This academic paper presents a new framework for predicting temperature fields in real-time during metal additive manufacturing (AM) processes. The authors argue that accurate temperature prediction is essential to prevent overheating, adjust process parameters, and ensure process stability. The proposed framework uses a physics-informed neural network, which includes a physics-informed input and loss function, and a Convolutional Long Short-Term Memory (ConvLSTM) architecture. The model uses real-time temperature data to predict future 2D temperature fields across different geometries, deposition patterns, and process parameters. The framework was tested in two scenarios and exhibited errors below 3% and 1%, demonstrating its potential application across diverse scenarios.
Publication date: 5 Jan 2024
Project Page: not provided
Paper: https://arxiv.org/pdf/2401.02403
