基于BP神经网络的高速公路交通事件检测

Traffic Incident Detection on Highway Based on BP Neural Network

回顾了几种传统的交通事件检测算法，提出基于多层前向人工神经网络的检测模型，并运用BP算法予以实现。在将BP算法与传统算法进行比较之后，发现BP算法具有检测率高、误报率低、平均检测时间短的优点，同时也存在不足之处，指出了今后进一步研究的方向。

Several traditional algorithms used in traffic incident detection are reviewed in the paper. A detection model based on multi-layer artificial neural network is proposed and implemented with a BP algorithm. By comparing BP algorithm with traditional California algorithm, it is found that BP algorithm is of higher detection rate, lower false alarm rate, shorter mean detection time, though there are some disadvantages with it. The further research directions are also given.