To improve the quality and efficiency in satellite design, the satellite multidi sciplinary design optimization is presented with the obje ct of minimizing total mass considering orbit, payload, power supply and structu re. The paper combs the coupling among the four subsystems and establishes the d etailed and practical disciplinary analysis model. The optimization strategy is proposed based on support vector regresstion(SVR) method. The strateg y is applied to multidisciplnary design optimization (MDO) of the marine satelli te. The studied satellite takes Ocean 1 satellite (HY-1) as a reference, and the total mass of the satellite is reduced by 14.1% after optimization compared with the initial value . Th e optimization result proves the rationality of the satellite′s MDO model and t he effectiveness of the optimization strategy. It is of vital importance to the future work of the application of SVR metamodel in satellite MDO design.