A YOLO-based Deep Learning Approach for Vibration-based Rotating Shaft Imbalance Detection

One of the prevailing causes of vibrations in machines is rotor imbalance. Rotor balancing can be used to fix the majority of rotating machinery issues. When it comes to high-speed running equipment, even a slight imbalance can lead to serious issues and decrease the operational efficiency of rotating machinery. This work proposed a deep learning approach for the detection of binary and multiclass imbalance in rotating shafts. A YOLOv11 model-based approach is developed to detect imbalance and identify unbalanced rotor positions. To precisely identify unbalanced positions, this method trains the YOLOv11 model using numerous sets of measured response data and simulated data from unbalanced rotor positions. Analyses of two scenarios were conducted: (i) binary classification of balanced vs. imbalanced rotors and (ii) multiclass classification of the level of imbalance. The proposed model achieved 99.77% and 96.74% testing accuracy for the binary and multiclass classification scenarios, respectively. The findings imply that both binary and multiclass classification problems can be successfully identified by the suggested deep learning architecture. This study offers a reliable framework for identifying rotating equipment shaft imbalance and could be applied in industrial applications as a real-time problem detection technique.