Fault detection and diagnosis in photovoltaic panels by radiometric sensors embedded in unmanned aerial vehicles
Photovoltaic solar energy is increasing its capacity in the global electric market due to its lower operating costs and higher efficiency, together with the support of the governments. Photovoltaic solar panels require high initial investments, and it is necessary to use advanced and efficient metho...
| Autores: | , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2021 |
| País: | España |
| Institución: | Universidad de Castilla-La Mancha |
| Repositorio: | RUIdeRA. Repositorio Institucional de la UCLM |
| OAI Identifier: | oai:ruidera.uclm.es:10578/36797 |
| Acceso en línea: | https://doi.org/10.1002/pip.3479 https://hdl.handle.net/10578/36797 |
| Access Level: | acceso abierto |
| Palabra clave: | Condition monitoring system Fault detection and diagnosis Pattern recognition Photovoltaic Radiometric sensor Unmanned aerial vehicle |
| Sumario: | Photovoltaic solar energy is increasing its capacity in the global electric market due to its lower operating costs and higher efficiency, together with the support of the governments. Photovoltaic solar panels require high initial investments, and it is necessary to use advanced and efficient methods that lead to the maintainability and reliability of these systems, extending their life cycle and productivity. New condition monitoring systems are being applied to reduce the cost of inspections and ensure efficient data collection. The main contribution of this paper is a new efficient and low-cost condition monitoring system based on radiometric sensors. The thermal patterns of the main photovoltaic faults (hot spot, fault cell, open circuit, bypass diode, and polarization) are studied in real photovoltaic panels. Different scenarios are considered, analyzing online the main patterns of the faults by Internet of Things. The accuracies of the results are statistically analyzed and compared with fault-free scenarios. The validation of the approach is done by thermographic analysis of the experiments and employing a different radiometric sensor. The detection of faults with statistical analysis is achieved in 100% of the cases, and the identification of the specific fault in 96% of the scenarios. An artificial neural network and classification learner algorithms are employed to validate the results, obtaining similar results for fault identification, 95% and 94%, respectively, and 100% for fault detection. |
|---|