Abstract:  In order to reduce the influence of machining and assembly errors on the accuracy of antenna truss，and solve the optimization problems with random variables，the influence of assembly error of the normal distribution on the accuracy was considered.And the equilibrium state of truss antenna unit was solved by the principle of minimum potential energy.The influence of the change of the bar on the surface accuracy was analyzed by the Monte Carlo method，and the average accuracy under the random error was obtained.For improving the computational efficiency，an approximate model of the truss antenna tetrahedron unit with the bar length tolerance as the design variable was established，with the truss surface accuracy and overall processing cost as the objective function.Tolerance optimization design was carried out through the multiobjective optimization method，which obtained multigroup tolerance design scheme with the approximate model error less than 8%.In addition，the minimum average accuracy is 0.013mm，and the analysis result shows that the influence of the web rod is significantly greater than that of the bottom rod.

Key words: assembly tolerance, truss antenna, multi-objective optimization, surface accuracy, radial basis function