Comparative analysis of machine learning algorithms for power factor prediction in smart grid applications

The escalating complexity of electrical energy systems demands innovative approaches to power management and predictive analysis. This groundbreaking study introduces a novel machine learning methodology for power factor prediction within a university campus setting, leveraging a sophisticated data normalization technique to address asymmetries in data collection. By systematically comparing multiple machine learning algorithms—including Logistic Regression, Random Forests, Support Vector Machines (SVM), k-Nearest Neighbors, and Multi-Layer Perceptron Neural Networks—the research provides insights into predictive performance. Notably, the research also developed an innovative data standardization mechanism using arithmetic functions, which effectively mitigates data asymmetry challenges. These results not only advance our understanding of power factor dynamics but also offer a robust framework for energy management in complex electrical systems, particularly in regions with adversarial operational characteristics.