In order to solve the problems of poor compatibility, weak environmental adaptability and low discharge efficiency of traditional liquid metal diaphragm tank, the research on multi-objective parameter regulation of liquid metal structure optimization design was put forward. In this paper, the structural design of spherical metal diaphragm with a certain volume is carried out, and the arc length method is used for numerical simulation. By using the second generation genetic algorithm, the multi-objective optimization design of the spherical liquid metal diaphragm with this configuration was carried out with the objective of reducing the eccentric displacement and increasing the buckling load at the initial position, and the local optimal bottom radius, the fixed-point thickness of the diaphragm and the pre-flanging radius were obtained, which reduced the eccentric displacement of the diaphragm apex and increased the buckling load at the initial position. The experimental results show that the buckling load P_buckle is increased by 1.13% and the eccentric displacement T_xz is decreased by 36.07% compared with that before optimization, and the performance of membrane turnover after optimization is better.