Machine Learning-Based Fault Detection in Solar PV Systems
Abstract
Terrestrial stability of solar photovoltaic (PV) systems is great in order to enable maximization of energy output as well as providing adequate long life of the systems. The given paper introduces the machine learning (ML)-based fault detection system on solar PV installations with the help of an ESP8266 microcontroller, voltage sensors, and ThingSpeak IoT system. The sensor-collected voltage data in the PV panel is transmitted using ESP8266 and saved and visualized in ThingSpeak on the cloud. The obtained data is further analyzed in MATLAB where ML tools are trained to recognize abnormal patterns which were connected to the presence of common PV fault including partial shading, panel degradation, and connection problems. The suggested system allows diagnosing any fault early to maintain costs efficiently and avoid energy wastage. The experimental outcomes prove that the ML model is able to recognize the normal and faulty operation state with high accuracy. This is a smart monitoring and predictive maintenance of solar PV system IoT-enabled solution that is inexpensive and scalable.