This paper deals with the automation capabilities of defective solar modules detection by thermographic camera mounted on a dron or on the car roof. During thermography imaging analysis of large photovoltaic power plants there are captured large numbers of images. These images are then analyzed by human operators. Given the vast amount of images, which are sometimes very similar, and given of the specificity of some defects, the work of operators can be replaced by the automation recognizing of anomalies software. Two methods for the automatic detection of defects in thermal imaging pictures were developed and validated. First one method is based on the geometric information of tested modules and on the assumption that defective cell has an increased temperature along its whole surface and therefore will appear as a regular geometric shape which is recognizable by geometric comparisons. The second method does recognition by usage of trained artificial neural network.

No abstract

No abstract