Aiming at the problem that the spatial characteristics of ship traffic flow data are seldom considered in ship traffic flow prediction models,a deep learning prediction model is proposed.The model is composed of Chebyshev network(ChebNet)and the long short-term memory(LSTM),so it is called ChebNet-LSTM.ChebNet K-order convolution operator is helpful to extract the spatial characteristics of ship flow data,and LSTM is used to learn the temporal characteristics of ship flow data.The ship flow prediction experiment is carried out in three areas with different ship flow in Zhoushan waters.The results show that,the proposed ChebNet-LSTM model can effectively extract the temporal and spatial characteristics of ship flow data,its performance on various evaluation indices is better than that of the comparison model,the prediction accuracy is greatly improved,and it can provide data support for intelligent navigation of water traffic.
万方数据
维普
