Abstract: In the clamping process of thin-walled parts, different parameters such as positioning layout, clamping layout and clamping sequence lead to different degrees of deformation of thin-walled parts. Most of existing research by the method of node to the objective function is minimum deformation, deformation degree and ignores all directions of different regions of the stress distribution of the clamping effect, puts forward the minimum strain energy as the target air thin-walled clamping scheme optimization, the most hours, when the whole strain energy thin-walled deformation degree is small, the stress distribution is more uniform.Latin hypercube sampling and finite element software are used to generate sample data.A BP neural network prediction model is established to minimize the global strain energy, and the location of positioning elements is the decision variable. In the clamping process of thin-walled parts, positioning and clamping are interrelated processes. Based on the optimization of positioning layout, the nonlinear mapping relationship between clamping sequence, clamping position and overall strain energy is constructed. The established mapping relationship is iteratively optimized to determine the optimal clamping sequence and clamping position. The correctness and effectiveness of the proposed method are verified by an application example of a curved thin-walled piece.