The optimization target of traditional light-weight methods is a single part generally, which can not make the whole component achieve the design goal of lightest weight and highest surface figure precision. It can not meet performance indexes due to the strict limitation of the carrying capacity of actuator and mirror's material. A Primary Mirror Segment Assembly ( PMSA ) is designed integrated to reduce the load of the segment actuator and ensure the surface figure and stability of the splicing primary mirror. The points surpport on the mirror back is selected as the support method through the topology optimization, the mass distribution of the back of the mirror and the distribution of the removable part are obtained. The initial structure of the mirror is determined based on back with triangular hole for light-weighting. A multi-axis flexible support structure with slotted beams based on the support method is put forward. Component integration design introduces too many parameter variables, resulting in a large amount of calculation, long iteration time and difficulty in convergence. A combinatorial optimization algorithm based on multi-island genetic and gradient algorithm to solve the problem of too many variables and difficulty in convergence is designed, and it is applied to optimization design of the segment. The combinatorial optimization algorithm's tactics is that the sensitive area of the design space is roughly located by the multi-island genetic algorithm, and the exact positions are optimized by using the gradient optimization algorithm in the area. The tactics makes full use of the advantages of global optimization algorithm in overall design space traversal and gradient optimization algorithm in local area convergence speed faster. It is established that multiple-object optimization model which combines structure mass, the surface figure error under the influence of gravity along Y-axis ( dectction direction ) and the surface figure error under the effect of 5℃ temperature change as objectives. The optimization model is carried out on the initial structure of the primary mirror segment assembly, and the statics of the optimized structure under the influence of gravity along Y-axis ( dectction direction ) and 5℃ temperature change are analyzed. The Root Mean Square (RMS) of the mirror’s surface figure error is controlled within 5 nm, which is tested at last and is far better than design requirement. The segment weights 1.74 kg, and primary frequency of X, Y, Z is more than 400 Hz. Its surface density is as low as 27.32 kg/m³. The surface figure error is test and the primary mirror segment assembly with RMS is as low as 0.019λ (λ=632.8 nm ), which meets the design requirement better than λ/50. The design method of integrated optimization is reasonable, and it provides a guiding for design of mirror assembly of space camera. In order to evaluate the optimization efficiency of the combinatorial algorithm, it is compared with multi-island genetic algorithm.The multi-island genetic algorithm is used to optimize the primary mirror segment assembly.The combinatorial optimization algorithm iterates 394 times, and the multi-island genetic algorithm iterates 601 times. The result shows that combinatorial optimization algorithm based on multi-island genetic and gradient optimization has better solving ability and can effectively improve the optimization efficiency of integral optimization design of the primary mirror segment assembly.