基于改进ARIMA算法的船舶纵摇运动预测  

Ship pitching motion prediction based on improved ARIMA algorithm

作　　者：张婷[1,2] 邱亮 王培良[1,2] ZHANG Ting;QIU Liang;WANG Peiliang(Navigation College,Shandong Transport Vocational College,Weifang 261206,China;Merchant Marine Collge,Engineering Research Center of Simulation Technology of the Ministry of Education,Shanghai Maritime University,Shanghai 201306,China)

机构地区：[1]山东交通职业学院航海学院,山东潍坊261206 [2]上海海事大学商船学院航运仿真技术教育部工程研究中心,上海201306

出　　处：《舰船科学技术》2023年第21期64-69,共6页Ship Science and Technology

基　　金：国家青年自然科学基金资助项目(51909155);潍坊市软科学研究计划项目(2021RKX125)。

摘　　要：针对现有自回归移动平均模型(ARMA)在时间序列预测不具有普适性问题,提出构建差分自回归移动平均模型(ARIMA)对船舶摇摆角度值进行预测,并对比分析不同差分次数时的预测效果。首先,分析船舶纵摇角度值原始序列值,检验其平稳性,进行差分运算,明确ARIMA模型的适用性;然后,依据AIC准则确定模型的阶数,并求解模型参数。最后,进行角度值序列预测,结合预测精度和误差分析,确定适用于预测船舶纵摇角度值的ARIMA模型。研究结果表明:以误差和均方差为模型评价指标,优化差分次数获得的ARIMA模型具有更好的预测效果。Aiming at the problem that the existing autoregressive moving average model(ARMA)does not have universal applicability in time series forecasting,it is proposed to construct the autoregressive integrated moving average model(ARIMA)to predict the ship's pitch angle values,and to compare and analyze the predictions effect on differential times.Firstly,the original ship pitch angle values are analyzed.The values stationarity is checked.The values are performed difference calculations to clarify the applicability of the ARIMA model.Then,the model order is determined according to the AIC criterion and the model parameters are solved.Finally,the ship's pitch angle values are predicted,and the final ARIMA model combined with prediction accuracy and error analysis is obtained.

关 键 词：ARIMA模型 最优差分次数 船舶运动预测 集装箱船 

分 类 号：U661.42[交通运输工程—船舶及航道工程]