摘要
船闸在长期服役过程中会因为温度、水压等因素影响产生形变,严重危害通航安全。为实现精准的船闸位移预测,构建高效的船闸预测模型,文章引入深度学习方法,基于某大坝船闸的历史观测数据,利用长短记忆神经网络构建了船闸位移预测模型。结果显示文章所提模型最终的预测效果MAE达到了0.0081 mm,AEmax达到了0.0154 mm,RMSE达到了0.0099 mm,均远优于传统的多元线性回归方法。说明该模型具有良好的预测性能,为实现船闸的安全预警提供了一种新方法。
The ship locks will be deformed during long-term service due to temperature,water pressure and other factors,which seriously endangers navigation safety.In order to achieve accurate ship lock displacement prediction and build an efficient ship lock prediction model,this paper introduces the Deep Learning method and constructs a ship lock displacement prediction model based on the historical observation data of a dam ship lock using LSTM neural network.The results show that the final prediction effect of MAE reaches 0.0081 mm,AEmax reaches 0.0154 mm and RMSE reaches 0.0099 mm,which are better than the traditional multivariate linear regression method.The model proposed in this paper has good prediction performance and provides a new method to realize the safety warning of ship locks.
作者
丁腾腾
DING Tengteng(Pearl River Water Resources Institute of Pearl River Water Conservancy Commission,Guangzhou 510611,China)
出处
《现代信息科技》
2023年第16期155-158,共4页
Modern Information Technology
关键词
船闸
LSTM
位移预测
ship lock
LSTM
displacement prediction