Electrical faults detection for the intelligent diagnosis of a photovoltaic generator

the work presented in this paper is dedicated to improving the methods of detection and diagnosis of faults affecting production systems, particularly photovoltaic systems. We proposed a new intelligent algorithm for the detection and diagnosis of PV installations, capable of detecting and resonate to define the type of defects that can affect this type of system. This new algorithm is based on the notion of pattern recognition, for that it is able to prepare the representation space and the decision space on the one hand, and on the other hand, the classification of all new observations collected during the functioning of the system. This algorithm mainly based on the method of knearest neighbor and two tools of artificial intelligence to improve this method and increasing the rate of its classification, which are fuzzy logic to optimize the location of the centers of gravity of classes and also the new observations, and the neural network that can classify the case of discharges ambiguity and releases distance which presents the limitations of the method of the k-nearest neighbor. We tested the performance of our algorithm on a database of a photovoltaic system at the research unit of GHARDAIA Algeria.