论文部分内容阅读
本文采用仅有隐层非线性神经元的ＢＰ神经网络拟合船舶型线，避免了样条函数对某些特殊型线拟合的不足，并对ＢＰ（反向传播）算法进行了有益的了探讨，最终给出了拟合某船横剖线的算例．证明了本方法是可行的．
In this paper, a BP neural network with only hidden-layer nonlinear neurons is used to fit ship profiles, which avoids the shortcomings of the spline function fitting for some special profiles and makes a good use of the BP (backpropagation) algorithm Finally, an example of fitting a ship’s cross section is given. It is proved that this method is feasible.