Optimization method for clamping layout of refractory thin-wall parts based on IAGA-Elman

Abstract

Abstract Thin-walled parts made of refractory materials are widely used in the sintering field of lithium battery powder materials. To solve the problem of deformation and fracture caused by the low stiffness of refractory thin-wall parts, a sandwich-layout proxy optimization model based on the IAGA-Elman neural network and sub-optimization mechanism was proposed. Based on ABAQUS, finite element simulation models under different clamping layouts were obtained. The sample set required for neural network training was established, a sub-optimization mechanism with MSP+EI combination plus point criterion as the target task was constructed, and the IAGA algorithm was used to optimize the Elman neural network for optimal parameter seeking. The deep nonlinear mapping relationship between clamping layout parameters and clamping deformation is explored by proxy optimization model. The experimental results show that the proposed method can predict the clamping deformation of thin-walled parts with fewer simulations and higher fitting accuracy, and can provide a basis for the optimal design of the clamping layout of refractory thin-walled parts.