基于GRNN的船舶阻力预测  

Ship Resistance's Prediction Based on GRNN

作　　者：郎济才[1] 谢行[1] 范蠡 蔡世泽 

机构地区：[1]哈尔滨工程大学船舶工程学院,黑龙江哈尔滨150001 [2]广州船舶及海洋工程设计研究院,广东广州510250

出　　处：《武汉船舶职业技术学院学报》2014年第3期18-20,24,共4页Journal of Wuhan Institute of Shipbuilding Technology

摘　　要：文章通过介绍广义回归神经网络(GRNN)的原理及其非线性拟合的能力,建立了基于GRNN的船舶阻力预测模型。本文将该模型运用到Taylor系列船舶剩余阻力预测中,得出预测结果。通过与实际阻力比较,得出预测阻力的相对误差在5%之内的百分比为93.98%,最后本文通过预测插值后的数据,得出相对误差在5%以内占85.38%。由此可以看出,GRNN预测船舶阻力有很高的精度和较强的泛化能力,为船舶阻力的预测提供了一种简便的预测方法。This paper introduces the Generalized Regression Neural Network（GRNN）and the principle of nonlinear fitting ability,establishing aprediction model of ship resistance based on GRNN.In this paper,the model is applied to the Taylor series marine residual resistance prediction,and we get the prediction results.By comparing them with the actual resistance,the percentage of the relative errors of prediction of resistance within the range of5%is 93.98%.At the end of this paper,through the data coming from the prediction of interpolation,we get a percentage of the relative error within 5% of 85.38%.From this we can see that the GRNN prediction of ship resistance has high precision and good generalization ability,providing a simple prediction method for the prediction of ship resistance.

关 键 词：广义回归神经网络 船舶阻力 预测误差 

分 类 号：U661.31[交通运输工程—船舶及航道工程]