基于GRU-DconvLSTM的船舶轨迹预测

Ship Trajectory Prediction Based on GRU-DconvLSTM

利用深度学习方法预测船舶未来航行趋势,对海上交通安全以及船舶管理具有重要意义。在船舶自动识别系统(AIS)中已知的经度、纬度、航速数据基础上,提出一种基于门控循环单元结合双卷积层长短期记忆神经网络(GRU-DconvLSTM)预测模型。根据原始数据的变化趋势,采用标准差法对数据中的异常值进行处理,得到最终试验数据。该模型一方面通过门控循环单元(GRU)学习船舶历史数据上的运动规律;并采用双卷积层与长短期记忆神经网络(LSTM)结合的形式充分提取数据深层信息,提高模型对时序数据深层次特征的挖掘能力。将该模型与卷积长短期记忆神经网络(CNN-LSTM)、卷积门控循环神经网络(CNN-GRU)以及卷积层长短期记忆(Conv-LSTM)神经网络等3个模型进行对比,将均方根误差、平均绝对误差、平均绝对百分比误差作为评价标准,结果表明,GRU-DconvLSTM模型在经度和纬度预测上误差较小,精确度较高。

Deep learning methods are used to predict the future navigation trend of ships and are of great significance to maritime traffic safety and ship management.Based on the known longitude,latitude,and speed data from the automatic identification system(AIS),a prediction model based on Gated Recurrent Unit-Double convolutional Layers of Long Short-Term Memory(GRU-DconvLSTM)is proposed.According to the changing trend of the original data,the standard deviation method is used to deal with the outliers in the data to obtain the final experimental data.On the one hand,the Gated Recurrent Unit(GRU)is used to learn the movement law of the ship on the historical data.The combination of double convolutional layers and a Long Short-Term Memory neural network(LSTM)is used to fully extract the deep information of the data,to improve the model's ability to mine the deep features of time series data.The model is combined with Convolutional Neural Network-Long Short-Term Memory,CNN-LSTM,Convolutional Neural Network-Gated Recurrent Unit,convolutional neural network-gated recurrent unit,convolutional neural network-gated recurrent unit(CNN-GRU)and Convolutional Long Short-Term Memory(Conv-LSTM)neural network,the root mean square error,mean absolute error and mean absolute percentage error are used as evaluation criteria.The results show that,the GRU-DconvLSTM model has smaller error and higher accuracy in longitude and latitude prediction.