对船舶姿态的准确预测有利于保障船舶航行安全、提高船载仪器的数据质量.针对船舶运动姿态预测与分析问题,将特征选择(Feature Selection,FS)方法与长短期记忆神经网络(Long Short Term Memory,LSTM)相结合,以提高LSTM网络对船舶运动姿态的预测精度,为船舶运动姿态的补偿奠定基础.以船舶运动姿态预测为目标,基于无人艇采集到的包含船舶运动状态信息的数据集,构建并训练了适应性较强的FS-LSTM神经网络预测模型.最终结果表明,本文所提出的方法具有较高的预测精度.
Ship motion state prediction based on FS-LSTM network
The accurate prediction of ship attitude is conducive to ensuring ship navigation safety and improving the data quality of shipborne instruments.Aiming at the problem of ship motion attitude prediction and analysis,this paper com-bines the Feature Selection(FS)method with the Long Short Term Memory(LSTM)neural network to improve the accur-acy of ship motion attitude prediction by LSTM network.It lays the foundation for the compensation of ship motion attitude.Based on the data set of ship motion state information collected by unmanned boat,this paper constructs and trains the FS-LSTM neural network prediction model with strong adaptability.The final results show that the proposed method has high prediction accuracy.
万方数据
维普
