Comparison of the effectiveness among different photovoltaic fault detection techniques

Abstract

The main objective of this research was the quantitative comparison of the effectivity of three failure detection techniques in PV systems: a) VI, b) IRT, and c) EA; focused on the detection of partial shadows, soiling, and electrical failures. For this, an experiment was developed in a PV installation at the San Carlos Local Technological Campus of the Tecnológico de Costa Rica. The VI and IRT techniques were made using an unmanned aerial system and the electrical analysis compared the generated power with the estimated power using an embedded system Raspberry Pi 3 model b; each technique was implemented using low-cost instrumentation. The results showed that there were no significant differences in the total amount of failures detected by the three techniques; however, the VI was the best at detecting soiling and the worst at detecting electrical failures. Also, partial shadows were the type of failure that was most detected by all the techniques. Therefore, to detect as many faults as possible, VI should be combined with IRT or EA. This research contributes to the search for better fault detection techniques, also, the methodology developed can be replicated to compare other techniques and other types of faults. Finally, the results create objective indicators for the generation of suitable maintenance plans that will achieve a greater contribution to the environment and more profitable generation projects.