基于NN的无信号混合交通交叉口自行车冲突避让行为建模仿真  

Modeling and Simulation of Bicycle Accident Avoidance at Mixed Non-Signalized Traffic Intersections Based on NN

作　　者：田雨[1] 黄玲[2] 陈罡[2] 

机构地区：[1]沈阳体育学院体育信息技术系,辽宁沈阳110102 [2]华南理工大学土木与交通学院,广东广州510640

出　　处：《广东工业大学学报》2015年第1期103-108,共6页Journal of Guangdong University of Technology

基　　金：国家自然科学基金资助项目(51408237);广东省重大科技专项资助项目(2012A010800007)

摘　　要：在我国城市无信号交叉口,自行车与混合交通流的其他车辆(如机动车、自行车、行人)的冲突避让行为,难以通过传统数学方法建立准确的仿真模型.本文在以往交叉口自行车微观模型基础上,采用冲突车辆与自行车的相对位置矢量、相对速度矢量以及期望速度与实际速度相对矢量作为输入,自行车二维空间加速度矢量作为双输出的神经网络建立数学模型.利用实地采集自行车-机动车冲突、自行车-自行车冲突和自行车-行人冲突样本数据,进行神经网络训练和仿真验证,并对骑车者性别、冲突车辆类型和冲突速度进行了模型灵敏度分析.结果表明,模型能较好地反映无信号交叉口自行车与其他车辆及障碍物的冲突避让行为.The modeling and simulation of bicycle accident avoidance with other vehicles（ motorcar ／ other bicycles ／ pedestrians） at non-signalized intersections is of great importance. However,it is very difficult to simulate the accident avoidance of individual bicycle because of the great variation,characteristics and traffic environment in cycling. The current study proposes a back propagation neural network（ BPNN）approach to simulate the bicycle accident avoidance at non-signalized intersections. A computer-based four-layered（ NN） model was developed for bicycle accident avoidance modeling. The NN model was trained,validated with field data and then compared with other models. It was found that the NN model performed well. Results showed that the NN model could produce reasonable estimates for individual bicycle at non-signalized intersections.

关 键 词：自行车 冲突避让行为 神经网络 无信号交叉口 

分 类 号：U491[交通运输工程—交通运输规划与管理]