摘要
研究交通流状态的分类、识别与预测,建立了基于模糊聚类及模式识别的交通流状态自适应模糊神经推理系统。对大量交通流历史特征数据采用模糊聚类法进行状态分类并进行模式识别,得到系统的原始输入输出数据集。建立交通流状态预测的自适应模糊神经系统,以交通流特征数据及其识别结果作为训练数据集进行系统参数及模糊规则的训练与确定,直到误差在控制范围内,并进行系统检测和复核。仿真及其检测和复核结果表明系统预测的准确率在95%以上。
The clustering, recognition, and prediction of traffic flow patterns were studied. An adaptive neuro-fuzzy inference system （ANFIS） was established based on fuzzy clustering and pattern recognition of traffic flow patterns. Firstly, a large quantity of traffic flow data was classified and identified by fuzzy clustering method and recognition rules. The initial input-output data of ANFIS were obtained. Then, the adaptive neuro-fuzzy system of traffic flow patterns was established. The system trained itself with the data and constructed fuzzy inference rules until the prediction error was under control. At last, the whole system was tested and checked. The results of simulation, testing, and checking illustrate that the accuracy of the prediction system is above 95%.
出处
《交通与计算机》
2007年第4期46-49,共4页
Computer and Communications
关键词
交通工程
交通流状态预测
模糊聚类
模式识别
自适应模糊神经推理
traffic engineering
traffic flow pattern prediction
fuzzy clustering
pattern recognition
adaptive neuro-fuzzy inference
