[1]赵玮,徐良杰,冉斌,等.基于深度学习DBN算法的高速公路危险变道判别模型[J].东南大学学报(自然科学版),2017,47(4):832-838.[doi:10.3969/j.issn.1001-0505.2017.04.031]
 Zhao Wei,Xu Liangjie,Ran Bin,et al.Dangerous lane-change detecting model on highway based on deep learning DBN algorithm[J].Journal of Southeast University (Natural Science Edition),2017,47(4):832-838.[doi:10.3969/j.issn.1001-0505.2017.04.031]
点击复制

基于深度学习DBN算法的高速公路危险变道判别模型()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
47
期数:
2017年第4期
页码:
832-838
栏目:
交通运输工程
出版日期:
2017-07-20

文章信息/Info

Title:
Dangerous lane-change detecting model on highway based on deep learning DBN algorithm
作者:
赵玮12徐良杰1冉斌3汪济洲1
1武汉理工大学交通学院, 武汉 430063; 2内蒙古科技大学经管学院, 包头 014010; 3东南大学交通学院, 南京 210096
Author(s):
Zhao Wei12 Xu Liangjie1 Ran Bin3 Wang Jizhou1
1School of Transportation, Wuhan University of Technology, Wuhan 430063, China
2School of Economics and Management, Inner Mongolia University of Science and Technology, Baotou 014010, China
3School of Transportation, Southeast University, Nanjing 210096, China
关键词:
危险变道判别 模拟车试验 智能交通 深度信任网络 自动驾驶 车联网
Keywords:
dangerous lane-changing discriminant vehicle simulation experiments intelligent transportation deep belief network automatic driving connected vehicle
分类号:
U491.2
DOI:
10.3969/j.issn.1001-0505.2017.04.031
摘要:
针对危险变道过程影响交通安全这一问题,提出一种基于深度学习DBN(deep belief networks)算法与分类分析方法的新型危险变道量化判别模型,以解决现存车辆变道过程不可被量化分析及准确判别的问题.招募28名被试者,利用模拟器仿真平台开展实际场景下试验,获取详细的车辆行驶数据及驾驶环境数据作为训练的模型输入.采用SVM算法作为输出层的分类器,建立了DBN-SVM判别模型及基于样本下模型的一般训练方法.该模型的识别精度为93.78%,较朴素贝叶斯模型和BP-ANN神经网络分别提高20.11%和14.45%,并且调整参数后判别结果稳定.DBN-SVM模型可以根据驾驶员历史驾驶数据对即将发生的危险变道做出预测及判别,对驾驶员提出警告,从而减少交通事故的发生.此外,该研究为车联网环境下变道判别及警示的研究提供了理论支持.
Abstract:
Aiming at the problem that the vehicle lane-changing process cannot be quantitatively analyzed and accurately discriminated, a new quantitative discriminant model based on the DBN(deep belief networks)algorithm and the classification analysis method is presented. Twenty-eight participants were recruited. The participators took part in real-scene simulation experiments using the simulation driving platform. The detailed data of vehicle traveling and driving environment was required and used as the input of the model. With the SVM(support vector machine)algorithm as the classifier of the output layer, the discriminant model DBN-SVM and corresponding training method are set up. The discriminant accuracy of the DBN-SVM is 93.78%, increasing by 20.11% and 14.45% compared with the Naïve Bayes model and BP-ANN(back propagation-artificial neural networks), respectively. And, the results are stable with adjusted parameters. The DBN-SVM model can predict and discriminate coming dangerous lane-changing according to drivers’ driving history data, and warn drivers. As a result, it can reduce the chance of traffic accidents. This study provides theoretical support for lane-changing discrimination and warning under the connected vehicle environment.

参考文献/References:

[1] Jewett A, Shults R A, Banerjee T, et al. Alcohol-impaired driving among adults—United States, 2012[J]. MMWR Morbidity & Mortality Weekly Report, 2015, 64(30): 814-817. DOI:10.15585/mmwr.mm6430a2.
[2] Dharmaratne S D, Jayatilleke A U, Jayatilleke A C. Road traffic crashes, injury and fatality trends in Sri Lanka: 1938—2013[J]. Bulletin of the World Health Organization, 2015, 93(9): 640-647. DOI:10.2471/BLT.14.150193.
[3] World Health Organization. Global status report on road safety-2013: Supporting a decade of action[M]. World Health Organization, 2013: 286.
[4] Hidas P. Modelling lane changing and merging in microscopic traffic simulation[J]. Transportation Research Part C: Emerging Technologies, 2002, 10(5): 351-371. DOI:10.1016/s0968-090x(02)00026-8.
[5] Cassidy M J, Rudjanakanoknad J. Increasing the capacity of an isolated merge by metering its on-ramp[J]. Transportation Research Part B: Methodological, 2005(10): 896-913. DOI:10.1016/j.trb.2004.12.001.
[6] Zheng Z, Ahn S, Chen D, et al. Freeway traffic oscillations: Microscopic analysis of formations and propagations using wavelet transform[J]. Transportation Research Part B: Methodological, 2011, 45(9): 1378-1388.
[7] Laval J A, Daganzo C F. Lane-changing in traffic streams[J]. Transportation Research Part B: Methodological, 2006, 40(3): 251-264. DOI:10.1016/j.trb.2005.04.003.
[8] Zheng Z. Recent developments and research needs in modeling lane changing[J]. Transportation Research Part B: Methodological, 2014, 60: 16-32. DOI:10.1016/j.trb.2013.11.009.
[9] Moridpour S, Sarvi M, Rose G, et al. Lane-changing decision model for heavy vehicle drivers[J]. Journal of Intelligent Transportation Systems, 2012, 16(1): 24-35. DOI:10.1080/15472450.2012.639640.
[10] Chen S W, Fang C Y, Tien C T. Driving behaviour modelling system based on graph construction[J]. Transportation Research Part C: Emerging Technologies, 2013, 26: 314-330. DOI:10.1016/j.trc.2012.10.004.
[11] Dahl G E, Yu D, Deng L, et al. Context-dependent pre-trained deep neural networks for large-vocabulary speech recognition[J]. IEEE Transactions on Audio Speech & Language Processing, 2012, 20(1): 30-42.
[12] Zhang Y, Zhang E, Chen W. Deep neural network for halftone image classification based on sparse auto-encoder[J]. Engineering Applications of Artificial Intelligence, 2016, 50: 245-255. DOI:10.1016/j.engappai.2016.01.032.
[13] Hinton G E, Osindero S, Teh Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(7): 1527-1554. DOI:10.1162/neco.2006.18.7.1527.
[14] Fischer A, Igel C. Training restricted Boltzmann machines: An introduction[J]. Pattern Recognition, 2014, 47(1): 25-39. DOI:10.1016/j.patcog.2013.05.025.
[15] Kollar T, Tellex S, Roy D, et al. Toward understanding natural language directions[C]//2010 5th ACM/IEEE International Conference on Human-Robot Interaction(HRI). Osaka, Japan,2010. DOI:10.1109/hri.2010.5453186.
[16] Mendoza M Á, Pérez de la Blanca N, Marín-Jiménez M J. Fitting product of HMM to human motions[J]. Computer Analysis of Images and Patterns, 2009, 5702: 824-831. DOI:10.1007/978-3-642-03767-2_100.
[17] Yu R, Abdel-Aty M. Utilizing support vector machine in real-time crash risk evaluation[J]. Accident Analysis & Prevention, 2013, 51: 252-259. DOI:10.1016/j.aap.2012.11.027.
[18] Bengio Y. Learning deep architectures for AI[J]. Foundations and Trends? in Machine Learning, 2009, 2(1): 1-127. DOI:10.1561/2200000006.
[19] Chen J, Huang H, Tian S, et al. Feature selection for text classification with Naïve Bayes[J]. Expert Systems with Applications, 2009, 36(3): 5432-5435. DOI:10.1016/j.eswa.2008.06.054.
[20] Saravanan N, Ramachandran K I. Incipient gear box fault diagnosis using discrete wavelet transform(DWT)for feature extraction and classification using artificial neural network(ANN)[J]. Expert Systems with Applications, 2010, 37(6): 4168-4181. DOI:10.1016/j.eswa.2009.11.006.

备注/Memo

备注/Memo:
收稿日期: 2016-11-06.
作者简介: 赵玮(1988—),男,博士生,讲师;徐良杰(联系人),女,博士,教授,博士生导师,laurrie119@163.com.
基金项目: 教育部社科青年基金资助项目(16YJCZH157)、国家重点基础研究发展计划(973计划)资助项目(2012CB725405)、内蒙古科技大学创新基金资助项目(2015QDL27).
引用本文: 赵玮,徐良杰,冉斌,等.基于深度学习DBN算法的高速公路危险变道判别模型[J].东南大学学报(自然科学版),2017,47(4):832-838. DOI:10.3969/j.issn.1001-0505.2017.04.031.
更新日期/Last Update: 2017-07-20