abstract
Daylight photoluminescence (DPL) methods exhibit high performance and low operation cost for the fault detection in solar modules during daylight solar energy harvesting. The key point for increasing the quality of these imaging techniques is to filter out the influence from the varying surrounding reflections. Optical ultranarrow bandpass filters can directly eliminate the solar irradiation but are hardly practical for widespread application due to their excessively high manufacturing cost. Another approach to remove the background signal is the application of an external electrical signal variation exerted by the inverter to change the working point of the solar module from the open-circuit voltage to the maximum power point. However, the energy loss during the measurement and resulted output instability have to be minimized with simultaneous optimization of image quality, which is not trivial. In this paper, we present an optimized algorithm for processing DPL images during inverter operation like IV characterization, arc detection, DC switch-off, and dynamic shadowing while MPP-tracking. We employ the structural similarity index to evaluate the similarity between acquired DPL images and laboratory electroluminescence images for quantitative comparison between the methods. A combination of inverter power on/off switch for overview imaging and dynamic shadowing for detailed single module imaging with non-normalized Pearson correlation coefficients allows untrained persons to measure DLP images without rewiring or access to the inverter settings.