基于逐步回归分析和BP_Adaboost算法的危险驾驶行为辨识
|
摘要
为准确辨识车辆在行驶过程中可能出现的异常加减速,压线行驶,右侧超车驾驶行为,以便于及时给予驾驶员信息反馈和安全预警,使车辆保持安全的运行状态.首先通过虚拟驾驶仿真实验平台,采集驾驶行为的48种车辆运行数据对实验数据进行预处理,获得实验样本1492组;其次利用逐步回归分析对原始数据进行降维处理,并选取其中的最优回归模型获得特征参数;将提取的特征参数数据输入到BP_Adaboost多分类网络中,训练BP_Adaboost多分类网络,对上述驾驶行为进行识别;最后该模型与BP神经网络进行识别结果对比分析.结果表明模型识别率相较于BP神经网络提高了8.81%,达到92.93%,能进行更加有效的安全预警.
In order to recognize abnormal acceleration and deceleration,line pressing,overtaking on the right side and give the driver Information feedback and safety early warning in time and keep the vehicle in a safe running state.Firstly,48 kinds of drivers driving behavior data and 1492 sets of experimental samples were gathered by driving simulation experiments,the experimental data were pretreated.Secondly,using stepwise regression analysis to reduce the dimension of the original data and selecting the optimal regression model to obtain the characteristic parameters.The the characteristic parameters were input in a BP_Adaboost multi-classification identification model.BP-Adaboost multi-classification identification network was trained,and an identification model was built for driving behavior.Finally,The model is compared with stepwise regression analysis and BP neural network combination model.The results show that the recognition rate of this model is 8.81% higher than that of BP neural network,reaching 92.93% and the model is more effective in early safely warning.
In order to recognize abnormal acceleration and deceleration,line pressing,overtaking on the right side and give the driver Information feedback and safety early warning in time and keep the vehicle in a safe running state.Firstly,48 kinds of drivers driving behavior data and 1492 sets of experimental samples were gathered by driving simulation experiments,the experimental data were pretreated.Secondly,using stepwise regression analysis to reduce the dimension of the original data and selecting the optimal regression model to obtain the characteristic parameters.The the characteristic parameters were input in a BP_Adaboost multi-classification identification model.BP-Adaboost multi-classification identification network was trained,and an identification model was built for driving behavior.Finally,The model is compared with stepwise regression analysis and BP neural network combination model.The results show that the recognition rate of this model is 8.81% higher than that of BP neural network,reaching 92.93% and the model is more effective in early safely warning.
引文
[1]WU Bingfei, CHEN Yinghan,YEH C H,et al.Reasoning-based framework for driving safety monitoring using driving event recognition[J].IEEE Transactions on Intelligent Transportation Systems,2013,14(3):1231-1241.
[2]JEONGE O H C,KIM I.Detection of lateral hazardous driving events using in-vehicle gyro sensor data[J].KSCE Journal of Civil Engineering,2013,17(6):1471-1479.
[3]肖献强,王其东.基于隐马尔可夫模型的驾驶行为预测方法研究[J].中国机械工程2013,24(.21):2972-2976.
[4]髙印寒,唐荣江,梁杰等.汽车声品质的GA-BP神经网络预测与权重分析[J].光学精密工程2013,21(2):462-468.
[5]张敬磊,王晓原·基于非线性组合模型的交通流预测方法[J].计算机工程,2010,36(5):202-204.
[6]LI X,WANG L,SUNG E.Ada Boost with SVM-based component classifiers[J].Engineering Applications of Artificial Intelligence,2008,21(5):785-795.
[7]王小川,史峰,郁磊,李洋.MATLAB神经网络43个案例分析[M].北京:北京航空航天大学出版社,2013:42-43.
[8]Freund Y,Schapire R E.Experiments with a new boosting algorithm[C]//Proceedings of the 13~(th)International Conference on Machine Learning,Bari,Italy,july 3-6,1996:148-156.
[9]吕雁飞,侯子骄,张凯.多分类BP_Adaboost算法研究与应用[J].高技术通讯,2015,25(5):437-444.
[2]JEONGE O H C,KIM I.Detection of lateral hazardous driving events using in-vehicle gyro sensor data[J].KSCE Journal of Civil Engineering,2013,17(6):1471-1479.
[3]肖献强,王其东.基于隐马尔可夫模型的驾驶行为预测方法研究[J].中国机械工程2013,24(.21):2972-2976.
[4]髙印寒,唐荣江,梁杰等.汽车声品质的GA-BP神经网络预测与权重分析[J].光学精密工程2013,21(2):462-468.
[5]张敬磊,王晓原·基于非线性组合模型的交通流预测方法[J].计算机工程,2010,36(5):202-204.
[6]LI X,WANG L,SUNG E.Ada Boost with SVM-based component classifiers[J].Engineering Applications of Artificial Intelligence,2008,21(5):785-795.
[7]王小川,史峰,郁磊,李洋.MATLAB神经网络43个案例分析[M].北京:北京航空航天大学出版社,2013:42-43.
[8]Freund Y,Schapire R E.Experiments with a new boosting algorithm[C]//Proceedings of the 13~(th)International Conference on Machine Learning,Bari,Italy,july 3-6,1996:148-156.
[9]吕雁飞,侯子骄,张凯.多分类BP_Adaboost算法研究与应用[J].高技术通讯,2015,25(5):437-444.