Analysis and Classification of Partial Shading Conditions in Photovoltaic Arrays

Abstract

With the escalating global transition toward renewable energy, ensuring the operational stability and reliability of photovoltaic (PV) systems has become a critical concern. Among the various factors that impair PV system performance, partial shading emerges as a particularly challenging issue. This study conducts an in-depth investigation of partial shading effects by analyzing the power–voltage (P–V) response of PV arrays. A set of novel mathematical formulations is developed to characterize the extremum points on P–V curves under diverse shading scenarios. These formulations offer essential insights into the dynamic behavior of PV arrays subjected to non-uniform irradiance, facilitating accurate detection and classification of partial shading conditions. The proposed approach enhances fault diagnosis and contributes to more effective control strategies for maintaining system performance. Furthermore, four diagnostic features derived from the P–V curve analysis are employed to distinguish between simple and complex shading patterns using two distinct machine learning classifiers. The results confirm that the proposed models reliably discriminate between different levels of shading complexity, even when using a single extracted feature.