Abstract: To improve the prediction accuracy of port cargo throughput and further provide data support for port construction, Elman neural network with the capability of dealing with dynamic information was introduced. Elman neural network was applied to the forecast of cargo throughput in Ningbo-Zhoushan port, and the time series data were constructed by using the first six months data to recursively predict the next month data. Meanwhile, the prediction results were compared with those of BP neural network and RBF neural network. The results show that: in terms of port cargo throughput forecast, compared with BP neural network and RBF neural network, Elman neural network can better adapt to the characteristics of the throughput data varying with time, and its prediction values are closer to the actual values, whose prediction performance is better. And it can better reflect the actual state of the port.

Key words: traffic and transportation engineering, port, cargo throughput, Elman neural networks, forecasting, dynamic learning

摘要： 为提高港口货物吞吐量的预测精度，进而为港口建设提供数据支持，引入具有处理动态信息能力的Elman神经网络。将Elman神经网络应用于宁波舟山港的货物吞吐量预测，采用前6个月数据递归预测后一个月数据的方式构建时间序列数据，同时与BP神经网络以及RBF神经网络的预测结果进行分析比较。结果表明：在港口货物吞吐量预测方面，相比于BP神经网络以及RBF神经网络，Elman神经网络更能适应吞吐量数据随时间变化的特性，其预测值更接近实际值，其预测性能更优，且更能体现港口实际状态。

关键词: 交通运输工程, 港口, 货物吞吐量, Elman神经网络, 预测, 动态学习