ABSTRACT
In this paper, a stress-based topology optimization method using deep convolution neural network (CNN) is proposed, which can give a real-time prediction of the optimized structure without any iteration. The popular U-Net network structure is adopted to improve the edge extraction ability of the neural network. To train the network, the method of moving asymptotes (MMA) is employed to generate a dataset with random loading conditions and volume constraints. The efficiency and accuracy of the proposed method are compared with the conventional method, and the performance of the proposed method is evaluated. The results show that the proposed method can significantly reduce the computational cost with little sacrifice on the performance of the design scheme. The proposed method has great potential and broad application prospects for topology optimization in the practice of structural design in the future.
