潜艇操纵水动力小波神经网络预报方法研究  

A Study on Methods of Submarine Manoeuvre Hydrodynamic Coefficients Prediction Using Wavelet Neural Networks

作　　者：曾广会[1] 朱军[2] 刘洁[1] 

机构地区：[1]海军潜艇学院,青岛266042 [2]海军工程大学船舶与海洋工程系,武汉430033

出　　处：《中国造船》2013年第3期132-138,共7页Shipbuilding of China

基　　金：国家自然科学基金(51179199)

摘　　要：针对潜艇操纵性优化设计中水动力系数预报问题,在潜艇水动力预报中引入艇体肥瘦指数概念,确定了潜艇艇体几何描述的五参数模型。提出采用小波神经网络方法预报潜艇水动力,确定了神经网络的结构,利用均匀试验设计方法,设计了神经网络的学习样本。在验证CFD预报艇体水动力有效的基础上,完成了样本水动力系数的CFD计算;通过对样本进行学习,完成了潜艇艇体操纵性水动力系数小波神经网络预报。研究结果表明,只要确定适当的输入参数,选择适当的学习样本和网络结构,利用小波神经网络方法对潜艇水动力进行预报可以达到较高的精度。With respect to the prediction of hydrodynamic coefficients applied to maneuver performance optimization in submarine design, a fat-thin index of submarine hull is introduced, and a five parameters model is conformed. A method of predicting hydrodynamic coefficients is proposed by using wavelet neural networks. The structure of the network is confirmed, and by using uniform design method, a series of submarine hull model is designed as the sample for network to study. After checking the availability of CFD method in predicting hydrodynamic coefficients, hydrodynamic coefficients of the sample are calculated. Through the sample study the prediction of hydrodynamic coefficients of submarine hull is completed. The results indicate that hydrodynamic coefficients of submarine hull can be predicted well by using the wavelet neural network with suitable input parameters, study sample and net structure.

关 键 词：潜艇 操纵 水动力 小波神经网络 

分 类 号：U661.33[交通运输工程—船舶及航道工程]