Application of artificial intelligence in magnetic performance prediction of sintered Nd-Fe-B magnets

This study aims to accurately predict the magnetic properties of sintered NdFeB magnets using artificial intelligence technology. Specifically, a BP neural network was adopted to build a prediction model. In the process of model construction, we specifically considered the complexity of sintered NdFeB magnet compositions and the influence of processing on the residual magnetization (Br) and coercivity (Hcj) magnetic properties. Therefore, we selected element compositions and processing parameters as feature variables. Through optimization of model hyperparameters, we successfully obtained a BP neural network prediction model with high prediction accuracy. The research results demonstrate that the proposed prediction model can effectively predict the magnetic properties of sintered NdFeB magnets, providing important guidance for process optimization and magnet composition design.