Comparative Analysis of Machine Learning Algorithms for Identifying Partial Shading Conditions on PV Array

The rapid global expansion of photovoltaic (PV) technologies has made solar power systems a pivotal component in reducing CO2 emissions and enhancing energy security. However, the efficiency of PV systems is often compromised by partial shading, which can significantly reduce power output and lead to accelerated degradation of solar panels. Detecting and assessing these shading conditions is crucial for maintaining PV system performance and longevity. This paper explores the application of machine learning (ML) techniques to identify and classify partial shading in PV systems. Six ML models are applied, namely K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Random Forest (RF), Artificial Neural Networks (ANN), Gradient Boosting, and Adaptive Boosting (AdaBoost) Models. The study uses Wavelet Packet Transform (WPT) and Empirical Mode Decomposition (EMD) to extract Intrinsic Mode Functions (IMFs) that capture the characteristic frequencies associated with shading. These features are then used to train and validate the ML models. Among the models tested, the Random Forest model demonstrated superior performance, achieving a detection accuracy of 98.43% and a classification accuracy of 97.6%. The results highlight the effectiveness of the Random Forest model in reliably diagnosing shading faults, thereby enhancing the reliability and efficiency of PV systems. The performance of six machine-learning algorithms is validated through the cross-validation method.