孙世超 董曜 李娜 郑勇

摘要：為准确预测集卡的港内周转时间，进而提升整个物流系统的作业效率，通过对集装箱码头闸口数据进行深入分析，得到3种不同任务类型的集卡港内周转时间序列，并在此基础上提出一种基于小波分解和自回归移动平均（autoregressive moving average， ARMA）模型的集卡港内周转时间预测方法。该方法首先利用小波分解技术对集卡港内周转时间序列的多维变化特征进行逐层分离，再利用ARMA模型对分离后的多个时间序列分别进行拟合，然后对拟合结果进行合并，以此近似模拟原序列的时变规律，继而实现集卡港内周转时间的短期预测。为验证该方法的有效性，将数据样本划分为训练集（75%）和测试集（25%），训练集用于拟合多维ARMA模型，测试集用于检验ARMA模型的预测结果误差。研究结果表明，对于3种任务类型，该模型均可以精确预测集卡的港内周转时间，为物流企业调整集卡运输计划提供相应的技术支持。

关键词：  水运管理; 集卡周转时间预测; 小波分解; 自回归移动平均（ARMA）模型; 码头闸口数据

中图分类号：  U691+.3

文献标志码：  A

收稿日期： 2021-03-25

修回日期： 2021-06-21

基金项目： 国家自然科学基金（71702019）

作者简介：

孙世超（1988—），男，辽宁大连人，副教授，博士，研究方向为交通运输工程，（E-mail）dlmu_sunshichao@163.com

Meeting of the Waterborne Transport Division， World Transport Convention 2021 （WTC 2021）

Truck turnaround time prediction in a port based on wavelet decomposition

SUN Shichao， DONG Yao， LI Na， ZHENG Yong

（College of Transportation Engineering， Dalian Maritime University， Dalian 116026， Liaoning， China）

Abstract： In order to accurately predict the in-a-port turnaround time of trucks and improve operation efficiency of the whole logistics system， three different task types of in-a-port turnaround time series of trucks are obtained through the analysis on the gate data of container terminals. On this basis， a method for predicting the in-a-port turnaround time of trucks based on the wavelet decomposition and the autoregressive moving average （ARMA） model is proposed. This method initially employs the wavelet decomposition technology to separate the multi-dimensional change characteristics of the in-a-port turnaround time series of trucks， and then applies the ARMA model to fit the time series after separation， respectively. Subsequently， the fitted results are merged to simulate approximately the time-varying law of the original series， and then realize the short-term prediction of the in-a-port turnaround time of trucks. In order to verify the effectiveness of the method， this study divides the data sample into a training set （75%） and a test set （25%）. The training set is used to fit the multi-dimensional ARMA model， and the test set is used to test the prediction error of the ARMA model. The results show that for the three task types， the model can accurately predict the in-a-port turnaround time of trucks， and it can provide corresponding technical support for the adjustment of truck transportation plan of logistics enterprises.

Key words： water transportation management; prediction of truck turnaround time; wavelet decomposition; autoregressive moving average （ARMA） model; terminal gate data

0 引 言

随着外贸经济的不断发展，近些年我国几个主要港口的集装箱吞吐量增长迅速，如深圳盐田港的集装箱吞吐量已连续三年突破1 400万TEU。在此背景下，提高港口的现代化管理水平，提升集装箱码头设备的作业效率，减少集卡的周转时间，成为推动我国港口绿色化和智慧化发展需关注的重要领域之一。……