一种交通流状态智能推理系统

An Intelligent Inference System of the Traffic Flow Status

设计一种基于模糊聚类及模式识别的交通流状态自适应神经模糊推理系统,用来研究交通流状态的分类、识别及预测。首先,对大量交通流历史特征数据采用模糊聚类的方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。然后,建立交通流状态预测的自适应神经模糊系统,以交通流特征数据及其识别结果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。最后,进行系统检测和复核。仿真及其检测和复核结果表明了系统良好的应用性能。

This paper aims at the clustering, recognization and prediction of traffic flow patterns. The paper presents an intelligent system combining fuzzy clustering, pattern recognization and adaptive neuro-fuzzy inference systems （ANFIS）. Firstly, a large quantity of traffic flow data is classified and identified by the fuzzy C-means method and recognition rules. The result is the initial input-output data of ANFIS. Then, the system trains itself with the data and constructs the fuzzy inference rules until the prediction errors are in control. Finally, we test and check the whole system. The simulation and its test and check results illustrate that the system is applicable.