Research On UAV Maneuvering Decision Method Based On Game Theory In Complex Air Combat

With the development of UAV technology,new requirements for the air combat maneuver of UAV are put forward.Based on game theory and combined with intelligent algorithm,this paper studies the maneuvering decision-making method of UAV in complex air combat.The main research contents are as follows:Aiming at the maneuver confrontation problem of unmanned aerial vehicles in fuzzy information environment,a maneuvering decision-making method based on intuitionistic fuzzy game is proposed.Firstly,establishing an intuitionistic fuzzy payment matrix to evaluate the maneuver strategy of both sides with the application of the multi-attribute intuitionistic fuzzy evaluation method.Second,a maneuvering anti-intuitionistic fuzzy game model that satisfies the intuitionistic fuzzy full-order relationship is proposed,and the existence proof and solution method of the solution of this game model are given.Then,the individual evolutionary algorithm,which is used to solve the model is improved by using individual control parameters and genetic algebra adaptive strategy.Finally,a new idea for solving the air combat decision problem in the fuzzy environment is offered by the simulation verification model and the rationality and effectiveness of the algorithm.Aiming at the problem of active defense of UAV under unknown system model,a data-driven adaptive dynamic programming algorithm(ADP)is proposed to solve the optimal control strategy.First,a linear differential game model for active defense problems is established.Secondly,the solution to the model is transformed into couple game algebraic Riccati equations(CGARE).Then,with the strategy iterative method,a data-driven adaptive dynamic programming algorithm is proposed to solve the approximate solution of CGARE,and the convergence of the algorithm is proved.The simulation verifies the correctness of the algorithm and the model.The results show that under the condition of the unknown attacker's maneuver time constant,the defender can always intercept the attacker's attack target.Aiming at the problem of UAVS confrontation in large-scale UAV swarms environment,a UAV swarms confrontation method based on the mean field game model is proposed.First,the complex interaction between the UAV swarms is described by a simple potential field superposition principle.Then,the mean field theory is introduced to establish the mean field game model of the UAV swarms.The UAV's choice of strategy will depend on its own state and the overall state distribution of the swarms.Finally,based on the idea of reinforcement learning,the MFSG-Q learning algorithm is proposed for the learning of UAV swarms confrontation.The simulation results show that the UAV swarms can show some complex clusters based on simple maneuvering strategies,and the MFSG-Q learning algorithm has higher success rate than the single agent learning algorithm.