System Faults Diagnosis in a Photovoltaic Generator Using Artificial Neural Network Approach

Thanks to acquisition systems, fault diagnosis methods in Photovoltaic Generators (PVG) find themselves confronted with large quantities of voluminous data that are sometimes unused. The international scientific community, through the application of artificial intelligence (AI) methods, considered it as a promising solution. They allow large data sets to be efficiently managed in order to diagnose possible complex industrial system related faults such as in PVG. In this article, we present a method for diagnosing the most frequently encountered system faults. Better knowledge of the behavior of the PVG has been carefully constructed and added to a algorithm based on Artificial Neural Networks (ANN). This configuration starts from a collection of data from robust and quality bases linked to our study system, with regard to the function of the objectives to be achieved. Finally, the analysis of the impact of system faults on the performance is made and the return to a good operating state of the PVG. The results obtained are presented and well interpreted, highlighting the advantages and limitations of the approach already implemented by other authors. The results finally prove the advantage of our method, for undertaking corrective or preventive maintenance actions on the faulty part of the generator. This will undoubtedly allow a rapid return to good operating condition.