![]() In this paper, an autonomous UCAV air combat maneuvering decision method based on LSHADE-TSO optimization in a model predictive control framework is proposed, along with enemy trajectory prediction. Empirically, we demonstrate that the proposed method outperforms state-of-the-art algorithms with an at least 67.4% relative winning rate in a high-fidelity air combat simulation environment. After this particular relationship simplification process, representative collaboration tactics emerged via subsequent intention communication and joint decision making mechanisms. Specifically, TRISONIC creates a Graph Neural Networks (GNNs) and expert knowledge composite approach to jointly reason out the key relationships into an Abstract Relationship Graph (ARG). In view of this, a novel Multi-Agent Deep Reinforcement Learning (MADRL) and expert knowledge hybrid algorithm named Transitive RelatIonShip graph reasOing for autoNomous aIr combat Collaboration (TRISONIC) is proposed, which solves the large-scale autonomous air combat problem with complex relationships. ![]() However, previous studies have encountered significant difficulties in dissecting large-scale air confrontations with such complex relationships. These relationships often present numerous, multi-relational, and high-order characteristics. Large-scale air combat is accompanied by complex relationships among the participants, e.g., siege, support. The simulation results show that this algorithm can carry out a multi-aircraft air combat confrontation drill, form new tactical decisions in the drill process, and provide new ideas for multi-UCAV air combat. Finally, the idea of centralized training and distributed implementation is adopted to improve the decision-making ability of the unmanned combat aircraft and improve the training efficiency of the algorithm. Secondly, to overcome the sparse return problem of traditional reinforcement learning, according to the angle, speed, altitude, distance of the unmanned combat aircraft, and the damage of the missile attack area, this paper designs a comprehensive reward function. ![]() In order to simulate the real beyond-visual-range air combat, the missile attack area model is established, and the probability of damage occurring is given according to both the enemy and us. Firstly, the model of the unmanned combat aircraft is established on the simulation platform, and the corresponding maneuver library is designed. To solve the problems of autonomous decision making and the cooperative operation of multiple unmanned combat aerial vehicles (UCAVs) in beyond-visual-range air combat, this paper proposes an air combat decision-making method that is based on a multi-agent proximal policy optimization (MAPPO) algorithm.
0 Comments
Leave a Reply. |