A Novel Dynamic Hypergraph Attention Framework for the Prediction of Removal Rate in Chemical Mechanical Polishing Process

With the rapid advancement of machine learning technology, numerous algorithms have been employed to enhance the prediction accuracy of material removal rate (MRR). In this paper, a novel prediction framework is introduced based on a dynamic hypergraph attention mechanism, providing accurate MRR predictions. This approach utilizes the multi-level attention mechanism within the hypergraph architecture to effectively capture the relationships between nodes and hyperedges in the polishing machine, enabling precise MRR predictions, while maintaining a low data processing cost. Experimental results show that our method achieved an RMSE of 1.42 and an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.963, outperforming current state-of-the-art models.