基于ARIMA的短时交通量预测模型

Short-term traffic flow forecast based on ARIMA model

为了对交通控制与诱导系统提供更可靠的数据,有必要对实时、动态的数据进行短时交通量在线预测。本文对短时交通量预测的ARIMA建模方法进行了介绍,用2008年10月7日以3min为单元的短时交通量数据作为训练数据,将其平稳化处理并标定模型为ARIMA(5,1,6),采用最大似然估计法估计模型的参数,残差白噪声检验验证其有效性,用静态预测法对8~10日三天的数据进行实时动态预测,结果表明ARIMA模型在短时交通量预测方面可移植性较强,可以在历史数据较少的情况下准确的捕捉其内在规律,并对大量数据进行静态预测,可以在线、实时、动态预测下一时刻的数据,给交通管理系统提供更可靠的实时数据。

In order to provide more reliable data for traffic control and guidance system,it is necessary to carry out short-term traffic volume online prediction for real-time and dynamic data.Firstly,ARIM A modeling method for short-term traffic volume prediction is introduced.The short-term traffic volume data on October 7,2008 in 3 min units is used as training data and smooth it out and calibrated as ARIM A(5,1,6).Then the maximum likelihood estimation method is used to estimate the parameters of the model,which use the white noise test of residual error to verify its effectiveness.Later the static prediction method is used to carry out real-time dynamic prediction on the data on the 8th,9th and 10th.The results showthat ARIM A model has strong portability in short-term traffic volume prediction and ARIM A model can accurately capture its inherent laws and carry out static prediction on a large amount of data under the condition of less historical data.It can also predict the data at the next moment in real time and dynamically,and provide more reliable real-time data for traffic management system.