Abstract.In this paper, a SWATH USV is taken as the research object, a series of static roll attenuation tests are carried out under different draft and different initial roll angles. By changing the composition of the restoring moment and damping, 4 mathematical models of roll attenuation motion system identification are established. And based on system identification theory and a variety of optimization algorithms programming ideas, C# language is used to rewrite the system identification software including genetic algorithm and particle swarm algorithm. The identification and analysis of the rolling data of unmanned craft are carried out, and a better mathematical model of rolling motion is obtained by identification and calculation, and the variation of identification parameters under different draught and inclination angles is compared. The relative error between the experimental value and the forecast value predicted by the software, in good agreement with experimental values, proved the reliability of the system identification method, can forecast the future of unmanned navigation parameters and motion attitude.