基于TVF-EMD和ARIMA模型的短时交通量预测研究

Research on Short-Term Traffic Volume Prediction Based on TVF-EMD and ARIAM Models

摘要: 为描述短时交通量数据中的非线性和非平稳性成分，以提高短时交通量预测精度，进而促进智能交通系统的发展，本文提出了一种基于时变滤波经验模态分解(TVF-EMD)方法与差分整合移动平均自回归(ARIMA)模型的混合预测模型，即TVF-EMD-ARIMA模型。首先利用TVF-EMD对处理后的交通量数据进行分解，再对分解后的序列建立ARIMA模型进行预测。研究结果表明：相比于经验模态分解(EMD)方法和变分模态分解(VMD)方法，TVF-EMD方法分解得到的交通量序列更加平滑；混合预测模型TVF-EMD-ARIMA与单一ARIMA模型相比，其平均绝对误差、平均绝对百分比误差和均方根误差分别降低了3.6700、0.0775、5.3539。

Abstract: In order to describe the nonlinear and nonstationary components in short-term traffic volume data, so as to improve the accuracy of short-term traffic volume prediction and promote the de-velopment of intelligent transportation systems, this paper proposes a hybrid prediction model based on time-varying filtering empirical mode decomposition (TVF-EMD) method and autoregressive integrated moving average (ARIMA) model, namely TVF-EMD-ARIMA model. Firstly, TVF-EMD is used to decompose the processed traffic data, and then an ARIMA model is established for prediction of the decomposed sequence. The results show that compared with the empirical mode decomposition (EMD) method and the variational mode decomposition (VMD) method, the traffic sequence obtained by TVF-EMD method decomposition is smoother. Compared with the single ARIMA model, the mean absolute error, mean absolute percentage error and root mean square error of the hybrid prediction model TVF-EMD-ARMA are reduced by 3.6700, 0.0775 and 5.3539, respectively.