摘要
提出了一种基于径向基函数 ( RBF)神经网络的机载数字地图压缩方法 ,并给出了网络训练的具体算法。与一般 RBF网络的构造方法不同 ,该网络结构的所有参数是通过学习方式同时获得的 ,因此其泛化性大大增强。数字仿真表明该方法具有自动适应地形、参数配置合理、机载计算量小等特点。
The compression and reconstruction of digital terrain data is vital to aerial combat vehicle in carrying out air to ground attack. We present a RBF neural network model and the algorithm of network learning, which are suitable for the compression of onboard digital map. Section 1 explains the RBF neural network. Subsection 1.2 explains the RBF neural network model we establish to make the onboard reconstruction equipment as simple as possible. Subsection 1.3 is very important; it explains in much detail how to constitute all the parameters of an RBF neural network synchronously via learning. Subsection 1.4 explains that only fewer representative points are needed for the storage of smooth areas. The explanation in section 1 leads to a significant compression of the data as well as a highly reduced onboard memory requirement. Section 2 gives a numerical simulation example; simulation results, given in Figs.5, 6,and 7,show preliminarily that our method, based on RBF neural network, is satisfactory for compression of onboard digital map.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2003年第3期325-328,共4页
Journal of Northwestern Polytechnical University
基金
航空基金 (2 0 0 0 CB0 6 0 1)
国防"973"项目 (2 0 0 1HS0 6 37)
关键词
数字地图
径向基函数
空对地攻击
机载
onboard digital map,radius basis function(RBF),air to ground attack
