Modelling and simulation of distributed target allocation based on improved NSGA-II in multi-UUV collaborative attack

In multi-UUV collaborative attack, a reasonable weapon-target allocation scheme is very important to improve the attack efficiency. The quality and computation time of the solution are the key indices to evaluate the performance of the weapon-target algorithm. At present, the swarm intelligence optimization algorithm represented by genetic algorithm is often used to solve the weapon-target allocation problem, but this method usually adopts the centralized programming solution. Although it can achieve better solution quality, the solution process has disadvantages such as large computation and long time, which limits its application in practice. To solve these problems, this paper proposes a distributed improved non-dominated sorting genetic algorithm-II (DINSGA-II), which includes adaptive crossover and mutation strategies and distributed elite interaction strategies. In addition, we have developed a multi-UUV coordinated attack simulation platform based on XSimStudio using component-based modelling techniques. Finally, the proposed distributed target allocation algorithm is compared with the traditional centralized NSGA-II algorithm on the developed simulation platform. The results show that both algorithms can find a reasonable allocation scheme, but in terms of computational speed, the running time of the DINSGA-II algorithm is about 12 times faster than that of the traditional centralized NSGA-II algorithm with the same problem size.