Abstract: This dissertation focus on the optimization method for composite stiffened panel structure. Due to the high number and different type (discrete and continuous) of design variable, hierarchy optimization is introduced in this method.The mass of structure was fixed to be constant, while the locations of reinforced beams, cross-sectional dimensions, number of piles and stacking sequences were optimized to improve the buckling performance of overall structure. Artificial plies were introduced so that the optimization of the other variables can be done without considering the stacking sequence. Then design of factorial experiment was conducted to define the primary and secondary variables that influence the structure buckling load in the remaining variables. Level 1 and Level 2 optimizations were conducted independently to obtain the optimums for two types of design variables. For computing efficiency improvement, according to the type of variables and value range in the process of optimization, radial basis agent model was set up for some variables.Genetic algorithm was implemented in Level 3 optimization to search for the best stacking sequence.Optimization results show that the resultant buckling load is 3.21 times higher than that of the original one.