It seems that monitoring of solar panels will get even smarter over time. Forschungszentrum Jülich GmbH has developed a self-referencing algorithm that can predict and identify string yield loss and underperformance without additional weather data.

The researchers evaluated the yield data of 9MWp PV station by applying the algorithm to determine underperformance. The facility has three-phase inverters deployed across 1719 strings. Such inverters have three strings on one mounting rack located in three rows.

The algorithm correlates with irradiance-dependent performance ratio and doesn’t need to consider the weather data. Such performance ratio that captures regular degradation is influenced by grid despatch, inverter limitations and environmental influences. However, such a ratio is much less effective than self-referencing algorithms in the long term perspective. The algorithm takes into account plant layout and extraordinary aging of components and can study all kinds of units of a solar power plant.

The algorithm can also predict and identify yield loss in underperforming strings. Also, it can identify and quantify such local differences as topography changes, layout, soiling, shading, row position and others by analyzing underperforming strings. However, the further advantages of this technology is still to be determined as the researchers are going to apply it “for degradation studies of time-series and root-cause analysis with machine learning and combination with different data sources, for example, imaging data” as per research paper.

Self-referential algorithm uses PV system’s string with maximum output as a reference for measuring the performance. Researchers benchmark the self-referential ratio against the established performance ratio that requires solar irradiance for calculation.

If this invention will be widely applied in the future, it can significantly improve the precision of solar panel efficiency calculations and thus will allow us to eliminate the cause of underperformance quicker.