Detection and assessment of partial shading in photovoltaic arrays

The paper presents a methodology for detection and assessment of partial shading conditions in photovoltaic (PV) arrays based on artificial neural networks (ANN) as a preliminary step toward automatic supervision and monitoring. The PV array is modeled under normal and partial shading conditions for performance comparison. ANN is designed, trained, and tested for full identification of the partial shading condition. One ANN detects the presence of partial shading and distinguishes it from the uniform change in environmental conditions. If the first ANN detects partial shading on the PV array, other two ANN agents determine the shading factor and infers the number of shaded modules of the array, consequently. Results show excellent performance of ANN on the detection and assessment of partial shading.