Fault Diagnosis of Photovoltaic Panels Using a Low-Cost Edge Device

In this article, a novel method for fault diagnosis of photovoltaic (PV) panels based on infrared thermography images was presented. The examined faults are: partial shading effect, dust accumulation, short-circuited PV module and damaged bypass diode in a PV module. Deep convolutional neural network (DCNN) classifiers were developed and embedded into a low-cost microprocessor (Raspberry Pi 4). Simulation results showed the ability of the developed classifiers to detect and classify faults with good accuracy of 96.52% and 92% for detection and classification, respectively. Moreover, experimental tests  demonstrated the feasibility of the developed method. A graphical user interface (GUI) was also developed.