| ABSTRACT |
PD SENSE: AI-DRIVEN PARTIAL DISCHARGE RECOGNITION IN HIGH VOLTAGE ROTATING EQUIPMENT PD Sense discloses an AI-driven system for automated recognition of partial discharge in high-voltage rotating equipment. The system analyzes Phase-Resolved Partial Discharge (PRPD) images using a trained deep learning model based on a convolutional neural network incorporating an attention mechanism to enhance extraction of discriminative spatial and channel features. A graphical user interface enables guided selection of trained models, loading of PRPD images, execution of automated classification, and display of predicted discharge types with associated confidence values. The system further generates an explainable visualization that highlights regions of the PRPD image contributing to the recognition outcome, thereby improving transparency and interpretability of the diagnostic results. By integrating attentionenhanced artificial intelligence, explainable visualization, and an interactive software interface, the disclosed system provides a non-intrusive and efficient solution for insulation condition assessment, supporting early detection of degradation, improved diagnostic consistency, and enhanced reliability of high-voltage rotating equipment.
|
