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

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.