Dynamic effect web generation for heterogeneous UAV cluster using DQN-based NSGA-II: Methods and applications

Effect web will be an important combat means to achieve accurate, efficient, agile and reliable destruction of enemy targets. The use of Unmanned Aerial Vehicles (UAV) cluster in warfare has become a key element in the battle for military superiority between nations. The construction of UAV cluster effect web is a kind of combinatorial optimization in essence. By selecting the optimal combination in the limited equipment concentration, the whole network can be optimized. Firstly, in order to improve the combinatorial optimization efficiency of UAV cluster effect web, NSGA-II based on deep Q-network (DQN-based NSGA-II) is proposed. This algorithm is used to solve the Multi-Objective Combinatorial Optimization (MOCO) problem in the construction of effect web. Secondly, a dynamic generation method is devised to solve the problem caused by the possible destruction of enemy and our node under the fierce confrontation between the two sides. Finally, the simulation results show that the DQN-based NSGA-II is better than the genetic algorithm with single operator. The comparison experiment shows that the weight of evaluation indexes will have a corresponding influence on the optimization results.