A new kind of particle swarm optimization (PSO) algorithm is proposed to identify the inertia parameters of an onorbit satellite equipped with a class of MultiDOF robot. By diversity maintenance by limiting the definition of particle neighborhood, random reset of inefficient particles and sequential evaluation of particle errors, a better result is achieved in contrast with the classical PSO algorithm. Moreover, it doesn’t require additional fuel consumption, linear momentum measurement nor specific robot path planning. The simulation experiments show that the improved algorithm performs more accurately and efficiently.