Improved Classification of Partial Discharge in Rotating Machines via Feature Extraction, Selection and Integrated Random Forest

In modern industrial systems, the reliable operation of rotating machinery is critical to minimizing downtime and maintenance costs. Insulation degradation due to partial discharge (PD) events is one of the primary causes of failure in such machinery. Early detection of insulation defects is essential to prevent catastrophic failures, but accurately classifying these defects remains a challenging task due to the complexity and variability of PRPD patterns. To address this issue, this paper proposes a novel approach that combines the Random Forest (RF) algorithm with the Gray Level Co-occurrence Matrix (GLCM) for feature extraction and Grey Wolf Optimization (GWO) for feature selection. By reducing the feature space and focusing on the most relevant features, the proposed model significantly improves classification accuracy and efficiency. Experimental results show that the RF+GLCM+GWO model outperforms the RF+GLCM model, achieving substantial improvements in overall accuracy, macro precision, macro recall, macro F1-score, and macro specificity. This integrated approach provides a reliable and effective method for early detection of insulation degradation, enhancing the performance of diagnostic systems and supporting better maintenance strategies in rotating machinery.