Failure or degradation effects lead to power losses in solar panels during their field operation and are identified commonly by electroluminescence (EL) imaging. Some failures like potential-induced degradation (PID) and light and enhanced temperature induced degradation (LeTID) require an identification of the EL pattern over the entire solar panel. As the manual process of analyzing patterns is prone to error, we seek for an automatic detection of these failure types. We predict automatically the failure types PID and LeTID by adopting the principal component analysis (PCA) method in combination with a k-nearest neighbor (kNN) classifier. We increase the explained variance of the PCA by 4 %abs using a Gaussian blurring preprocessing step and gain insights into the basic mechanism of the machine learning algorithm by analyzing schematic EL images. The kNN classifier predicts the failure classes in the same way as the expert. Finally, we work with a larger test dataset of 40 similar images to mimic a field-typical user case and meet again the expert classification.
5 November 2019 in Progress in Photovoltaics: Research and Applications
We present a systematic study on the benefit of the implementation of poly‐Si on oxide (POLO) or related junctions into p‐type industrial Si solar cells as compared with the benchmark of Passivated Emitter and Rear Cell (PERC). We assess three aspects: (a) the simulated efficiency potential of representative structures with POLO junctions for none (=PERC+), one, and for two polarities; (b) possible lean process flows for their fabrication; and (c) experimental results on major building blocks. Synergistic efficiency gain analysis reveals that the exclusive suppression of the contact recombination for one polarity by POLO only yields moderate efficiency improvements between 0.23%abs and 0.41%abs as compared with PERC+ because of the remaining recombination paths. This problem is solved in a structure that includes POLO junctions for both polarities (POLO2), for whose realization we propose a lean process flow, and for which we experimentally demonstrate the most important building blocks. However, two experimental challenges—alignment tolerances and screen‐print metallization of p+ poly‐Si—are unsolved so far and reduced the efficiency of the “real” POLO2 cell as compared with an idealized scenario. As an intermediate step, we therefore work on a POLO IBC cell with POLO junctions for one polarity. It avoids the abovementioned challenges of the POLO2 structure, can be realized within a lean process flow, and has an efficiency benefit of 1.59%abs as compared with PERC—because not only contact recombination is suppressed but also the entire phosphorus emitter is replaced by an n+ POLO junction.
In this article, we present a new method to measure electroluminescence (EL) images of current half-cell PV modules without opening electrical contacts. The method uses the series parallel interconnection of the substrings in the module to power one module half (luminescence part) by illumination of the other module half (generator part). The defect interpretation of the resulting images changes slightly compared to usual EL images. Module defects in the generator part of the module influence the working point of the EL of the luminescence part. The new method has a practical limitation, as an electrical power of about 12.4 kW would be required for bringing the luminescent part into the operating point of the rated short-circuit current. Using infrared light emitting diodes and a working point of half of the nominal short-circuit power brings the needed power down to about 2.4 kW for a 144-cell module. For current Si-based PV modules, the method allows image acquisition times below 1 s per image with conventional silicon sensor-based cameras.
Failure or degradation effects lead to power losses in solar panels during their field operation and are identified commonly by electroluminescence imaging. Failures like potential induced degradation and light and enhanced temperature induced degradation require an identification of the electroluminescence pattern over the entire solar panel. As the manual process of analysing patterns is prone to error, we seek for an automatic detection of these failure types. We predict automatically the failure types potential induced degradation and light and enhanced temperature induced degradation by adopting the principle component analysis method in combination with a k-nearest neighbour classifier.