Research on Optimal Cooperative Attack Strategy of Multiple UAVs

Abstract

Aiming at the problem of the optimal coordinated attack strategy of multi-UAV system against multiple moving targets in a known environment, a PMMAXQ layered reinforcement learning algorithm applied to multi-UAV cooperative attack is proposed. Firstly, the problem of multi-UAV coordinated strike, the successful strike conditions and the movement strategy of moving targets are described. Then the moving targets are allocated rationally through the 0-1 planning method. Finally, the MAXQ algorithm is improved by the Bayesian probability statistical formula. The simulation results show that, compared with the MAXQ algorithm, the PMMAXQ algorithm converges faster and the average cumulative reward value that the UAV can obtain is higher.