A Structural Optimization Design Method Accelerated by Generative Adversarial Networks

Abstract

The invention discloses a method for accelerating structure optimization design by applying a generative adversarial network. The acceleration method can solve the problem that a traditional SIMP algorithm is high in calculation complexity in structural optimization design, the method comprises two parts of model obtaining and model using, and the main process of model obtaining comprises the steps that 1, a small number (100 sets) of optimization process diagrams are generated in advance through the SIMP method to serve as a training set and a test set; 2, expanding the data set by using a data enhancement technology; 3, using an encoder-decoder to construct a generator; 4, constructing a discriminator by using a deep convolutional network; 5, using the deformed pix2pix model to train a generator-discriminator and display a training result and stores a final training model. When the final model is used for structure optimization design, firstly, a small number of iteration steps are calculated through an SIMP method, an iteration result is input into the final model for calculation, and therefore rapid calculation of the final optimized structure is achieved.

Description

technical field [0001] The invention belongs to the field of artificial intelligence, and in particular relates to a structure optimization design technology accelerated by a generative confrontation network, and a method for reducing computational complexity to the greatest extent on the premise of ensuring correct generation results. Background technique [0002] In order to design an efficient material structure distribution under given load conditions, constraints and performance indicators, researchers at home and abroad apply topology optimization methods to optimize the design; the basic idea of ​​topology optimization is to seek the optimal topology of the structure. It is transformed into the problem of finding the optimal material distribution in a given design area. At present, the topological optimization methods mainly include SIMP algorithm, ESO algorithm, level set method, etc. The calculation amount of these methods depends on the scale of the grid. With the ...

AI Technical Summary

Problems solved by technology

[0002] In order to design an efficient material structure distribution under given load conditions, constraints and performance indicators, researchers at home and abroad use topology optimization methods to optimize the design; the basic idea of ​​​​topology optimization is to find the optimal topology of the structure Transformed into the distribution problem of seeking the optimal material in a given design area, the current topology optimization methods mainly include SIMP algorithm, ESO algorithm, level set method, etc. The calculation amount of these methods depends on the size of the grid. The continuous increase of the calculation amount increases exponentially, making it take a long time to obtain the optimal design results

Images