[1]刘 伟,刘浩学,朱 彤.基于ROC 曲线优化的车辆行驶状态估计BP模型[J].西安建筑科技大学学报:自然科学版,2014,(04):593-597.[doi:10.15986/j.1006-7930.2014.04.025]
 LIU Wei,LIU Haoxue,ZHU Tong.BP neural network for vehicle state estimation was based on ROC curve optimization[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2014,(04):593-597.[doi:10.15986/j.1006-7930.2014.04.025]
点击复制

基于ROC 曲线优化的车辆行驶状态估计BP模型()
分享到:

西安建筑科技大学学报:自然科学版[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2014年04期
页码:
593-597
栏目:
出版日期:
2014-08-31

文章信息/Info

Title:
BP neural network for vehicle state estimation was based on ROC curve optimization
文章编号:
1006-7930(2014)04-0593-05
作者:
刘 伟刘浩学朱 彤
(长安大学汽车学院, 陕西 西安 710064)
Author(s):
LIU Wei LIU Haoxue ZHU Tong
(School of Automobile, Chang’an University, Xi’an 710064, China)
关键词:
交通工程车辆行驶状态估计ROC 曲线优化
Keywords:
traffic engineering vehicle state estimation BP neural network ROC curve optimization
分类号:
U491;TP206
DOI:
10.15986/j.1006-7930.2014.04.025
文献标志码:
A
摘要:
针对现有行车状态估计器难以适应复杂非线性模型,结合 BP神经网络在解决非线性系统方面表现出优良的性能,采 用ROC曲线(受试者特征工作曲线)对BP神经网络算法进行优化,依据各个节点权重值的变化情况绘制学习机器相应的ROC 曲线,将ROC 曲线下方面积作为各个节点权重值选取的唯一准则,每次在同一节点进行变步长的搜索(大步长和小步长), 并根据不同步长的搜索结果确定下一次步长的大小,以确定最佳的权重值,最后以波动性较强的车辆横摆角速度作为样例对算法进行验证.研究结果表明:通过ROC 对其性能的评价,加速了BP 网络的收敛速度,在一定程度上避免了出现局部最小值的情况,提高了模型的容错能力;优化后的模型在5%误判率的情况下有较高的击中概率,表现出更强的泛化能力,适应性更强.
Abstract:
In view of the inadequate adaptability of the existing traffic estimator to the complex nonlinear models, the receiver operating characteristic (ROC) curves were adopted to evaluate the algorithm of BP neural network which demonstrates excellent performance in solving problems with nonlinear systems. According to each node weight value, the corresponding ROC curve of the learning machine is drawn. The area of ROC curve is the only criterion for the selection of each node weights. The method is used for the searching of the minimum value through variable steps, i.e. maximum and small, at the same nodes. According to the results, the size of the next step for determining the best weights can be fixed. Finally, the algorithm verification can be carried out on the basis of the strong volatility data of vehicle yaw-rate. The results show that the evaluation of the performance of ROC may accelerate the convergence speed of the BP network to a certain extent which making it possible to avoid the local minimum. Furthermore, the fault tolerance model can be improved and the optimal model error probability of the hit rate goes higher in 5% of the cases. The optimized model showed a strong generalization capacity as well as the fine adaptability.

参考文献/References:

[1] 宗长富, 胡丹, 杨肖 , 等.基于扩展Kalman滤波的汽车行驶状态估计[J].吉林大学学报: 工学版, 2009,39(1):7-11.

ZONG Changfu, HU Dan,Yang Xiao,et al. Vehicle Driving State Estimation Based on Extended Kalman Filter[J]. Journal of Jilin University: Engineering and Technology Edition, 2009,39(1):7-11.

[2] 高振海, 郑南宁, 程洪. 基于车辆动力学和Kalman滤波的汽车状态软测量[J].系统仿真报, 2004, 16(1): 22-24.

GAO Zhenhai, ZHENG Nanning, CHENG Hong. Soft sensor of vehicle state based on vehicle dynamics and Kalman filter[J]. Journal of System Simulation,2004,16(1):22-24.

[3] SUN Yujuan, WANG Yilei , LI Tao, et al. A new BP neural network model based on the random fuzzy theory[J].Fuzzy Systems and Knowledge Discovery,2007,8(1): 42-45.

[4] ZHAO Jiali, LUO Siwei, HAN Zhen. The Improvements of BP neural network learning algorithm[C]//Proceedings of ICSP, 2000:1647-1649.

[5] 张旭, 徐玉秀, 刘恩东.基于神经网络优化法的故障诊断应用研究[J].沈阳工业大学学报, 2004,26(3):313-315.

ZHANG Xu, XU Yuxiu, LIU Endong. Comparison between powell algorithm and BP algorithm in training neural network[J]. Journal of Shenyang University of Technology, 2004,26(3):313-315.

[6] 李帧, 徐凌宇.BP网络微观与宏观弊病的研究[J].计算机工程与设计, 2006, 27(20): 3779-3782.

LI Zhen, XU Lingyu. Research on microcosmic and macroscopical defects in BP-NET[J].Computer Engineering and Design, 2006,27(20):3779-3782.

[7] 李祚泳, 汪嘉杨, 郭淳. PSO 算法优化BP 网络的新方法及仿真实验[J].电子学报,2008, 36(11): 2224-2228.

LI Zuoyong, Wang Jiayang, Guo Chun. A new method of BP network optimized based on particle swarm optimization and simulation Test[J]. Acta Electronica Sinica, 2008,36(11): 2224-2228.

[8] 张宝金, 胡青苗, 曾梅光, 等.提高BP网络性能的一种算法[J].东北大学学报: 自然科学版,1999,20(1): 108-110.

ZHANG Baojin, HU Qingmiao, ZENG Meiguang, et al. Method of improving the properties of BP artificial neural network[J]. Journal of Northeastern University: Natural Science, 1999,20(1):108-110.

[9] 石昊苏.基于实例与MATLAB的ROC曲线绘制比较研究[J].电子设计与工程,2010,18(9):36-39.

SHI Haosu. Comparative research of the ROC curve drawing based on case and MATLAB[J]. Electronic Design Engineering, 2010,18(9):36-39.

[10] 邹洪侠, 秦锋, 程泽凯, 等.二分类器的 ROC 曲线生成算法[J].计算机技术与发展, 2009,19(6):109-112.

ZOU Hongxia, QIN Feng, CHENG Zekai, et al. Algorithm for generating ROC curve of two-classifier[J]. Computer Technology and Development, 2009,19(6):109-112.

[11] 罗成汉.基于 MATLAB 神经网络工具箱的 BP 网络实现[J].计算机仿真, 2004,21(5):109-115.

LUO Chenghan. Realization of BP network based on neural network tool Kit in MATLAB[J]. Computer Simulation, 2004,21(5):109-115.

[12] 孙帆, 施学勤.基于 MATLAB 的 BP 神经网络设计[J].计算机与数字工程, 2007, 35(8): 124-126.

SUN Fan, SHI Xueqin. Design of BP neural network based on MATLAB[J]. Computer & Digital Engineering, 2007,35(8):124-126.

相似文献/References:

[1]李晓伟,陈 红,周继彪,等.基于累积前景理论的城市公交线网灰关联优化[J].西安建筑科技大学学报:自然科学版,2012,(04):495.[doi:10.15986/j.1006-7930.2012.04.008]
 LI Xiao-wei,CHEN Hong,ZHOU Ji-biao,et al.Grey relevancy optimization model of urban public transportation network based on cumulative prospect theory[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2012,(04):495.[doi:10.15986/j.1006-7930.2012.04.008]
[2]李晓伟,陈 红,王肇飞,等.基于离差最大化的公路网灰关联投影评价模型[J].西安建筑科技大学学报:自然科学版,2012,(05):679.[doi:10.15986/j.1006-7930.2012.05.012]
 LI Xiao-wei,CHEN Hong,WANG Zhao-fei,et al.Grey relation projection evaluation model of highway network planning schemes based on maximizing deviation method[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2012,(04):679.[doi:10.15986/j.1006-7930.2012.05.012]
[3]王秋平,杨 茜,孙 皓.基于空间句法的西安市历史街区交通改善研究[J].西安建筑科技大学学报:自然科学版,2015,(04):487.[doi:10.15986/j.1006-7930.2015.04.005]
 WANG Qiuping,YANG Xi,Sun Hao.Study on improving traffic conditions for the historic district of Xi’an based on space syntax[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2015,(04):487.[doi:10.15986/j.1006-7930.2015.04.005]
[4]李晓伟1,2,陈 红1,等.基于AHP-熵复合物元的城市交通可持续发展评价[J].西安建筑科技大学学报:自然科学版,2011,(06):831.[doi:DOI:10.15986/j.1006-7930.2011.06.011]
 ,,et al.Evaluation model of urban traffic sustainable developmentbased on matter element with AHP and entrop[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2011,(04):831.[doi:DOI:10.15986/j.1006-7930.2011.06.011]
[5]李聪颖,等.雾霾天气对出行行为的影响机理研究[J].西安建筑科技大学学报:自然科学版,2015,(05):728.[doi:DOI:10.15986/j.1006-7930.2015.05.021]
 LI Congying,HUANG Yizhe,LI Gan,et al.Analysis of the influence of smog on the travel behavior[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2015,(04):728.[doi:DOI:10.15986/j.1006-7930.2015.05.021]
[6]王秋平,张 译,孙 皓.历史街区自行车交通特性研究[J].西安建筑科技大学学报:自然科学版,2016,(02):227.[doi:10.15986/j.1006-7930.2016.02.013]
 WANG Qiuping,ZHANG Yi,SUN Hao.Research on the historical block bicycle traffic characteristics[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2016,(04):227.[doi:10.15986/j.1006-7930.2016.02.013]
[7]李晓伟,王 炜,杨 敏,等.多模式综合交通客运方式选择行为差异性—基于强制与休闲型活动出行的对比分析[J].西安建筑科技大学学报:自然科学版,2016,(06):868.[doi:10.15986/j.1006-7930.2016.06.015]
 LI Xiaowei,WANG Wei,YANG Min,et al.Disparity of choice behavior for multi-mode passenger transport system—A comparative analysis of the compulsory and leisure activities travel[J].J.Xi’an Univ. of Arch. & Tech.:Natural Science Edition,2016,(04):868.[doi:10.15986/j.1006-7930.2016.06.015]

备注/Memo

备注/Memo:
基金项目:国家自然科学基金青年基金项目(51108036);中央高校基本科研业务费专项资金项目(CHD2011JC180)
收稿日期:2013-09-20 修改稿日期:2014-07-27
作者简介:刘伟(1979-),男,讲师,博士,主要从事交通安全与交通规划方面的研究.E-mail:liuwei@chd.edu.cn
更新日期/Last Update: 2015-10-06