基于神经网络的船舶倾覆预警系统  

Pre-warning system for ship capsizing based on natural network

作　　者：韩梦佳 曲铭雯 HAN Meng-jia;QU Ming-wen(School of Computing,Jiangsu University of Science and Technology,Zhenjiang 212003,China;School of Electronic Information,Suzhou University,Suzhou 215000,China)

机构地区：[1]江苏科技大学计算机学院,江苏镇江212003 [2]苏州大学电子信息学院,江苏苏州215000

出　　处：《电子设计工程》2020年第6期64-68,共5页Electronic Design Engineering

摘　　要：随着我国国际航运中心的不断建设和发展,水路通航密度增加,船舶大型化发展日益明显,加之海上恶劣极端天气呈常态化趋势,发生重大海上安全事故的风险增加,因此人们对于船舶运输的安全性提出了更高的要求。本文针对船舶市场发展现状和船载人员的安全需求,提出了一套基于神经网络的船舶倾覆预警系统,通过加速度传感器获取船舶实时倾斜数据并上传至服务器,使用TensorFlow神经网络进行数据训练,最后得到船舶倾覆模型。系统可将实时获取的数据与此模型进行比对,进而推演预测船舶是否有倾覆的可能性,可做到提前预警并及时联系就近的海事组织做好营救措施,有效保障船舶所载人员的生命安全,具有一定的研究价值。With the rapid development of the world economy and Internet,and the continuous improvement of ship automation technology,people put forward higher performance requirements for the safety of ship transportation,and at the same time,they also put forward new suggestions on whether the ship can send out rescue signals in time and effectively after distress.At present,there is no mature early warning system for ship capsizing in China,and the prediction of ship capsizing is still in the stage of research and simulation.In this paper,aiming at the existing safety problems and the development status of ship market,a ship capsizing early warning system based on natural network is proposed.The real-time ship tilting data are acquired by gyroscope sensor and uploaded to the server.The data are trained by TensorFlow neural network.Finally,the ship capsizing model is obtained.The system can compare the real-time data with the model,and then deduce whether the ship is likely to capsize.It can make early warning and timely contact the nearest IMO to do rescue measures,effectively guarantee the safety of the people on board.

关 键 词：船舶倾覆 预警 神经网络 Node-RED 树莓派 

分 类 号：TN919.5[电子电信—通信与信息系统]