基于改进的BP网络的高速公路事件检测  

Freeway traffic incident detection based on improved BP neural network

作　　者：陈旭生[1] 王保帅[2] 

机构地区：[1]信阳师范学院计算机与信息技术学院,河南信阳464000 [2]郑州铁路局郑州北车辆段,河南郑州450053

出　　处：《信阳农业高等专科学校学报》2009年第2期126-128,共3页Journal of Xinyang Agricultural College

摘　　要：在高速公路交通拥挤状态自动判别(ACI)系统中,高速公路事件检测的精度尤为重要。分析了引起高速公路交通事件的主要因素,提取人、车和环境三者中对交通事件影响比较大的因素的模糊评价值作为输入,然后运用BP神经网络对基本的交通参数进行预测,把预测值和实测值的比值也作为神经网络的输入。为了克服BP网络在学习训练过程中收敛速度慢、容易陷入局部极小的不足,采用了自适应的学习速率和附加动量法,对网络进行训练。最后,将此算法在MATLAB环境中进行仿真。仿真结果表明了算法的有效性。In the freeway traffic automatic congestion identification （ACI） System, the accuracy of the incident detection is especially important. It analyses the major cause of the freeway accidents, and extracts the factors which have more impact on the freeway accident from the factors of people, vehicle and environment. And then use the value of its fuzzy evaluation as the input. The ratio of the results of basic traffic parameters which was forecasted the by BP neural network and really value is also used as the inputs of neural network. In order to solve the Bp neural nctwork＇s inherent deficiency of slowly converging and easily falling into local minimum, we propose to use an algorithm combined self-adapting learning rate and extra momentum. Finally, simulate the algorithm with MATLAB. The result shows that the algorithm is effective.

关 键 词：ACI BP神经网络 交通事件 预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]