Unmanned aerial vehicle cluster twinborn model derivation method for reinforcement learning training

The invention discloses an unmanned aerial vehicle cluster twinborn model derivation method for reinforcement learning training, which belongs to the field of system modeling and simulation, and can derive the pose of an unmanned aerial vehicle cluster digital twinborn model and increase the quantity and quality of training data so as to improve the performance of a reinforcement learning algorithm in unmanned aerial vehicle cluster control. The method is based on distribution fitting test and comprises the following steps: firstly, establishing hypothesis distribution, obtaining unknown parameters of the hypothesis distribution through a maximum likelihood estimation method, secondly, selecting optimal fitting distribution through goodness-of-fit test, distribution comparison and distribution optimization, and then sampling the hypothesis distribution to obtain expanded simulation model parameters to realize model derivation. And finally, verifying the effectiveness of model derivation in a virtual-real comb