High-Dimensional Multiobjective Optimization Design for Magnetic Stealth of Underwater Vehicle Based on Improved MSOPS Algorithm

The underwater vehicle will generate multiple magnetic features such as peak, gradient, rate of change, etc. in multiple directions. In order to guarantee magnetic stealth performance and improve survivability, the low-dimensional optimization can not seek the true Pareto front of each objective, this paper proposes to carry out high-dimensional multi-objective optimization and control of the demagnetization system of the underwater vehicle, and introduce the high-dimensional algorithms to generate the Pareto front on the hyper-plane to provide the decision makers with choices. This paper proposes the IMSOPS that combines the parallel search idea of multiple single-objectives with a genetic algorithm, utilizes the ability of the genetic algorithm to retain the strong dominant individuals, adds initial boundaries, and realizes the high-dimensional multi-objective algorithm with fast convergence. And non-dominated sorting and congestion operations are added to further improve the diversity of the algorithm. Using underwater vehicle simulation experiments and comparing with other high-dimensional algorithms, the improved algorithm is analyzed for its superiority in convergence speed, convergence, diversity, and comprehensiveness, etc. Finally, the effectiveness of the IMSOPS algorithm implemented on the magnetic stealth technology of underwater vehicles is verified through the experiment of a real scaled-down model.