Photovoltaic Failure Detection Based on String-Inverter Voltage and Current Signals

Abstract

The existence of failures in photovoltaic systems causes energy losses, security problems, and damage to its components. Therefore, it is necessary to develop monitoring systems to improve their productivity, reliability, efficiency, and safety. This work proposes a method for detecting and indicating short-circuit failure and partial shading present in grid-connected photovoltaic modules. The novelty of this proposal is the processing of voltage and current signals generated (ripple signals) by the electrical interaction between the photovoltaic string, the photovoltaic inverter, the condition of the modules, temperature, and irradiance. The magnitudes of specific frequency components are obtained from these electrical signals using DFT, and Cartesian coordinates are formed in a three-dimensional plane. Coordinates belonging to each of the string conditions are mostly located in different octants from this plane. This distribution allows an assessment of the modules state. Certain scattering is solved with a K-nearest neighbors algorithm. The effectiveness of the methodology was experimentally validated in two PVG with different characteristics. In each test case, the method was adaptable to the real conditions of the photovoltaic system. The results show effectiveness greater than 90% in the first evaluation levels, there is a slightly lower detection certainty in anomalous conditions with coordinates that are very close to each other. However, the system has 100% certainty detecting the presence of an abnormal condition. The method allows adaptation to different conditions, and takes advantage of the electrical signals derived from the actual performance of the used devices.