SCIENCE CHINA Information Sciences, Volume 61 , Issue 11 : 119204(2018) https://doi.org/10.1007/s11432-018-9463-x

Missile aerodynamic design using reinforcement learning and transfer learning

More info
  • ReceivedFeb 28, 2018
  • AcceptedMay 3, 2018
  • PublishedSep 17, 2018


There is no abstract available for this article.


This work was supported by National Natural Science Foundation of China (Grant No. 61603210) and Aeronautical Science Foundation of China (Grant No. 20160758001).


Appendix A.


[1] Van Nguyen N, Tyan M, Jin S. Adaptive multifidelity constraints method for efficient multidisciplinary missile design framework. J Spacecr Rockets, 2016, 53: 184-194 CrossRef ADS Google Scholar

[2] Ni A, Zhang Y F, Chen H X. An improvement to NSGA-II algorithm and its application in optimization design of multi-element airfoil. Acta Aerodynamica Sin, 2014, 32: 252--257. Google Scholar

[3] Giacché D, Xu L, Coupland J. Optimization of bypass outlet guide vane for low interaction noise. AIAA J, 2014, 52: 1145-1158 CrossRef ADS Google Scholar

[4] Kutz J N. Deep learning in fluid dynamics. J Fluid Mech, 2017, 814: 1-4 CrossRef ADS Google Scholar

[5] Zhang W D, Wang Y B, Liu Y. Aerodynamic study of theater ballistic missile target. Aerosp Sci Tech, 2013, 24: 221-225 CrossRef Google Scholar

[6] Lillicrap T P, Hunt J J, Pritzel A, et al. Continuous control with deep reinforcement learning. 2015,. arXiv Google Scholar

[7] Oquab M, Bottou L, Laptev I, et al. Learning and transferring mid-level image representations using convolutional neural networks. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Columbus, 2014. 1717--1724. Google Scholar

[8] Biancolini M E, Costa E, Cella U. Glider fuselage-wing junction optimization using CFD and RBF mesh morphing. Aircraft Eng Aerosp Tech, 2016, 88: 740-752 CrossRef Google Scholar

  • Figure 1

    (a) Design parameters of the missile; (b) architecture of the deep learning approach for aerodynamic design; protect łinebreak (c) performance comparison of NCGA, NSGA-II, MOPSO, DDPG, and SL-DDPG in the hybrid environment.