As the core equipment of offshore floating liquefied natural gas (FLNG) installation system, LNG cryogenic flexible pipeline is an important link in the safe transportation of liquefied natural gas industry chain. It plays the role of link between the preceding and the following in the system. Therefore, the structural design of LNG cryogenic hose is very important, which determines the overall mechanical and safety performance of cryogenic pipeline in the system. In this paper, according to the structural characteristics of LNG cryogenic hose, the multi-layer structure is decoupled into lining layer structure, armored layer structure and non-load-bearing layer structure. Based on neural network (RBF), response surface (RSM) and Kriging modeling methods, the approximated model is established and the accuracy of different models is compared. It is found that the error of RBF modeling method is the least. In the optimization design, based on genetic algorithm, the hierarchical and multi-objective optimization design of three different functional structure layers of multi-layer LNG cryogenic hose is carried out. In the optimization of lining structure, the minimum mass and minimum bending stiffness are the optimization objective. In the optimization of armoring layer structure, the optimization objective is to minimize the mass and tensile stiffness. In the optimization of non-bearing layer structure, the optimization objective is to minimize the mass, bending stiffness and heat transfer rate. And the effectiveness of the optimization design method in this paper is verified.