基于SARIMA模型的船舶交通流量预测研究

Vessel Traffic Flow Prediction Based on the SARIMA Model

为提高船舶月交通流量预测精度,更合理地为港口规划和发展提供决策依据,选用季节性差分自回归滑动平均(seasonal autoregressive integrated moving average,SARIMA)模型对船舶月交通流量建立了预测模型,并利用Eviews软件,以2007年1月-2015年12月荆州港船舶交通流月均流量统计数据为样本进行了实证分析.对船舶月交通流量时间序列样本数据进行平稳化预处理,消除其趋势成分和季节因素;基于平稳化后的数据建立了SARIMA模型并对模型进行参数检验及最优模型选取;并利用所获得的最优模型SARIMA(2,0,0)(1,1,1)12对2008年1月-2016年3月荆州港船舶交通流月均流量进行预测,并将预测结果与AR(1)模型、季节指数模型的预测结果进行对比分析.对比分析结果表明,SARIMA的预测精度更高,更能反映船舶月交通流量的变化情况,利用该模型对船舶月交通流量进行建模预测具有较好的实用性.

To improve the predictive accuracy of vessel traffic flow and provide more reasonable deci-sion-making basis for port planning and development, Seasonal Autoregressive Integrated Moving Average （SARIMA） model is put forward to predict the monthly traffic flow of vessel. Based on the soft-ware Eviewsis, empirical analysis is carried out for the vessel traffic flow monthly statistical data of Jingzhou port during January 2007 - December 2015. Firstly, the sample data from the vessel traffic flow monthly statistics of Jingzhou port is executed stationary pre-process, in order to eliminate the trend component and seasonal factors of the statistical data. Afterwards, the SARIMA model based on the data through stationary pretreatment is set up. Then the model parameters are test and the optimal model SARIMA（2,0,0） （1,1, 1 ）12 is validated. Finally, the prediction of the vessel traffic flow during January 2008 - March 2016 of Jingzhou port is made, and the prediction results are compared with the those using AR （1） model and seasonal exponential model. The comparison results show that the SARIMA prediction accuracy is higher, and can reflect the monthly change characteristics of vessel traffic flow more accurately.