Photovoltaic system failure diagnosis based on adaptive neuro fuzzy inference approach: South Algeria solar power plant

Abstract The present work proposes a new solar power station surveillance approach based on the photovoltaic module failures diagnosis using an adaptive neuro-fuzzy inference approach. Indeed, the main aim of this proposed approach is to ensure and increased energy efficiency and improved reliability of the studied solar power station. This approach is used to generate faults indicators, to detect, locate and isolate the faults based on modeling of the main characteristic variables based on an adaptive neuro-fuzzy inference, where the main aim is the prediction of the expected studied system behavior based on the actual collected measurements of the studied system. Where, the investigation field of this work is implanted on an area of 60 ha it contains 120120 solar panels with an efficiency of 15–20% with a total power of 30 MW connected to the electrical network of 30 KV. The obtained results confirm the validity of the proposed approach in improving the reliability and the overall efficiency of the studied power system. It was proved experimentally that after a sandstorm, the normal operating mode thresholds were exceeded and absolute overshoots of 35.5, 5.6 and 1.3 were registered for the output power, the output voltage and the output current respectively. These registrations have permitted to identify the failures and to set up a decision for the cleaning of the photovoltaic module. Indeed, It has been proved in this work that the operation state mode can be maintained based on failure detection and its maintenance which can be achieved in time thanks to the proposed approach.