摘要
为用于船体型线设计中船体曲面重建和光顺,基于多分辨率小波网络原理,引入三次B样条函数作为小波基函数,三次B样条基函数作为尺度函数,利用小波网络的权值和激励函数的线性组合,构造了新的二维多分辨率B样条小波网络(multi-resolution B-splinewavelet net work,MRBWN).将曲面分解为相互正交的u、v方向函数,用该网络来拟合自由曲面;并利用梯度下降法对网络进行训练.最后对船体曲面进行了光顺,实例证明了该方法的有效性和正确性.
To reconstruct and smooth the ship surface in ship form design,based on the multi-resolution wavelet network （MRWN） principle,cubic B-spline function was introduced as the wavelet function and cubic B-spline basis function as the scale function and the linear combination of the weights and the activation function of wavelet network was used,a novel two-dimensional multi-resolution B-spline wavelet network （MRBWN） was constructed. The surface was decomposed into orthogonal u,v direction function,and new network was used to approximate it. The gradient algorithm was adopted to train the network. The smooth ship surface is obtained in the experiment of hull surface,which shows the feasibility and validity of the approach.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2010年第2期215-219,共5页
Journal of Dalian University of Technology
关键词
多分辨率B样条小波网络
船体曲面
光顺
重建
multi-resolution B-spline wavelet network
ship surfaces
smoothing
reconstruction
