In a thesis entitled "Deep Learning Based Stress Prediction for Bottom-Up Stereo-lithography (SLA) 3D Printing Process," a University at Buffalo student named Aditya Pramod Khadlikar describes a method of predicting stress distribution of parts in SLA 3D printing using a deep learning framework. The framework consists of a new 3D model database that captures a variety of geometric features that can be found in real 3D parts as well as "FE simulation on the 3D models present in the database that is used to create inputs and corresponding labels (outputs) to train the DL network."
Multiple samples were tested using CNN.Several parts with similar cross sections on a particular layer are examined to determine the stress distribution on a particular layer. Khadlikar and his colleagues found that different parts of a particular layer that had the same cross section had different stress distributions in that layer.
One important conclusion is that CNN is much faster than FEA simulations.The data sets created work effectively, helping to determine parameters such as peak stress and information dependent on the previous layer to determine the distribution of stress over the layer.In general, the deep learning model performs better than the simple neural network model used for pressure prediction.