Abstract: To address the poor adaptability to dynamic scenarios in maritime underwater moving target positioning, this study enhances the measurement accuracy and effective range of long-baseline (LBL) sonar arrays. An improved particle swarm optimization algorithm—GA-PSO—is proposed, integrating the crossover operator of the genetic algorithm (GA) with non-dominated solution selection based on an elite archive. A coordinated station layout model featuring “one dynamic measurement station plus five optimizable seabed-fixed stations” is constructed. Dual constraints—namely, regional boundary constraints and minimum base station spacing constraints—are incorporated, along with a linearly decreasing inertia weight (LDIW) strategy, to ensure both optimization feasibility and computational efficiency. Through iterative optimization, the average regional geometric dilution of precision (GDOP) is reduced from 5.1858 to 3.2360, while the regional coverage rate increases from 70.40% to 78.20%. This algorithm effectively mitigates key limitations of the standard particle swarm optimization (PSO) algorithm, including premature convergence to local optima and loss of population diversity. It provides an efficient solution for optimal base station layout design in underwater moving target positioning and supports the execution of marine survey missions.