An Intelligent Penetration Guidance Law Based on DDPG for Hypersonic Vehicle

A novel intelligent guidance law based on Deep Deterministic Policy Gradient (DDPG) is devised in this paper to deal with problems of penetration for hypersonic boost glide vehicle (HBGV). Firstly, an agent based on DDPG algorithm is designed by setting a model of Markov decision process. Interacting with environment, the agent is trained to produce continuous overload command to avoid interception by one interceptor and ensure successful penetration for HBGV. Compared with traditional penetration guidance law, the new guidance law can intelligently deal with the complex offensive and defensive game problem. In addition, the designed reward of agent avoids excessive overloading which can save energy of HBGV. Finally, simulations are carried out, which prove that the guidance law enables HBGV to avoid interception by interceptor. The results in different test scenarios also illustrate that the intelligent guidance law has great generalization ability which can meet need of penetration in other offensive and defensive situations.