The report presents the development of a digital twin based on machine learning to quickly determine the optimal programming to achieve the desired tactical behavior, similar to the behavior of a swarm of several drones. The goal is to create a system with fully autonomous behavior by optimizing the values of the operating parameters embedded in the coupled differential equations of the dynamics of a swarm of drones. Ways to achieve minimization of the "cost-error" function are presented. A discrete element method (DEM) has been developed for the rapid design of a quadcopter of any complex shape, the attachment of engines, and then its exposure to an aggressive environment in order to determine its performance.