有限视场角下基于改进比例导引和深度学习的多约束制导律  

Multi-constraint Guidance Law Based on Improved Proportional Navigation and Deep Learning under Limited Field-of-view

作　　者：程妍菲 刘泽石 杜江鹏 程昊宇 CHENG Yanfei;LIU Zeshi;DU Jiangpeng;CHENG Haoyu(Unmanned System Research Institute,Northwestern Polytechnical University,Xi’an 710072,China;Nation Key Laboratory of Unmanned Aerial Vehicle Technology,Northwestern Polytechnical University,Xi’an 710072,China;Integrated Research and Development Platform of Unmanned Aerial Vehicle Technology,Northwestern Polytechnical University,Xi’an 710072,China;Shenyang Aircraft Design and Research Institute,Shenyang 110035,China;Shanghai Electro-Mechanical Engineering Institute,Shanghai 201109,China)

机构地区：[1]西北工业大学无人系统技术研究院,西安710072 [2]西北工业大学无人飞行器技术全国重点实验室,西安710072 [3]西北工业大学无人机技术集成攻关大平台,西安710072 [4]沈阳飞机设计研究所,沈阳110035 [5]上海机电工程研究所,上海210029

出　　处：《无人系统技术》2024年第5期47-53,共7页Unmanned Systems Technology

基　　金：国家自然科学基金(62176214,62101590,62003268,62303380)。

摘　　要：针对考虑气动力模型的导弹精确打击问题,开展了基于改进比例导引和深度学习的制导律研究。首先生成随机训练数据,随后在训练数据基础上基于残差神经网络生成映射网络来对剩余飞行时间误差进行预测,最后将得到的预测结果引入改进比例导引制导律,使得在视场角约束下攻击角度误差和飞行时间误差均能实现有效收敛。仿真结果表明,该方法能够满足制导要求,证明了该方法的有效性。与传统方法相比,提出的基于改进比例导引和深度学习的制导方法在攻击角度误差和飞行时间误差方面表现更优。Targeting the strike issue of missiles under aerodynamic force models,research has been conducted on guidance law based on improved proportional navigation and deep learning.Firstly training data are generated randomly,followed by the creation of a mapping network based on the residual neural network to predict the error in remaining flight time.Then the predicted results are incorporated into the improved proportional navigation guidance law,allowing for effective convergence of both attack angle error and flight time error under the field-of-view angle constraint.Finally simulation results indicate that the guidance requirements are met by this method,demonstrating its effectiveness.Compared to traditional methods,this method exhibits superior performance in terms of attack angle error and flight time error.

关 键 词：制导律 飞行时间控制 剩余飞行时间预测 攻击角度控制 视场角约束 残差神经网络 

分 类 号：TJ765[兵器科学与技术—武器系统与运用工程]