摘要
由于机动伴随防空武器系统应用场景的特殊性,难以准确建立适应多变工况的挠曲变形模型,无法保证主子惯导传递对准的精度。为解决这一问题,采用无数据依赖智能技术,将传递对准数学模型中的挠曲变形模型由神经网络代替,神经网络的连接权系数扩充为传递对准模型的部分未知变量,使用非线性卡尔曼滤波对模型所有变量进行实时估计,从而获得主、子惯导之间的失准角,基于主惯导姿态信息进而完成行进中的高精度传递对准。仿真实验表明,该方法能在很短时间内估计出失准角,完成导弹惯导的高精度初始姿态装订。
Due to the particularity of the application scenarios of maneuver concomitant antiaircraft weapon systems,it is difficult to accurately establish a flexural deformation model that adapts to changing working conditions,and the accuracy of transfer alignment cannot be guaranteed.To solve this problem,a data-free intelligent technology is adopted,and the flexural deformation model in the transfer alignment mathematical model is replaced by a neural network.The connection weight coefficients of the neural network are expanded to part of the unknown variables of the transfer align⁃ment model.The nonlinear Kalman filter is used to estimate all the variables of the model in real time,so as to obtain the misalignment angle between the main and the sub-inertial navigation.Based on the attitude information of the main inertial navigation system,the high-precision speed alignment during the movement is completed.Simulation experiments show that this method can estimate the misalignment angle in a short time and complete the high-precision initial attitude binding of the mis⁃sile inertial navigation system.
作者
何荧
万佳庆
韦姗姗
胡晓强
HE Ying;WAN Jia-qing;WEI Shan-shan;HU Xiao-qiang(Jiangnan Electromechanical Design Institute,Guizhou Guiyang 550009,China;Guizhou Police College,Guizhou Guiyang 550009,China;Wenzhou University,College of Electrical and Electronic Engineering,Zhejiang Wenzhou 325035,China)
出处
《现代防御技术》
北大核心
2022年第6期50-58,共9页
Modern Defence Technology
关键词
无数据依赖
传递对准
神经网络
挠曲变形
姿态匹配
比力匹配
data-free
transfer alignment
neural network
flexural deformation
attitude matching
specific force matching