舰船运动姿态极短期预报算法研究  

Extreme short-term prediction algorithm for ship attitude motion

作　　者：吴爽[1] 焦淑红[1] 任慧龙[2] WU Shuang;JIAO Shuhong;REN Huilong(College of Information and Communication Engineering,Harbin Engineering University,Harbin 150001,China;College of Shipbuilding Engineering,Harbin Engineering University,Harbin 150001,China)

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001 [2]哈尔滨工程大学船舶工程学院,黑龙江哈尔滨150001

出　　处：《应用科技》2019年第4期6-10,共5页Applied Science and Technology

基　　金：国家自然科学基金项目(KY10100160075)

摘　　要：舰船运动姿态数据流的极短期实时预报可以帮助决策者在决策过程中更好地分析问题、评价和制定方案,具有很好的参考价值,故着重对该部分进行研究。考虑到六自由度运动中横摇的影响,故主要针对舰船横摇运动姿态数据流序列具有混沌属性并且连续量大的特点,结合数据流挖掘理论框架,提出了一种基于小波变换的递推最小二乘(recursive least square,RLS)的Volterra核估计算法,用于对监测系统中采集到的连续的横摇运动姿态数据流进行实时预报研究。该方法首先对姿态数据流概要结构进行获取,然后利用小波阈值降噪,最后将降噪处理的数据利用RLS的Volterra核估计算法进行实时预报。通过在舰船横摇运动姿态预报的实践验证表明,该算法可很好地解决运动姿态数据流在线自适应预报问题。The extreme short?term real?time prediction of ship motion attitude data stream can help decision?makers to better analyze,evaluate and formulate schemes in the decision?making process,which has good reference value.Therefore,this part is mainly studied.Aiming at the chaotic property and large continuity of ship motion attitude da?ta stream sequence,a Volterra kernel estimation algorithm based on recursive least squares(RLS)of wavelet trans?form is proposed,which is combined with the theoretical framework of data stream mining.It is used to predict the continuous motion attitude data stream collected in the monitoring system in real time.In this method,firstly,the outline structure of the attitude data stream was acquired,then the wavelet threshold was used to denoise,and fi?nally the de?noised data was predicted in real time using RLS Volterra kernel estimation algorithm.The practical verification in ship motion attitude prediction shows that the algorithm can solve the problem of on?line adaptive pre?diction of motion attitude data stream.

关 键 词：运动姿态序列 数据流 VOLTERRA RLS 小波变换 预报 

分 类 号：U662.9[交通运输工程—船舶及航道工程]