Simulation Research On Air Combat Confrontation Strategy Of Fixed-wing Aircraft Based On Hierarchical Contro

Unmanned aerial vehicles(UAVs)play an important strategic role in modern warfare.Faced with complex combat environments and multi task requirements,the air combat strategy of autonomous and collaborative unmanned combat is becoming increasingly important.This thesis conducts in-depth research on key issues in unmanned aerial vehicle collaborative countermeasures.Firstly,this thesis uses PID control algorithm combined with fixed wing flight characteristics to design a flight control module.This plan uses a PID algorithm to control the aileron lever of the aircraft so that the target point of the aircraft is located directly above the nose.At the same time,PID control is used to align the nose with the target,and PID control is combined with the tail rudder to make the aircraft turn.After multiple mechanism optimization,this scheme can enable the aircraft to achieve nose alignment with the target waypoint under different conditions and accurately reach the target point in testing.Based on this control scheme,this thesis also designed a one-on-one gun fighting scheme and a drum maneuvering scheme.Secondly,this thesis studies the flight control strategy of fixed wing aircraft close combat based on Deep reinforcement learning.This thesis uses the SAC algorithm as the experimental algorithm,and uses the nose vector,speed vector,sky vector,enemy aircraft relative direction vector,enemy aircraft relative speed vector and other states during the flight of the aircraft as the model input.The algorithm model directly outputs the aircraft lever position.In order to improve the combat performance and robustness of the algorithm,this thesis uses a robust enhancement training scheme.And during training,the initial distance,attitude,speed,altitude,and other parameters of both fighter planes are randomized to adapt the model to various complex combat situations.Finally,this thesis studies the multi aircraft air combat countermeasure algorithm based on Multi-Agent Deep Reinforcement Learning as the central control strategy of air combat.In order to reduce the algorithm input dimension,reduce the algorithm exploration space,and improve the data sampling efficiency,this thesis abstracts the aircraft involved in air combat into a mathematical statistical model according to their situation awareness and combat capabilities,and then uses the MAPPO algorithm to explore the countermeasures in the model.Subsequently,a multi aircraft air combat countermeasure strategy was designed based on a hierarchical control method,and the strategy proposed in this thesis was tested in high-precision air combat simulation software.It can be seen that the air combat confrontation strategy presented in this article demonstrates the synergy between aircraft,thereby winning the victory in air combat.This thesis is based on a hierarchical control structure,extending from the flight control design and adversarial performance improvement of a single fixed wing unmanned aerial vehicle to intelligent control algorithms for multi aircraft air combat.The flight control algorithm designed in this thesis achieves high-precision position control of unmanned aerial vehicles and improves the efficiency of upper level algorithm design; The flight control strategy of fixed wing aircraft in close combat based on Deep reinforcement learning achieves high aggressiveness and robustness; Multi aircraft air combat strategy based on multi-agent Deep reinforcement learning realizes autonomous cooperative confrontation of UAV,and integrates all modules of air combat strategy.