基于PCA与自适应BP算法的侵犯性驾驶行为识别  

Recognition of the Aggressive Driving Based on PCA and Adaptive BP Algorithm

作　　者：王丝丝[1] 张敬磊[1] 王晓原[1] 孙一帆[1] 陈慈 马春杰 WANG Si-si;ZHANG Jing-lei;WANG Xiao-yuan;SUN Yi-fan;CHEN Ci;MA Chun-jie(School of Transportation and Vehicle Engineering,Shandong University of Technology,Zibo 255049,China;College of Metropolitan Transportation,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]山东理工大学交通与车辆工程学院,山东淄博255049 [2]北京工业大学城市交通学院,北京100124

出　　处：《数学的实践与认识》2018年第13期163-170,共8页Mathematics in Practice and Theory

基　　金：山东省自然科学基金(ZR2017LF015);国家自然科学基金(61573009);山东省高等学校科技计划(J15LB07)

摘　　要：从机动车安全预警的角度，以有效识别侵犯性驾驶行为为目的，针对传统BP算法学习效率低、收敛速度慢等缺点，提出一种基于PCA与自适应学习速率的BP网络改进识别算法．首先借助人因系统及汽车驾驶平台进行仿真实验，获取驾驶人生理-心理及车辆运行数据集，然后利用主成分分析提取其特征指标，继而应用自适应学习速率BP网络改进算法对驾驶行为进行识别．结果表明：驾驶员呼吸、肌电、速度、油门、车道线偏距以及发动机转速受驾驶行为的影响较大；识别精度为96．17％，对比自适应学习速率BP网络算法、BP网络算法，算法能明显减小训练迭代次数、提高识别精度．In order to recognize the aggressive driving behaviors effectively for Automotive Active Safety, a PCA and improved BP network algorithm is proposed based on adaptive learning rate overcoming the former＇s low learning efficiency and slow convergence. Firstly, drivers＇ psychophysical and vehicle operational data, which is collected in automatic driving simulation experiments were extracted feature using PCA. And then driving behaviors are recognized with proposed algorithm. Results show that Respiratory, Electromyography, Speed, Accelerator and Lane offset are larger affected by driving behaviors. And the accuracy of identification can reach 96.17%. Compared with PCA and BP algorithm, BP algorithm, the proposed algorithm can obviously reduce the number of iterations and improve the recognition accuracy of training.

关 键 词：侵犯性驾驶 行为识别 自适应BP算法 人因系统 主动安全预警系统 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] U463.6[自动化与计算机技术—控制科学与工程] U491.254[机械工程—车辆工程]