vehicle by adopting a context time window to form statistical characteristics;
step three:extracting the statistical features, including: the type of vehicle, the length and width of the vehicle; a steering entropy value; time to counter collision (TTC) of parameter-1Headway time THW-1And reverse headwear distance DHW-1(ii) a When the vehicle has no lane changing intention, pressing the dotted line for a long time, compacting the line for a long time, and driving the vehicle out of the solid line for a long time; local traffic flow density, local speed differential, and local acceleration differential.
Step four: dividing the extraction result data in the step three, and randomly extracting no more than 5% of data from the extraction result data to carry out labeling to form a labeled data set; the rest data is a label-free data set and is used for label-free training and testing of semi-supervised learning;
step five: constructing a neural network; constructing an LSTM encoder-1 DCNN-LSTM decoder network architecture;
step six: unlabeled dataset is denoted as { XU}; the tagged data set is divided into a training set and a testing set, wherein the training set is marked as SL={XL,YL}; using tagged data sets SLPre-training a neural network;
then entering a self-learning stage, and setting the unlabeled data set { XUApply the pretrained network to generate a pseudo label (Y)P}; each generated pseudo label is provided with a certain confidence coefficient epsilon, and the confidence coefficient is compared with a threshold value epsilonthComparing, the set greater than the threshold is marked as SP t={XUh t,YPh tThe set smaller than the threshold is denoted as { Y }PL tT is iteration times; for the set { YPh tAccording to the manifold assumption, the false label is considered as a real label; will gather SLAnd SP tMerge to form a new set SL tThen the data is used for training the network; for { XU tRe-generating the pseudo label by applying the retrained network; for data set { X) not labeled in the final stage of self-learningU mstPredicting, and directly taking a predicted value as a real label of the predicted value; after completion, all the labels and trainings of the label-free data are obtainedAnd (5) training a network model.
2. The neural network model-based vehicle driving risk prediction method according to claim 1, wherein the penalty function of the neural network is:
Figure FDA0002716989510000021
wherein the probability mass function f of the distribution P can be defined as
Figure FDA0002716989510000022
Figure FDA0002716989510000023
AOBC(0,k)=AOBC(k)
Figure FDA0002716989510000024
Wherein N (t, k) ═ S in aobc (k)L,k tN (0, k) ═ S in obc (k)L,k 0L, |; n is the number of data in the mini-batch, m is the number of categories, yikIn order to be the true value of the value,
Figure FDA0002716989510000025
is a predicted value.
3. The neural network model-based vehicle driving risk prediction method of claim 2, further comprising a limit penalty function as shown in the following formula:
Figure FDA0002716989510000026
wherein ev is a limit value.
4. The neural network model-based vehicle travel risk prediction method of claim 1, characterized in that the steering entropy value SRE:
Figure FDA0002716989510000027
5. the neural network model-based vehicle travel risk prediction method of claim 4, wherein the time to collision TTC-1
Figure FDA0002716989510000031
6. The neural network model-based vehicle travel risk prediction method according to claim 1 or 5, characterized in that the local traffic flow density is specifically calculated according to the formula:
Figure FDA0002716989510000032
wherein, Xj=(xj,yj)TFor the vehicle of interest, μ ═ x, yTIs the coordinates of the center of the target vehicle,
Figure FDA0002716989510000033
wherein sigmaxAnd σyIs defined by the formula:
σx=|vx|+k1L
σy=|vy|+k2W
wherein v isx,vyIs the transverse and longitudinal speed of the vehicle, k1And k is2Is a compensation factor.
7. The neural network model-based vehicle travel risk prediction method of claim 1 or 6, wherein the local speed difference is calculated as follows:
Figure FDA0002716989510000034
8. the neural network model-based vehicle travel risk prediction method according to claim 1 or 7, characterized in that the local acceleration difference is calculated as follows:
Figure FDA0002716989510000035
CN202011076552.2A 2020-10-10 2020-10-10 Vehicle running risk prediction method based on neural network model Active CN112215487B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011076552.2A CN112215487B (en) 2020-10-10 2020-10-10 Vehicle running risk prediction method based on neural network model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011076552.2A CN112215487B (en) 2020-10-10 2020-10-10 Vehicle running risk prediction method based on neural network model

Publications (2)

Publication Number Publication Date
CN112215487A true CN112215487A (en) 2021-01-12
CN112215487B CN112215487B (en) 2023-05-23

Family

ID=74052983

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011076552.2A Active CN112215487B (en) 2020-10-10 2020-10-10 Vehicle running risk prediction method based on neural network model

Country Status (1)

Country Link
CN (1) CN112215487B (en)

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112849156A (en) * 2021-04-25 2021-05-28 北京三快在线科技有限公司 Driving risk identification method and device
CN113033902A (en) * 2021-03-31 2021-06-25 中汽院智能网联科技有限公司 Automatic driving track-changing planning method based on improved deep learning
CN113222609A (en) * 2021-05-07 2021-08-06 支付宝(杭州)信息技术有限公司 Risk identification method and device
CN113326825A (en) * 2021-08-03 2021-08-31 新石器慧通(北京)科技有限公司 Pseudo tag generation method and device, electronic equipment and storage medium
CN113326826A (en) * 2021-08-03 2021-08-31 新石器慧通(北京)科技有限公司 Network model training method and device, electronic equipment and storage medium
CN113436432A (en) * 2021-05-24 2021-09-24 武汉理工大学 Method for predicting short-term traffic risk of road section by using road side observation data
CN113919633A (en) * 2021-09-06 2022-01-11 同济大学 Risk prediction method for running state of hazardous chemical substance transport vehicle
CN114333416A (en) * 2021-12-24 2022-04-12 阿波罗智能技术(北京)有限公司 Vehicle risk early warning method and device based on neural network and automatic driving vehicle
CN114397474A (en) * 2022-01-17 2022-04-26 吉林大学 Wind parameter measurement method of arc ultrasonic sensor array based on FCN-MLP
CN114820216A (en) * 2022-05-09 2022-07-29 吉林大学 Driving risk quantification method, classification method and prediction method
CN114897295A (en) * 2022-03-30 2022-08-12 中国第一汽车股份有限公司 Model training and driving behavior evaluation method, device, equipment and storage medium
CN115082905A (en) * 2022-06-07 2022-09-20 青岛海尔科技有限公司 Method and device for detecting operation behavior, storage medium and electronic device
CN114820216B (en) * 2022-05-09 2025-07-01 吉林大学 A driving risk quantification method, classification method and prediction method

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107215335A (en) * 2017-06-01 2017-09-29 北京交通大学 Traffic safety risk feedback early warning system and method for early warning based on microcosmic driving
CN109094482A (en) * 2018-07-25 2018-12-28 哈尔滨工业大学 Vehicle operation risk assessment information acquisition system, risk evaluating system and method
US20190114544A1 (en) * 2017-10-16 2019-04-18 Illumina, Inc. Semi-Supervised Learning for Training an Ensemble of Deep Convolutional Neural Networks
CN109933619A (en) * 2019-03-13 2019-06-25 西南交通大学 A semi-supervised classification prediction method
CN109952583A (en) * 2016-11-15 2019-06-28 谷歌有限责任公司 The semi-supervised training of neural network
CN110610271A (en) * 2019-09-17 2019-12-24 北京理工大学 A Multiple Vehicle Trajectory Prediction Method Based on Long Short-Term Memory Network
CN111368879A (en) * 2020-02-17 2020-07-03 南京邮电大学 Track data mining method based on deep semi-supervised neural network

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109952583A (en) * 2016-11-15 2019-06-28 谷歌有限责任公司 The semi-supervised training of neural network
CN107215335A (en) * 2017-06-01 2017-09-29 北京交通大学 Traffic safety risk feedback early warning system and method for early warning based on microcosmic driving
US20190114544A1 (en) * 2017-10-16 2019-04-18 Illumina, Inc. Semi-Supervised Learning for Training an Ensemble of Deep Convolutional Neural Networks
CN109094482A (en) * 2018-07-25 2018-12-28 哈尔滨工业大学 Vehicle operation risk assessment information acquisition system, risk evaluating system and method
CN109933619A (en) * 2019-03-13 2019-06-25 西南交通大学 A semi-supervised classification prediction method
CN110610271A (en) * 2019-09-17 2019-12-24 北京理工大学 A Multiple Vehicle Trajectory Prediction Method Based on Long Short-Term Memory Network
CN111368879A (en) * 2020-02-17 2020-07-03 南京邮电大学 Track data mining method based on deep semi-supervised neural network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DONG-HYUN LEE: "Pseudo-Label : The Simple and E cient Semi-Supervised Learning Method for Deep Neural Networks", 《ICML 2013 WORKSHOP : CHALLENGES IN REPRESENTATION LEARNING(WREPL)》 *
HONGYU HU等: "Cost-sensitive semi-supervised deep learning to assess driving risk by application of naturalistic vehicle trajectories", 《EXPERT SYSTEMS WITH APPLICATIONS》 *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113033902B (en) * 2021-03-31 2024-03-19 中汽院智能网联科技有限公司 Automatic driving lane change track planning method based on improved deep learning
CN113033902A (en) * 2021-03-31 2021-06-25 中汽院智能网联科技有限公司 Automatic driving track-changing planning method based on improved deep learning
CN112849156B (en) * 2021-04-25 2021-07-30 北京三快在线科技有限公司 Driving risk identification method and device
CN112849156A (en) * 2021-04-25 2021-05-28 北京三快在线科技有限公司 Driving risk identification method and device
CN113222609A (en) * 2021-05-07 2021-08-06 支付宝(杭州)信息技术有限公司 Risk identification method and device
CN113436432A (en) * 2021-05-24 2021-09-24 武汉理工大学 Method for predicting short-term traffic risk of road section by using road side observation data
CN113326825A (en) * 2021-08-03 2021-08-31 新石器慧通(北京)科技有限公司 Pseudo tag generation method and device, electronic equipment and storage medium
CN113326826A (en) * 2021-08-03 2021-08-31 新石器慧通(北京)科技有限公司 Network model training method and device, electronic equipment and storage medium
CN113919633A (en) * 2021-09-06 2022-01-11 同济大学 Risk prediction method for running state of hazardous chemical substance transport vehicle
CN114333416A (en) * 2021-12-24 2022-04-12 阿波罗智能技术(北京)有限公司 Vehicle risk early warning method and device based on neural network and automatic driving vehicle
CN114397474B (en) * 2022-01-17 2022-11-08 吉林大学 FCN-MLP-based arc ultrasonic sensing array wind parameter measurement method
CN114397474A (en) * 2022-01-17 2022-04-26 吉林大学 Wind parameter measurement method of arc ultrasonic sensor array based on FCN-MLP
CN114897295A (en) * 2022-03-30 2022-08-12 中国第一汽车股份有限公司 Model training and driving behavior evaluation method, device, equipment and storage medium
CN114820216A (en) * 2022-05-09 2022-07-29 吉林大学 Driving risk quantification method, classification method and prediction method
CN114820216B (en) * 2022-05-09 2025-07-01 吉林大学 A driving risk quantification method, classification method and prediction method
CN115082905A (en) * 2022-06-07 2022-09-20 青岛海尔科技有限公司 Method and device for detecting operation behavior, storage medium and electronic device

Also Published As

Publication number Publication date
CN112215487B (en) 2023-05-23

Similar Documents

Publication Publication Date Title
CN112215487B (en) Vehicle running risk prediction method based on neural network model
CN111104969B (en) A method for predicting the possibility of collision between an unmanned vehicle and surrounding vehicles
CN110084151B (en) Video abnormal behavior discrimination method based on non-local network deep learning
CN113486822B (en) Surrounding vehicle track prediction method and system based on driving intention
Li et al. Short-term vehicle speed prediction based on BiLSTM-GRU model considering driver heterogeneity
Hu et al. Efficient mapping of crash risk at intersections with connected vehicle data and deep learning models
CN110949398A (en) Method for detecting abnormal driving behavior of first-vehicle drivers in vehicle formation driving
CN111738337B (en) Driver distraction state detection and identification method in mixed traffic environment
CN108091132B (en) A kind of traffic flow prediction method and device
CN109345832B (en) An urban road overtaking prediction method based on deep recurrent neural network
CN109455178B (en) An active control system and method for road traffic vehicle driving based on binocular vision
Xue et al. A context-aware framework for risky driving behavior evaluation based on trajectory data
CN111907523A (en) Vehicle following optimization control method based on fuzzy reasoning
CN118306403B (en) Vehicle track prediction and behavior decision method and system considering driving style
CN110619340B (en) A method for generating lane changing rules for autonomous vehicles
CN112308136B (en) Driving distraction detection method based on SVM-Adaboost
Sun et al. Vision-based traffic conflict detection using trajectory learning and prediction
CN113642114B (en) An error-prone anthropomorphic random car-following driving behavior modeling method
CN116946183A (en) Commercial vehicle driving behavior prediction method considering driving capability and vehicle equipment
CN115981302A (en) Vehicle following lane change behavior decision-making method and device and electronic equipment
CN114818981A (en) An imitation learning method for driving behavior
CN115186583A (en) A decision-making method for car-following behavior based on historical time series features
Ma et al. Lane change analysis and prediction using mean impact value method and logistic regression model
CN118553108B (en) A trajectory prediction method considering dynamic interactions between vehicles
CN119417671B (en) Intelligent driving scene self-adaptive teaching method and system based on reinforcement learning

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