基于样本船的兴波阻力智能预报方法

Intelligent prediction method of wave-making resistance based on sample ship

为了实现设计空间中船体兴波阻力的快速预报,对样本船型生成、兴波阻力预报和近似模型构建展开研究,建立基于样本船的兴波阻力智能预报方法。首先描述曲面变形技术的理论方法,将改进平移法和自由变形技术两种方法应用于船体变形;其次给出兴波理论和数值计算方法,验证本文兴波阻力计算程序的可靠性;并描述基于连续局部枚举的试验样本点设计方法和基于BP神经网络的近似模型;最后以KCS船型特征参数设计为实例,通过近似模型训练和参数分析,验证该近似模型的计算精度。研究结果表明:根据60组测试样本的结果验证,相关系数为0.9998,均方误差为3.14E-04,运行时间为2 ms,本文智能预报方法的预测精度高且速度快。

In order to realize the fast prediction of hull wave resistance in design space,the intelligent predic⁃tion method of wave-making resistance based on sample ships was established by analyzing the generation method of sample ship form,prediction of wave-making resistance and construction of an approximate model.Firstly,the theoretical method of surface deformation technology was described,and two methods of improved translation and free deformation technology were applied to hull deformation.Secondly,the wave-making the⁃ory and numerical calculation method were given to verify the reliability of the wave-making resistance calcu⁃lation program.The design method of experimental sample points based on continuous local enumeration and the approximate model based on BP neural network were described.Finally,with the design of KCS hull form characteristic parameters taken as an example,the accuracy of the approximate model was verified through the training and parameter analysis of the approximate model.The results show that,according to the results of 60 groups of test samples,the correlation coefficient is 0.9998,the mean square error is 3.14E-04,and the running time is 2 ms.The intelligent forecasting method has a high precision and a high speed.