Composite ultra-thin boom can be folded elastically. Moreover, such booms are able to self-deploy by releasing stored strain energy, which can be applied in deployable antenna, solar sail, and optical telescopes. Surrogate models for imperfection-sensitive quantities of interest and multi-objective optimization are developed for the design of a new N-shape cross-section composite ultra-thin deployable boom. The proposed optimal design method integrates four general steps: (1) design of experiments, wherein the sampling designs of the N boom are created on the basis of the two-factor five-level full factorial design of experiments method; (2) efficient computational analyses of each design sample, wherein the post-buckling behavior of the N boom are analyzed under three different axial directions using nonlinear finite element ABAQUS/Explicit solver; (3) establishing the surrogate models of bending stiffness around the x-and y-axes and torsional stiffness around the z-axis by response surface method (RSM); (4) performing the multi-objective optimization design using modified non-dominated sorting genetic algorithm to realize the optimal design. The bending stiffness around the x-and y-axes and the torsional stiffness around the z-axis are set as the objectives, mass is set as the constraint, and the bonded web height and the central angle of the middle tape spring of the N boom are set as the variables. The typical surrogate modeling method can be applied to different problems in structural and material design.