论文部分内容阅读
为了提高机动车闯红灯警示系统判别准确率,提出适用于该系统的机动车闯红灯判别模型并进行评价。首先综述闯红灯警示系统的产生与发展现状,根据系统服务需求与工作流程,提出使用Logistic回归与神经网络方法建立判别模型预测机动车停止-通过行为,通过城市交通实测数据对模型进行评价。数据采集自上海城市道路交叉口进口道,通过Autoscope视频检测系统提取机动车运行数据。评价结果表明,Logistic回归与神经网络模型正确判别率均达到90%以上,神经网络具有更好的判别准确率。相对于速度,当前检测地点与上游地点30 m处车速差或速度比更宜作为模型中的判别参数。
In order to improve the discrimination accuracy of the red light warning system of a motor vehicle, a judging model of red light of a vehicle running on the system is proposed and evaluated. Based on the system service requirements and workflow, this paper proposes to use Logistic regression and neural network method to establish discriminant model to predict vehicle stop-pass behavior and to evaluate the model through the measured data of urban traffic. The data was collected from the entrance road of Shanghai urban road intersection, and the data of motor vehicle operation was extracted by Autoscope video detection system. The evaluation results show that the correct discriminant rates of Logistic regression and neural network models are above 90%, and the neural network has a better discrimination accuracy. Relative to the speed, the current speed difference between the detection site and the upstream site at 30 m or the speed ratio is more suitable as a discriminant parameter in the model.