REAL-TIME DIAGNOSIS OF BATTERY CELLS FOR STAND-ALONE PHOTOVOLTAIC SYSTEM USING MACHINE LEARNING TECHNIQUES

Battery as a critical element in the stand-alone photovoltaic system remains without an appropriate protection fuse for short-circuit failure inside it. Therefore the safety is threatened and the lifetime of the battery is reduced. To address this problem, supervision of battery internal short-circuit is proposed using a machine learning anomaly detection and support vector machine (SVM) as fault detection and diagnosis respectively. Simulation of Stand-alone photovoltaic system with battery is carried-out to obtain data learning. In addition, a real profile of irradiance and temperature captured from Centre de Development des Energies Renouvelables (CDER), Algeria, during nine days is used as input of the system simulation. The developed anomaly detection and SVM diagnosis model show their ability to detect and diagnose the faults with high accuracy in test real-time data.