Research on a Model for the Structural Optimization of Freeform Surfaces Based on Deep Learning

Curved surface shell structures have attracted much attention due to their excellent mechanical properties and broad prospects for engineering applications. However, their optimal design faces challenges such as complex multi-objective nonlinear problem solving and low computational efficiency. This paper proposes an optimization model for freeform surface shell structures based on deep learning. By constructing a complex mapping relationship between geometric parameters, stress distribution, mass and displacement, the efficient optimization of freeform surface shell structures is realized. In this paper, non-uniform rational B-splines (NURBS) are used to generate diverse initial geometry samples, and finite element analysis (FEA) is combined to obtain high-quality training data. A fully connected neural network (FCNN) is trained to predict the performance of the optimized structure. The model optimizes both mass and displacement performance through a comprehensive loss function, and its prediction accuracy and optimization effect are verified by simulation and experimental verification. The results show that compared with traditional methods, the deep learning-based optimization model significantly improves computational efficiency and optimization performance. After optimization, the average mass of the shell is reduced by 34.2 %, the maximum displacement is reduced by 28.7%, and the stress distribution is more uniform. The method in this paper provides a new technical path for the efficient optimization and design of complex engineering structures.