Optimising Solar Power Plant Reliability Using Neural Networks for Fault Detection and Diagnosis

This study introduces an intelligent method to monitor grid-connected solar power stations, focussing on detecting problems in their energy output through the use of artificial neural networks (ANN). The main goal is to improve energy efficiency and bolster the reliability of solar power plants by forecasting their performance through real-time data analysis and modelling essential operational variables. The research was carried out in a solar field in AOULEF-ADRAR (South of Algeria), which covers six hectares and consists of 20,460 solar panels with an efficiency of 15 % to 20 %. The cumulative installed capacity is 5 MW, and the system is connected to a 30 kV electrical grid. The experimental findings validated the efficacy of the suggested ANN-based fault detection method. Subsequent to a sandstorm, the system exceeded standard operational limits, culminating in a total power overshoot of 200 KW. This procedure facilitated the identification of system faults and the execution of corrective measures, including the cleaning of PV modules to restore efficiency. The research highlights the importance of artificial intelligence (AI)-based monitoring systems to reduce downtime and maintenance expenses and guarantee consistent operation of photovoltaic plants under various environmental conditions. Research advocates for the integration of artificial neural networks with other machine learning methodologies, such as support vector machines, to improve fault prediction precision. Augmenting the data set by integrating data from various PV stations in different regions may improve the adaptability of the model to different environmental conditions. This method improves the creation of intelligent self-diagnosing solar power systems, promoting increased reliability and efficiency in the integration of global renewable energy.