Subdivision load identification of ball mill based on multi-domain feature extraction and UMAP-BOXGBoost.

Multi-residual unit fusion and Wasserstein distance-based deep transfer learning for mill load recognition.