A data-driven modelling and optimization framework for variable-thickness integrally stiffened shells

Abstract

The integrally stiffened shells possess the advantages of high specific strength, high specific rigidity and excellent sealing performance, which have been widely used in the load-carrying structure of next-generation manned spacecraft and cargo spacecraft. However, the automatic modelling and optimization of the variable-thickness (VT) integrally stiffened shell are challenging due to the diverse stiffener configurations and the VT skin at multiple corners. To address this issue, a novel data-driven modelling and optimization framework is proposed for the VT integrally stiffened shell in this paper. Firstly, a novel mesh deformation method is proposed for modelling the VT integrally stiffened shell combining the radial basis function (RBF) surrogate model and nonuniform rational B-splines (NURBS). By means of the RBF surrogate model, the mapping relationship between the background mesh domain of curved shell and the target mesh domain of flat plate is trained, and then the equal-thickness (ET) stiffened flat plate is transformed into the VT stiffened flat plate by moving coordinates of nodes. In order to realize the accurate description of the VT stiffened flat plate, NURBS method is used to describe the coordinates of nodes. Finally, the VT stiffened flat plate is transformed into the VT integrally stiffened shell by the mapping relationship. In order to verify the accuracy of the proposed method, a skin-stiffener verticality detection method is proposed for stiffened curved shells. Moreover, based on the deep neural network (DNN) method, a data-driven stiffener layout optimization method is established to minimize the structural weight of the VT integrally stiffened shell. To illustrate the effectiveness of the proposed method, an example of the integrally stiffened shell under internal pressure is carried out. The average value of the detection results of skin-stiffener verticality for the integrally stiffened shell under internal pressure is less than 0.29°, indicating the high accuracy of the proposed modelling method. Compared with the initial result and the optimal result of the equal-thickness integrally stiffened shell, the optimal result of the VT integrally stiffened shell achieves a structural weight reduction of 14.77% and 10.31%, respectively. Optimization results indicate the effectiveness of the proposed optimization framework and the advantage of VT integrally stiffened shells in lightweight design.