【摘要】 港口码头设计中主要考虑的问题之一是系泊船舶的运动量。首先开展了26.6万m~3LNG船舶系泊运动物理模型试验，测量了船舶在试验条件下的六自由度运动量；同时，考虑船舶运动响应的主要影响因素，根据横浪作用下26.6万m~3LNG船运动六分量的试验数据，构建了189组容量的训练集，基于小波分析和神经网络，研究确定三层隐含层神经元数分别为8、12和22,进而建立系泊LNG船舶运动量的预测模型。预测结果表明，小波神经网络模型具有输出多参数的算法优势，能够综合考虑系泊船非线性系统中不易量化的众多影响因素(波浪波高、周期、波长以及船舶自身特性等),给出相对精确的预测结果。小波神经网络模型在研究系泊船舶运动量预测方面具有良好适用性，可以有效地预测船舶的运动量，为实际工程设计提供参考。更多还原

【Abstract】 One of the main problems to be considered in the design of port wharf is the movement of mooring ships. Firstly, the mooring motion physical model test of 266 000 m~3 LNG ship was carried out, and the six-degree-of-freedom motion of the ship was measured under the test conditions. At the same time, considering the main influencing factors of ship motion response, a training set of 189 capacity groups was constructed according to the six component test data of 266 000 m~3 LNG ship movement under the action of transverse waves. Based on the wavelet analysis and neural network, the number of three hidden layers of neurons was determined to be 8, 12 and 22 respectively, and then a prediction model of the movement of moored LNG ships was established. The prediction results show that the wavelet neural network model has the advantage of multi-parameter output algorithm, and can comprehensively consider many influential factors(wave height, period, wavelength and ship′s own characteristics, etc.) which are not easy to be quantified in the nonlinear system of mooring ship, and give relatively accurate prediction results. The wavelet neural network model has good applicability in the study of the motion prediction of mooring ships, which can effectively predict the motion prediction of ships and provide reference for practical engineering design.更多还原