【摘要】 为提高船舶交通事故的预测精度,提出将自回归综合移动平均(autoregressive integrated moving average, ARIMA)模型与BP神经网络组合的船舶交通事故预测方法。该方法考虑船舶交通事故的复杂性和非线性因素,充分结合ARIMA模型与BP神经网络的优势,分别从简单加权和残差优化角度对ARIMA模型与BP神经网络的不同组合方法进行比较研究,并将其应用于2000—2018年英国籍船舶交通事故预测中。结果表明:与ARIMA模型、BP神经网络和ARIMA-BP的简单加权组合预测方法进行对比,ARIMA-BP的残差优化组合预测方法的预测精度最高,其均方根误差、平均绝对误差和平均绝对百分比误差分别为7.16、6.0和4.9%。本文提出的船舶交通事故预测方法可以为相关人员的决策提供指导。更多还原

【Abstract】 To improve the prediction accuracy of ship traffic accidents, a prediction method of ship traffic accidents is proposed, which combines the autoregressive integrated moving average(ARIMA) model with BP neural network. In this method, the complexity and nonlinear factors of ship traffic accidents are considered, the advantages of ARIMA model and BP neural network are combined fully, and the different combination methods of ARIMA model and BP neural network are compared from the perspective of simple weighting and residual optimization, respectively. Traffic accidents of British ships from 2000 to 2018 are predicted by different methods. The results show that, compared with ARIMA model, BP neural network and the simple weighted combination prediction method of ARIMA-BP, the residual optimization combination prediction method of ARIMA-BP is of the highest prediction accuracy, and the root mean square error, the mean absolute error and the mean absolute percentage error are 7.16, 6.0 and 4.9%, respectively. The ship traffic accident prediction method proposed in this paper can provide guidance for the decision-making of the related personnel.更多还原