Abstract: In order to improve the prediction accuracy of city traffic flow and overcome the shortcomings that a single prediction model can’t well reflect the essential characteristics of traffic flow, based on the chaotic characteristic of traffic flow, the coupled Kalman filtering theory and the principles of phase space reconstruction phase method is proposed, a Kalman filtering traffic flow forecasting model based on phase space reconstruction is established. This model is based on phase points of the phase space reconstruction as state vector phase point description, using the Kalman filtering theory in real time prediction and correction phase point of future evolution, the simulation based on the traffic flow data of a section of Jiaozuo city is carried out. By comparing the related performance index analysis, results show that Kalman filtering prediction index model based on phase space reconstruction is better than the single model unimproved, the prediction accuracy is improved by 16.75%.

Key words: chaotic characteristics, Kalman filtering, phase space reconstruction, short-term traffic flow, prediction model

摘要： 为了提高城市交通流的预测精度，克服单一预测模型不能很好反映交通流本质特征的缺点，在交通流混沌特性的基础上，提出将卡尔曼滤波理论与相空间重构原理相耦合的方法，建立基于相空间重构的卡尔曼滤波交通流预测模型。此模型以相空间重构的相点作为状态向量构成相点的状态空间描述，运用卡尔曼滤波理论实时预测并校正相点的未来演化规律，并根据焦作市某路段的交通流数据进行实例仿真。通过相关性能指标对比分析，结果表明，基于相空间重构的卡尔曼滤波预测模型各项指标明显优于未改进的单一模型，使预测精度提高了16.75%。

关键词: 混沌特性, 卡尔曼滤波, 相空间重构, 短时交通流, 预测模型