基于BP神经网络对横浪作用下系泊油船的运动量预测分析  

Analysis for predicting motion of oil tanker moored ships under cross wave based on BP neural network

作　　者：贾子锌 柳淑学[1] 李金宣[1] 饶正刚 JIA Zi-xin;LIU Shu-xue;LI Jin-xuan;RAO Zheng-gang(State Key Laboratory of Coastal and Offshore Engineering,Dalian University of Technology,Dalian 116024,China)

机构地区：[1]大连理工大学海岸和近海工程国家重点实验室,大连116024

出　　处：《水道港口》2022年第4期430-436,共7页Journal of Waterway and Harbor

基　　金：国家自然科学基金项目(51879037,51739010)。

摘　　要：系泊船舶的运动量预测是工程设计的关键因素之一。针对1万t级油船的系泊进行试验,构建了305组容量的训练集,结合系泊船舶运动变化的理论分析确定其影响因素。基于BP神经网络,利用反向传播算法训练建立5层隐含层人工神经网络模型,对油船系泊运动的运动量进行预测。模型的输入层包括6个参数,即入射波浪波高、周期、船舶吃水、水深、船舶横摇周期及纵摇周期。输出层为系泊油船运动量的六分量,即纵移、横移、升沉、横摇、纵摇、回转。结果表明,BP神经网络模型具有输出多参数的算法优势,能够综合考虑众多系泊船非线性系统中不易量化的影响因素,给出相对精确的预测结果。The prediction of mooring ship′s motion is one of the key factors in engineering design.The 10000 t mooring oil tanker experiment was designed as research object.A training set of 305 groups was constructed.Combined with the theoretical analysis of mooring ship motion,the influencing factors were determined.Based on BP neural network,a 5 hidden layer artificial neural network model was established by using back-propagation algorithm training to forecast quantity of mooring ship′s motion.The input layer of the model included six parameters,namely incident wave′s height and period,draught and water depth,roll period and pitch period.The output layer was the six component of mooring tanker′s motion,namely surge,sway,heave,roll,pitch and yaw.The results show that the BP neural network model could output multi-parameter simultaneously.It can comprehensively consider many factors which are not easy to be quantified in the nonlinear system of mooring ships,and give relatively accurate prediction results.

关 键 词：油船系泊运动 运动量 神经网络模型 预测 

分 类 号：U661[交通运输工程—船舶及航道工程]