摘要
为解决SINS动基座传递对准的快速精确问题,将混合优化的RBF神经网络应用于此。首先运用递阶遗传算法优化RBF神经网络的拓扑结构,并对网络其余参数进行全局粗调;在此基础上运用H∞滤波算法对网络其余参数进行在线自适应精调。其计算机仿真结果与扩展卡尔曼滤波比较表明:该算法在精度、实时性方面与扩展卡尔曼滤波相比提高了将近50%。
The hybrid optimal RBF neural network is precision of SINS moving base transfer alignment. At applied to resolve the problem about celerity and first, Optimizing the topological frame of RBF neural networks and adjusting sketchily the other parameters in the picture of the whole in the application of Hierarchical Genetic Algorithm（HGA）, on the basis of which , the other parameters are regulated accurately online adaptive by H∞ filter algorithm. It shows that this algorithm increases about 50% in precision and real-time by comparing computer simulation with Extend Kalman Filtering（EKF）.
出处
《火力与指挥控制》
CSCD
北大核心
2010年第1期104-107,135,共5页
Fire Control & Command Control
基金
国防科工委基金(J1300B004)
南京理工大学科研发展基金(XKF05031)
弹道国防科技重点实验室基金资助项目(514530601052S3301)
