基于优化X-12-ARIMA模型的船舶交通流量预测  

Prediction of ship traffic flow based on optimized X-12-ARIMA model

作　　者：陶鹤 TAO He(School of Economics,Lanzhou Technology and Business College,Lanzhou 730101,China;Gansu Province State Key Laboratory of Regional Recycling Economy,Lanzhou 730101,China)

机构地区：[1]兰州工商学院经济学院,甘肃兰州730101 [2]甘肃省高校区域循环经济重点实验室,甘肃兰州730101

出　　处：《高师理科学刊》2022年第2期25-30,共6页Journal of Science of Teachers＇College and University

基　　金：甘肃省社科规划项目(20-002D);甘肃省高校区域循环经济重点实验室开放基金课题项目(QXKJ2020-004)。

摘　　要：月度船舶交通流量数据具有较强的季节性,在提高数据预测精准度的同时,应提取其季节波动和长期趋势加以分析,而不是单纯预测未来发展趋势.为从数据中获得更多有效信息,利用时间序列ARIMA模型对原始数据进行拟合,使用残差平方和、均方根误差、AIC函数和SBC函数衡量模型拟合效果,选取局部最优模型.经比较后,选取X-12-ARIMA季节乘法模型拟合月度船舶交通流量数据,得到了季节波动、长期趋势和不规则变动随时间而发生的变化,并预测了未来12期的船舶交通流量.在此基础上,调用径向基神经网络函数对数据进行仿真研究.结果表明,采用优化的X-12-ARIMA模型预测船舶交通流量时,预测精度有了较大提高.Monthly ship traffic data should have strong seasonality.While improving the accuracy of data prediction,seasonal fluctuations and long-term trends should be extracted for analysis,rather than simply predicting future development trends.In order to obtain more effective information from the data,time series ARIMA model was used to fit the original data,residual sum of squares,root mean square error,AIC function and SBC function were used to measure the model fitting effect,and the local optimal model was selected.After comparison,X-12-ARIMA seasonal multiplication model is selected to fit monthly ship traffic data,and the changes of seasonal fluctuations,long-term trends and irregular changes over time are obtained,and the ship traffic flow in the next 12 periods is predicted.On this basis,RBF neural network function was used to simulate the data.The results show that the prediction accuracy has been greatly improved when the optimized X-12-ARIMA model is used to predict ship traffic flow.

关 键 词：船舶交通流量 季节性 X-12-ARIMA模型 RBF神经网络 

分 类 号：O29[理学—应用数学] U69[理学—数学]