Deep learning driven real time topology optimisation based on initial stress learning

Topology optimisation can facilitate engineers in proposing efficient and novel conceptual design schemes, but the traditional FEM based optimization demands significant computing power makes real time impossible. Based on convolutional neural network (CNN) method, a new deep learning approximate algorithm for topology is proposed. The learns from initial stress (LIS), which defined as major principal matrix obtained finite element analysis first iteration of classical optimisation. structure used to replace load cases boundary conditions independent variables, produce topological prediction results with high accuracy relatively small number samples. Compared method similar result without repeated iterations. A classic short cantilever problem was an example, optimized predicted successfully by established algorithm. By comparing structural it discovered that two are highly approximate, verifies validity Furthermore, evaluation proposed evaluate effects using different methods select samples performance topology, were promising concluded end.