摘要
将前向神经网络用于捷联惯性导航系统(SINS)的对准问题。首先,运用递阶遗传算法(HGA)优化神经网络(NNW)的拓扑结构,并对网络其余参数进行全局粗调;然后运用H滤波算法对具有最优结构的神经网络的其余参数在线自适应精调,并对这一过程与常规算法进行了计算机仿真比较。仿真结果表明:该算法能根据实际问题自适应确定网络结构,而且精度、实时性与常规方法相仿。
According to the characteristic of SINS, the paper presents a feedforward neural networks in which Hierarchical Genetic Algorithm(HGA) optimizes the networks topological frame and adjusts the other parameters globally, and then makes more accurate adjustment for the parameters with optimal networks frame by H filter algorithm. Simulation results show that the hybrid algorithm can adaptively determine the network structure, and has similar precision and real-time performance to that of former systems.
出处
《中国惯性技术学报》
EI
CSCD
2004年第1期5-9,共5页
Journal of Chinese Inertial Technology
