基于小波阈值去噪和函数主成分分析的交通流预测

Traffic Flow Prediction Based on Wavelet Threshold Denoising and Functional Principal Components Analysis

准确的交通流预测不仅有助于改善交通状况,而且有利于推动智能交通系统的发展。大多数当前的预测方法没有充分利用时间序列的潜在函数特性来进行预测。针对这个问题,本文提出了一个基于小波阈值去躁和函数主成分分析的交通流预测模型(Wavelet Threshold De-noising-Functional Principal Components Analysis, WTD-FPCA)。使用中国贵阳市真实的交通数据集对所提模型进行验证,并使用均方根误差(Root Mean Square Error, RMSE),平均绝对百分比误差(Mean Absolute Percentage Error, MAPE)和均方误差(Mean Square Error, MSE)来评价WTD-FPCA模型的预测性能。在预测性能比较中,我们考虑了季节差分自回归移动平均(Seasonal Autoregressive Integrated Moving Average, SARIMA)、长短期记忆(Long Short-Term Memory, LSTM)网络、循环神经网络(Recurrent Neural Network, RNN)和门控循环单元(Gate Recurrent Unit, GRU)。预测结果表明,WTD-FPCA模型的预测性能最优。

Accurate traffic flow forecasting not only helps to improve traffic conditions, but also facilitates the development of intelligent transportation systems. Most current forecasting methods do not make full use of the potential function property of time series for prediction. To address this problem, a traffic flow prediction model based on Wavelet Threshold Denoising and Functional Principal Components Analysis (WTD-FPCA) is proposed in this paper. The proposed model is validated using a real traffic dataset in Guiyang, China, and the Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and Mean Square Error (MSE) are used to evaluate the prediction performance of the WTD-FPCA model. For the prediction performance comparison, we considered Seasonal Autoregressive Integrated Moving Average (SARIMA), Long Short-Term Memory (LSTM) network, Recurrent Neural Network (RNN) and Gate Recurrent Unit (GRU). The prediction results show that the WTD-FPCA model has the best prediction performance.