Online Detection of PV Degradation Effects Through ANN Classifier

Photovoltaic (PV) system reliability and its service life are strongly dependent on the state of health of PV panels, thus methodologies and technical solutions for the accurate monitoring and the on-line diagnosis of the PV panels are fundamental for the maintenance of current PV installations and the development of new ones. In this paper a model based diagnostic technique combined with a neural network classifier has been developed for an early detection of PV panel degradation. The proposed approach is suitable for PV applications where each panel, or a small number of panels, is connected to a dedicated power converter for achieving both a distributed maximum power point tracking and a detailed PV monitoring. The diagnostic method has been tested on the Pynq-z2 platform based on the system-on-chip architecture, nevertheless different board could be used for implementing the proposed approach on embedded system for the on-line operation.