基于相空间重构—SVM的公交客流量预测研究

Predictive Study on Passenger Volume in Bus Stops Based on Phase Space Reconstruction and Support Vector Machine

提出了一种基于相空间重构与支持向量机预测公交客流量的新方法。应用互信息法计算公交客流量时间序列的最优时间延迟;应用Cao氏方法计算其最佳嵌入维数;然后计算出最大Lyapunov指数,证实客流量存在混沌现象。建立相空间重构—支持向量机预测模型并确定训练样本对,对公交客流量数据进行预测。实例证明,该方法能有效地进行客流量预测。

The paper presents a new method for forecasting passenger volume in bus stops based on phase space reconstruction and support vector machine（SVM）.Mutual information is used to calculate the optimal time lag of passenger volume time series.We adopt Cao to calculate the optimal embedding dimension.Then,we can calculate the Lyapunov exponent,and prove that the passenger volume chaos phenomena exist.A phase space reconstruction SVM model is constructed and pairs of training samples are determined to forecast the real passenger volume.The experimental result shows that this method can predict passenger volume effectively.